2010 ACS Content Test Evaluation Report - Parental Place of Birth

January 26, 2012

2012 AMERICAN COMMUNITY SURVEY RESEARCH AND EVALUATION REPORT
MEMORANDUM SERIES #ACS12-RER-04
MEMORANDUM FOR

ACS Research and Evaluation Steering Committee

From:

Enrique Lamas /Signed/
Chief, Population Division

Prepared by:

Luke J. Larsen, Elizabeth M. Grieco, and Patricia de la Cruz
Foreign-born Population Branch
Population Division

Subject:

2010 ACS Content Test Evaluation Report Covering Parental
Place of Birth

Attached is the final American Community Survey Research and Evaluation report for the 2010
ACS Content Test Evaluation Covering Parental Place of Birth. This report describes the results
of testing the parental place of birth questions in two locations on the ACS questionnaire.
If you have any questions about this report, please contact Elizabeth Grieco at (301)763-5275,
Luke Larsen at (301)763-2540, or Patricia de la Cruz at (301) 763-2377.
Attachment: (2010 ACS Content Test Evaluation Report Covering Parental Place of Birth)
cc:
ACS Research and Evaluation Steering Committee
ACS Research and Evaluation Team
Donna Daily
(ACSO)
Todd Hughes
Debbie Klein
David Raglin
Yuling Pan
(CSM)
Jennifer Tancreto
(DSSD)
Tony Tersine
Karen Humes
(POP)
Victoria Velkoff
Kurt Bauman
(SEHSD)

American Community Survey Research and Evaluation Program
January 31, 2012

2010 ACS Content Test
Evaluation Report Covering
Parental Place of Birth
FINAL REPORT

Luke J. Larsen
Elizabeth M. Grieco
Patricia de la Cruz
Population Division

Intentionally Blank

TABLE OF CONTENTS
EXECUTIVE SUMMARY ....................................................................................................................... iv
1. BACKGROUND
1.1 Motivation for the 2010 Content Test............................................................................................................................. 1
1.2 Previous Testing or Analysis .......................................................................................................................................... 2
1.3 Recommendations from Cognitive Testing .................................................................................................................... 2
1.4 Recommendations from the Expert Review Panel ......................................................................................................... 4

2. SELECTION CRITERIA
2.1 Critical Research Questions ............................................................................................................................................ 5
2.2 Supplementary Research Questions................................................................................................................................ 6
2.3 Additional Research Topics ............................................................................................................................................ 6

3. METHODOLOGY
3.1 Data Collection Methods ................................................................................................................................................ 7
3.2 Sample Design ................................................................................................................................................................ 8
3.3 Methodology Specific to Parental Place of Birth
3.3.1 Operational Definitions and Recodes ................................................................................................................... 9
3.3.2 Population Universes .......................................................................................................................................... 11
3.3.3 Analysis Framework and Statistical Testing Procedures .................................................................................... 12

4. LIMITATIONS ......................................................................................................................................13
5. CRITICAL RESEARCH QUESTIONS AND RESULTS
5.1 Response to the Content Test and Content Follow-up .................................................................................................. 14
5.2 Are the response distributions of parental place of birth and generational status comparable to existing data
sources? ......................................................................................................................................................................... 14
5.3 Which placement results in a lower missing data rate? ................................................................................................ 15
5.4 Do the two placements have similar or different response distributions?..................................................................... 16
5.5 Which placement results in more reliable estimates? ................................................................................................... 17
5.6 Does changing the placement of the parental place of birth questions from before to directly after the ancestry
question affect the item missing data rates, response distributions, or reliability for the ancestry question?................ 18
5.7 Does the placement of the parental place of birth questions directly before the school enrollment question affect
the item missing data rate, response distribution, or reliability for the school enrollment question? ............................ 20
5.8 Does the placement of the parental place of birth questions directly before the language spoken questions affect
the item missing data rates, response distributions, or reliability for the language spoken questions? ......................... 21
5.9 Summary of Critical Analysis ...................................................................................................................................... 22

6. SUPPLEMENTAL RESEARCH QUESTIONS AND RESULTS
6.1 For each mode of data collection, do the two placements have differential item missing data rates, response
distributions, or reliability of the data?.......................................................................................................................... 23
6.2 For each mail response stratum, do the two placements have differential item missing data rates, response
distributions, or reliability of the data?.......................................................................................................................... 25
6.3 Does either placement elicit respondent or interviewer behaviors that may contribute to interviewer or
respondent error? ........................................................................................................................................................... 26
6.4 Summary of Supplemental Analysis............................................................................................................................. 27

7. ADDITIONAL RESEARCH TOPICS AND RESULTS
7.1 Comparison of Content Test Data to ACS Data for Proximal Variables ...................................................................... 27
7.2 Additional Analysis of Proximal Variables by Mode in Content Test Data ................................................................. 29
7.3 Summary of Additional Research ................................................................................................................................. 30

8. SUMMARY ...........................................................................................................................................32
References ..................................................................................................................................................34

TABLE OF APPENDICES
Appendix A: Tables .............................................................................................................................. A-1
Appendix B: Images of the Mail Versions of the Control and Test Questions .....................................B-1
Appendix C: CATI and CAPI Versions of the Control and Test Questions .........................................C-1
Appendix D: Spanish Language Versions of the Control and Test Questions
(CATI/CAPI Modes) ...................................................................................................... D-1
Appendix E: Process Flowchart of the Content Follow-Up Survey Reinterview ................................. E-1
Appendix F: Parental Place of Birth Crosswalk for World Region of Birth Recode
using the American Community Survey .......................................................................... F-1
Appendix G: Parental Place of Birth Crosswalk for World Region of Birth Recode
using the Current Population Survey .............................................................................. G-1
Appendix H: Parental Place of Birth Crosswalk for Broad Place of Birth Recode
using the American Community Survey ......................................................................... H-1
Appendix I:

Parental Place of Birth Crosswalk for the Selected Place of Birth Recodes
using the American Community Survey ........................................................................... I-1

Appendix J:

Ancestry Crosswalk for Regional Ancestry Recode using the American
Community Survey ........................................................................................................... J-1

Appendix K: Ancestry Crosswalk for Selected Ancestry Recode using the American
Community Survey ......................................................................................................... K-1
Appendix L: Language Crosswalk for Selected Language Spoken at Home Recode
using the American Community Survey .......................................................................... L-1
Appendix M: Reliability Measures ....................................................................................................... M-1
Appendix N: Information Page ............................................................................................................. N-1

ACKNOWLEDGEMENTS
This report was prepared by Luke J. Larsen, Elizabeth M. Grieco, and Patricia de la Cruz
of the Foreign-Born Population Branch, Population Division, with assistance from Amy
Lauger and John Chesnut of the ACS Data Collections Method Branch, Decennial
Statistical Studies Division. The report was prepared under the general direction of
Jennifer Tancreto, ACS Data Collections Method Branch, Decennial Statistical Studies
Division; and Victoria Velkoff, Assistant Division Chief, Estimates and Projections Area,
Population Division. Overall direction was provided by Howard Hogan, Associate
Director for Demographic Programs, and Enrique J. Lamas, Chief, Population Division.
Within the Population Division, Yesenia Acosta, Christine Gambino, Tom Gryn, Edward
Trevelyan, and Nathan Walters assisted with data verification and statistical testing.
Angela Buchanan provided subject matter review and guidance. Principle editorial
review and guidance was provided by Frank B. Hobbs. Additional review and comments
were provided by Anthony G. Tersine, Jr., Decennial Statistical Studies Division; Karen
Humes, Population Division; Kurt Bauman, Social, Economic and Housing Statistics
Division; Yuling Pan, Center for Survey Measurement; and Todd Hughes, David Raglin,
Debra Klein, and Donna Daily, American Community Survey Office. Bob Kominski,
Alison Fields and Kin Corbin, Social, Economic and Housing Statistics Division,
completed coding and provided subject matter review and guidance.
Participants of the 2010 American Community Survey Content Test Inter-Agency
Subcommittee on Parental Place of Birth included Victoria Velkoff, Elizabeth M. Grieco,
Luke J. Larsen, Patricia de la Cruz, Tom Gryn, and David Dixon, Population Division;
Amy Lauger and John Chesnut, Decennial Statistical Studies Division; Theresa DeMaio
and Yuling Pan, Center for Survey Measurement; and representatives from the
Department of Homeland Security, Department of Agriculture, National Institute of
Health/National Institute of Child Health and Human Development, Bureau of Labor
Statistics, and Centers for Disease Control and Prevention.

EXECUTIVE SUMMARY
Test Objective
From August to December 2010, the Census Bureau conducted a field test of new and
revised questions in the 2010 American Community Survey (ACS) Content Test. The
results of that testing will determine the content to be incorporated into the ACS starting
in 2013.
The 2010 ACS Content Test questionnaire included two new questions on parental place
of birth (PPOB). Questions on PPOB are important because they divide the population
into “first generation” (the foreign born), “second generation” (the children of
immigrants), and “third-or-higher generation” (natives with no foreign-born parents)
categories, allowing policymakers and researchers to examine questions about adaptation
and integration of immigrants and their descendants over time. Also, questions on PPOB,
because they clearly define the second generation, are required to examine the social and
economic characteristics of the children of immigrants.
At present, the Current Population Survey (CPS) is the principal source of information on
the population by generational status. However, the CPS is generally confined to
national-level analysis and can only provide limited data at the sub-national level where
immigrants are settling and populations are changing rapidly. Because the ACS has a
larger sample size and the data can be combined into multi-year estimates for low levels
of geography, the study of immigrant assimilation in the United States would greatly
benefit from including PPOB questions on the ACS survey instrument.
Methodology
As a result of early questionnaire development efforts, Census Bureau analysts identified
one primary set of PPOB questions to test in two locations on the ACS questionnaire. A
research and evaluation plan was developed that detailed the specific course of analysis
necessary to determine whether or not the proposed changes should be recommended for
inclusion on the ACS questionnaire. For the PPOB questions, the project plan focused on
three broad topic questions:
1. Do the PPOB questions “work” on the ACS questionnaire, providing reasonable
and reliable data?
2. Does the placement of the PPOB questions on the questionnaire affect the data
produced?
3. Does the placement of the PPOB questions affect the data produced by the
ancestry, school enrollment, and language questions?
This report outlines the research questions, methodologies, and metrics used to answer
the three broad topic questions and describes the analytical results that helped form the
recommendation for inclusion on the ACS.

iv

Research Questions and Results
Are the response distributions of PPOB and generational status (i.e., first, second, and
third-or-higher generation) comparable to existing data sources?
Yes, the response distributions for father’s place of birth, mother’s place of birth, and
generational status on both panels of the Content Test were very similar to distributions
derived from the CPS.
Which placement results in a lower missing data rate?
The placement used by the Control panel resulted in a lower item missing data rate for
both father’s and mother’s place of birth (6.9 percent and 6.0 percent, respectively) than
the placement used by the Test panel (7.4 percent and 6.6 percent, respectively).
Do the two placements have similar or different response distributions?
In general, there were no notable significant differences in the response distributions of
father’s and mother’s place of birth between the two placements used by the Control and
Test panels. Thus, the two placements yielded similar distributions.
Which placement results in more reliable estimates?
In general, there were no notable significant differences in the reliability measures of
father’s and mother’s place of birth between the two placements used by the Control and
Test panels. Thus, neither placement yielded more reliable estimates than the other.
Does changing the placement of the PPOB questions from before to directly after the
ancestry question affect the item missing data rate, response distribution, or reliability
for the ancestry question?
In general, there were no notable significant differences in the response distributions or
reliability measures of first reported ancestry or second reported ancestry between the
two placements of PPOB used by the Control and Test panels. However, the placement
used by the Test panel resulted in a lower item missing data rate for first reported
ancestry (13.3 percent) than the placement used by the Control panel (14.5 percent).
Does the placement of the PPOB questions directly before the school enrollment question
affect the item missing data rate, response distribution, or reliability for the school
enrollment question?
In general, there were no notable significant differences in the response distributions or
reliability measures of school enrollment status between the two placements of PPOB
used by the Control and Test panels. However, the placement used by the Test panel
resulted in a lower item missing data rate for school enrollment status (4.8 percent) than
the placement used by the Control panel (5.4 percent).
v

Does the placement of the PPOB questions directly before the language spoken questions
affect the item missing data rates, response distributions, or reliability for the language
spoken questions?
In general, there were no notable significant differences in the response distributions or
reliability measures of any of the language questions between the two placements of
PPOB used by the Control and Test panels. However, the placement used by the Control
panel resulted in lower item missing data rates for non-English-language-spoken-at-home
status and English speaking ability (4.9 percent and 1.5 percent, respectively) than the
placement used by the Test panel (6.9 percent and 2.1 percent).
For each mode of data collection, do the two placements have differential item missing
data rates, response distributions, or reliability of the data?
In general, there were no notable significant differences among the response distributions
or reliability measures of father’s and mother’s place of birth between the two placements
of PPOB used by the Control and Test panels, regardless of the mode of data collection.
However, the placement used by the Control panel resulted in lower item missing data
rates in the mail mode for both father’s and mother’s place of birth (8.0 percent and 6.9
percent, respectively) than the placement used by the Test panel (8.6 percent and 8.0
percent, respectively). When considering only the households whose data was collected
by CATI or CAPI, there was no significant difference in item missing data rates for either
PPOB question between the two placements used by the Control and Test panels.
For each mail response stratum, do the two placements have differential item missing
data rates, response distributions, or reliability of the data?
In general, there were no notable significant differences among the response distributions
or reliability measures of father’s and mother’s place of birth between the two placements
of PPOB used by the Control and Test panels, regardless of the mail response stratum.
However, the placement used by the Control panel resulted in lower item missing data
rates in the high response area stratum for both father’s and mother’s place of birth (6.4
percent and 5.7 percent, respectively) than the placement used by the Test panel (7.2
percent and 6.5 percent, respectively). When considering only the households in the low
response area stratum, there was no significant difference in item missing data rates for
either PPOB question between the two placements used by the Control and Test panels.
Does either placement elicit respondent or interviewer behaviors that may contribute to
interviewer or respondent error?
No, there were no notable significant differences among either respondent or interviewer
behaviors between the Control and Test panels for either the father’s or mother’s place of
birth questions.

vi

Recommendations
The results of the 2010 ACS Content Test demonstrated that questions on PPOB are
clearly understood by respondents and provide consistent and reliable data. Based on
these results, the Census Bureau recommended that both questions on parental place of
birth be included in the ACS starting in 2013 using the question format tested and in the
placement used by the Control panel (i.e., between the year of entry and school
enrollment questions). The results of the ACS 2010 Content Test have also demonstrated
that, if the Control placement were to be used, there is currently enough space on the
ACS questionnaire for both questions on parental place of birth.
The Census Bureau believes there is added value in collecting information about PPOB,
though some may feel that this topic is somewhat duplicative when collected in
connection with existing survey questions on race, Hispanic origin, and ancestry. Adding
the PPOB questions to the questionnaire in 2013 would be done as part of a multi-year
process to further examine the relationship of the data for these topics. The ACS data
would also be evaluated in connection with results from the 2010 Census Alternative
Questionnaire Experiment, and this combined research would be used in determining
recommendations for which questions would remain on the ACS at the conclusion of this
process. The Census Bureau plans to provide various opportunities for public comment
as well as dialogue with groups that are especially interested in these data as we refine the
plans and share results on this cross-topical research.

vii

1. BACKGROUND
1.1 Motivation for the 2010 ACS Content Test
To evaluate proposed changes to the content of the American Community Survey (ACS),
the Census Bureau conducted the 2010 ACS Content Test. 1 The objective of the Content
Test, for both new and existing questions, was to determine the impact of changing
question wording, response categories, and redefinition of underlying constructs on the
quality of the data collected.
Through the Office of Management and Budget (OMB) Interagency Committee on the
ACS, subject matter experts from the Census Bureau and key data users from other
federal agencies collaborated in identifying revised and new questions for inclusion in the
Content Test. The suggested new and revised questions affected both the housing and
detailed person sections of the ACS questionnaire.
In the housing section, the food stamps question was altered to reflect a name change for
the food stamps program. In addition, a series of new questions were added related to
household computer ownership and Internet subscription.
Several changes were made in the detailed person section. First, a change in data needs
for the veteran series led to a revised set of response categories for the veteran status and
period of military service questions. Second, the question wording of the cash public
assistance income question was modified to address under-reporting of assistance on
behalf of children and single payment recipients. Third, to simplify the income questions
related to wages (wages, salary, commissions, bonuses, or tips) and property income
(interest, dividends, rental income, royalty income or income from estates and trust),
these questions were broken up into smaller questions for the Computer-Assisted
Telephone Interviewing (CATI) and Computer-Assisted Personal Interviewing (CAPI)
instruments only. Fourth, a set of new questions on parental place of birth was added to
enable data users to divide the population into “first generation” (the foreign born),
“second generation” (the children of immigrants), and “third or higher generation”
(native born with no foreign-born parents).
To meet the test objective of the Content Test, analysts evaluated changes to question
wording, response categories, instructions, and examples relative to a control version of
the question or relative to another version or placement for new questions. This report
discusses the results of the analysis of new questions on parental place of birth (PPOB).

1

This report uses the terms “2010 ACS Content Test” and “Content Test” interchangeably.

1

1.2 Previous Testing and Analysis
The 2010 ACS Content Test represents the first time that questions on PPOB were tested
for possible inclusion on the ACS questionnaire. Two questions on PPOB have been
included on the Current Population Survey (CPS) since 1994: 1) In what country was
your father born? and 2) In what country was your mother born? However, due to
constraints associated with the size of the CPS sample, analysis is generally confined to
the national level only and, for the foreign born, limited to broad regions of birth.

1.3 Recommendations from Cognitive Testing
Prior to conducting the Content Test, the Research Triangle Institute (RTI), Westat, and
Research Support Services (RSS) conducted cognitive testing under contract from the
Census Bureau to assist in identifying a final set of questions for the field test. Three
versions of each question topic were tested with the goal of choosing the best version for
the revised questions and the best two versions for the new questions. The questions
were tested in the three modes used in the ACS data collection (paper, telephone
interview, and personal interview) in English and Spanish. Cognitive interviews
consisted of one-on-one interviews using the proposed questions in the context of the
ACS survey. Survey methodologists also conducted respondent debriefings. 2
To develop possible question formats for cognitive testing, the Census Bureau formed a
Subcommittee on Parental Place of Birth, which included analysts from the Census
Bureau as well as other federal agencies interested in the topic. 3 The Subcommittee
developed three versions of the PPOB questions for consideration.
The first version was the simplest version – the respondent was asked the country of birth
of the parent – and was the version most similar to the existing PPOB questions used in
the CPS. The second version was more complex – the respondent was first asked to
indicate whether the parent was born in or outside the United States; if the parent was
born outside the United States, then the respondent was asked to identify the parent’s
country of birth. The third version was a slight variation on the second – the respondent
was asked whether the parent was born in the United States; if the response was “no,”
then the respondent was asked to identify the parent’s country of birth. The three
question versions used on the self-administered mail-back questionnaire are displayed in
Figure 1 in both English and Spanish. 4

2

Full details about the cognitive test are available in the final report produced by the contractors (Hinsdale
et al, 2009).
3
Throughout this report, the Subcommittee on Parental Place of Birth is often referred to simply as “the
Subcommittee”. Federal agencies other than the Census Bureau represented on the Subcommittee included
the Department of Homeland Security, Department of Agriculture, National Institute of Health/National
Institute of Child Health and Human Development, Bureau of Labor Statistics, and Centers for Disease
Control and Prevention.
4
The question wording was slightly different for all three test versions in the interviewer-administered
mode. This set of questions can be found in the cognitive testing final report (Hinsdale et al, 2009).

2

Figure 1. PPOB Question Versions Considered in Cognitive Testing
Question topic Version 1
Version 2
Version 3
Was this person’s FATHER
Was this person’s FATHER
Father’s place In what country was this
person’s FATHER born?
born in or outside the United
born in the United States?
of birth
__ Yes
Print name of country, or
States? Mark (x) one box.
(English)
__ Born in the United States
__ No – Print name of
Puerto Rico, Guam, etc.
__ Born outside the United
States – Print name of
foreign country, or Puerto
Rico, Guam, etc.
_________________

foreign country, or Puerto
Rico, Guam, etc.
________________

Mother’s place
of birth
(English)

In what country was this
person’s MOTHER born?
Print name of country, or
Puerto Rico, Guam, etc.

Was this person’s MOTHER
born in or outside the United
States? Mark (x) one box.
__ Born in the United States
__ Born outside the United
States – Print name of
foreign country, or Puerto
Rico, Guam, etc.
_________________

Was this person’s MOTHER
born in the United States?
__ Yes
__ No – Print name of
foreign country, or Puerto
Rico, Guam, etc.
________________

Father’s place
of birth
(Spanish)

¿En qué país nació el
PADRE de esta persona?
Escriba en letra de molde el
nombre del país o Puerto
Rico, Guam, etc.

¿Nació el PADRE de esta
persona en los Estados
Unidos o fuera de los
Estados Unidos? Marque (X)
UNA casilla.
__ Nació en los Estados
Unidos
__ Nació fuera de los
Estados Unidos – Escriba en
letra de molde el nombre del
país extranjero o Puerto
Rico, Guam, etc.
__________________

¿Nació el PADRE de esta
persona en los Estados
Unidos?
__ Sí
__ No – Escriba en letra de
molde el nombre del país
extranjero o Puerto Rico,
Guam, etc.
________________

Mother’s place
of birth
(Spanish)

¿En qué país nació la
MADRE de esta persona?
Escriba en letra de molde el
nombre del país o Puerto
Rico, Guam, etc.

Nacio la MADRE de esta
persona en los Estados
Unidos o fuera de los
Estados Unidos? Marque
(X) UNA casilla.
__ Nació en los Estados
Unidos
__ Nació fuera de los
Estados Unidos – Escriba en
letra de molde el nombre del
país extranjero o Puerto
Rico, Guam, etc.
__________________

¿Nació la MADRE de esta
persona en los Estados
Unidos?
__ Sí
__ No – Escriba en letra de
molde el nombre del país
extranjero o Puerto Rico,
Guam, etc.
________________

The results of cognitive testing suggested no substantial conceptual differences,
respondent preferences, or response quality among the three question versions. As no
additional information would likely be obtained from further testing different question
versions on the Content Test, the Subcommittee on Parental Place of Birth recommended
1) testing a single version of the PPOB questions and 2) testing this version in two
different places on the Content Test questionnaire. This approach reflected concerns that
the presence of the PPOB questions (with their shifted focus from a person to the parents
3

of that person) may have negative effects upon the quality of data gathered from
questions in close proximity. The Subcommittee determined that keeping the wording of
the questions consistent across experimental panels but varying the location of those
questions would allow analysts to measure the “placement effect” upon the PPOB
questions and the surrounding questions.
Of the three versions under consideration, the Subcommittee chose Version 1 for
inclusion on the Content Test questionnaire. This choice was supported by the results of
cognitive testing in two ways. First, Spanish-speaking respondents reported that the
Version 1 questions were easier to answer than the longer and more complex questions
included in Versions 2 and 3. Second, Version 1 reduced the problems associated with
identifying island areas as separate from the United States. In addition, the
Subcommittee noted that: the Version 1 questions were very similar in wording to the
PPOB questions used by the CPS; they were similar in structure to the place of birth
(POB) questions on the ACS questionnaire; and, as space limitations are always an issue,
they would require less room on the questionnaire than longer and more complex
questions from Versions 2 and 3.
The Subcommittee recommended two placements for the PPOB questions on the Content
Test questionnaire. One placement was between questions on year of entry and
educational attainment, and the other was between questions on field of degree and
ancestry/ethnic origin.

1.4 Recommendations from the Expert Review Panel
Following cognitive testing, an expert review panel composed of government survey
methodologists reviewed and suggested changes to the final question versions proposed
to move forward from cognitive testing into the field test. The proposed changes for each
question topic were approved by the corresponding OMB interagency subcomittee
responsible for initiating the research. The OMB provided final approval of the proposed
changes.
Regarding the PPOB questions, the expert review panel concurred with the
recommendations made by the Subcommittee on question format and placement.
However, the second placement was later revised by the Census Bureau, such that the
PPOB questions were placed after the question on ancestry and before the questions on
language spoken at home and English-speaking ability. In the original placements, the
PPOB questions preceded the question on ancestry in both panels. The Census Bureau
recognizes that the questions on PPOB – at least in part and for some sub-populations –
conceptually overlap with the question on ancestry. The decision to use the alternative
second placement was made due to concerns that, if the PPOB questions preceded
ancestry in both panels, it would be difficult to assess the effect that these questions
might have on the quality of the ancestry data.

4

2. SELECTION CRITERIA
2.1 Critical Research Questions
After cognitive testing was completed and the expert review panel had provided its
recommendations, Census Bureau analysts developed research and evaluation project
plans detailing the specific analyses necessary to determine whether or not the proposed
changes should be recommended for inclusion on the ACS questionnaire. For the PPOB
questions, the project plan focused on seven critical research questions that can be
organized under three broad topic questions:
First, do the PPOB questions “work” on the ACS questionnaire, providing reasonable and
reliable data?
•

Are the response distributions of father’s place of birth, mother’s place of birth,
and generational status (i.e., first, second, and third-or-higher generation)
comparable to existing data sources?

Next, does placement of the PPOB questions on the questionnaire affect the data
produced?
•
•
•

Which placement results in a lower missing data rate?
Do the two placements have similar or different response distributions?
Which placement results in more reliable estimates?

Finally, does the placement of the PPOB questions affect the data produced by the
ancestry, school enrollment, and language questions?
•
•
•

Does changing the placement of the PPOB questions from before to directly after
the ancestry question affect the item missing data rate, response distribution, or
reliability for the ancestry question?
Does changing the placement of the PPOB questions from before to directly after
the school enrollment question affect the item missing data rate, response
distribution, or reliability for the school enrollment question?
Does changing the placement of the PPOB questions from before to directly after
the language spoken questions affect the item missing data rate, response
distribution, or reliability for the language questions?

The results of the statistical analyses for each of the critical research questions (discussed
in Section 5) were generally in agreement and were used to form the final
recommendation sent by the Census Bureau to the Office of Management and Budget and
the American Community Survey Office. However, had the analyses provided
conflicting results, the research plan prioritized the work associated with topic questions
1 through 3, in that order, to guide the final decision-making process.

5

2.2 Supplementary Research Questions
In addition to the seven critical research questions, three supplementary questions were
included in the project plan to provide additional information on by-treatment differences
among the PPOB questions within response mode types and within mail response strata,
as well as potential error based on interviewer and/or respondent behavior. The goal of
this set of questions was to provide additional information about the performance of the
PPOB questions on the Content Test questionnaire. The statistical analyses completed to
answer these questions (discussed in Section 6) were used to inform, but not determine,
the recommendations for including the PPOB questions on the ACS questionnaire. The
supplementary research questions included:
•
•
•

For each mode of data collection (i.e., by mail, by phone, and in person), do the
two placements have differential item missing data rates, response distributions,
or reliability of the data collected from the PPOB questions?
For each mail response stratum (i.e., high and low), do the two placements have
differential item missing data rates, response distributions, or reliability of the
data collected from the PPOB questions?
Does either placement elicit respondent or interviewer behaviors that may
contribute to interviewer or respondent error?

2.3 Additional Research Topics
Although not part of the original project plan, two additional research topics were
included to determine if the presence of PPOB questions on the Content Test
questionnaire had any deleterious effects on the data quality produced by subsequent
questions. The additional research topics included:
•

•

For both placements, how do the item missing data rates and response
distributions for variables that followed the PPOB questions on the Content Test
questionnaire compare with the allocation rates of those same questions as
derived from the ACS ?
For each mode of data collection, do the two placements have differential item
missing data rates, response distributions, or reliability of the data for variables
that followed the PPOB questions?

The results of the statistical analyses associated with these questions (discussed in
Section 7) were used to inform, but not determine, the final recommendation for
inclusion of the PPOB questions on the ACS questionnaire.

6

3. METHODOLOGY
3.1 Data Collection Methods
The initial stages of the Content Test consisted of content determination, cognitive
laboratory pretesting, and expert reviews for the purpose of developing new and alternate
versions of question content. The field test portion of the ACS Content Test used the
data collection methodology currently used in the ACS (i.e., mail questionnaire, followup CATI, and follow-up CAPI) with an added reinterview conducted via a CATI
instrument known as the Content Follow-Up (CFU) survey. Additional data were
collected on respondent and interviewer behavior during the field test via Computer
Audio Recorded Interviewing (CARI) technologies for a subset of respondents during the
CATI and CAPI follow-up modes of data collection.
The Content Test followed the same schedule and procedures for the mail, CATI, and
CAPI operations as the September 2010 ACS production panel. Questionnaires were
mailed to sampled households at the end of August 2010. The Content Test used an
English-only mail form but the automated instruments (CATI, CAPI, and CFU) included
both English and Spanish versions. Households not responding by mail and for which a
phone number was available were contacted for a CATI interview during the month of
October 2010. In November 2010, Census Bureau field representatives visited a sample
of households that did not respond by mail or CATI to attempt a CAPI interview. The
CAPI operations ended December 2, 2010.
The field test included a CATI CFU reinterview to collect additional measures for the
study of response error. This operation started approximately two weeks after the initial
mail-out of questionnaires and ended two weeks after the end of the CAPI follow-up data
collection operation. The CFU included all occupied households for which the Census
Bureau received a response in the original interview and had a telephone number. A
response was defined as a case where the household provided data through at least the
first person’s place of birth question for mail cases or at least a sufficient partial interview
for CATI/CAPI interviews. The reinterview was conducted about 2 to 4 weeks after the
original interview and with the original respondent when possible. Note that the CFU
CATI reinterview was an abbreviated version of the original Content Test interview. The
CFU survey instrument included the basic demographic section and only those questions
preceding the questions being tested in the housing and the detailed person sections to
provide context (see Appendix E for the process flowchart of the CFU survey
instrument). For PPOB, the “ask-reask” reinterview method was used in the CFU.
The Content Test did not include all of the data collection operations and processes used
in the ACS. First, while the Telephone Questionnaire Assistance (TQA) program’s tollfree number was available to Content Test respondents for assistance, the CATI
instrument did not include content changes from the Content Test. Therefore, data
collected from Content Test respondents via TQA CATI interview were not included in
any part of the data analysis. Second, since the objective was to study response error

7

using unedited data, the Content Test excluded the Failed Edit Follow-up (FEFU) CATI
operation as well as the edit and imputation data processes.

3.2 Sample Design
The 2010 ACS Content Test consisted of a national sample of 70,000 residential
addresses in the contiguous United States. 5 The sample design for the Content Test was
largely based on the ACS sample design with some modifications to meet the test
objectives. The modifications included adding an additional level of stratification by
stratifying addresses into high and low mail response areas, over-sampling addresses
from the low mail response areas to ensure equal response from both strata, and sampling
units as pairs. The high and low mail response strata were defined based on ACS mail
response rates at the tract-level. The paired-sample selection formed pairs by first
systematically sampling an address within the defined sampling strata and then pairing
that address with the address listed next in the geographically sorted list. However, the
pair was not likely comprised of neighboring addresses. One member of the pair was
randomly assigned to the Control panel and the other member was assigned to the Test
panel. Those addresses assigned to the Test panel received the revised ACS questions
and the questions new to the ACS. The Control panel received the current questions on
the ACS as well as different versions of the new questions.
Another modification to the ACS sample design included the addition of a third sampling
stage. At the first stage, the 2010 ACS first-stage sample was used as the Content Test
first-stage sample. At the second stage, all housing units in the ACS first-stage sample
not selected in the 2010 ACS second-stage sample were selected as the Content Test
second-stage sample. In addition, any units that were selected to be in other operations
(e.g., training, other tests, etc.) were not selected in the Content Test second-stage
sample. At the third stage, addresses were selected using a sampling method similar to
the 2010 ACS second-stage sample design with the exception of adding the high and low
mail response stratification.

3.3 Methodology Specific to Parental Place of Birth
This section briefly reviews the operational definitions and recodes used throughout the
report, the population universes of pertinent variables, and the analytical framework used
to organize the statistical testing process.

3.3.1 Operational Definitions and Recodes
Generational status: Generational status is conceptually related to nativity status (i.e.,
whether a person is native born or foreign born). 6 Demographers who study the impact
of immigration on the U.S. population are not only concerned with the foreign born but
also the native-born children of the foreign born. These groups are referred to as the
5
6

The sample universe of the Content Test did not include Alaska, Hawaii, and Puerto Rico.
In this report, the terms “native” and “native born” are used interchangeably.

8

“first generation” and “second generation,” respectively. The children of the second
generation – in addition to their children, their children’s children, and so on – are part of
the “third-and-higher generation,” which can also be defined as the mathematical
difference between the size of the total native population and that of the second
generation.
In this report, generational status was determined from a person’s nativity status in
addition to place-of-birth information about that person’s parents. All foreign-born
persons were part of the first generation. If a native person had at least one foreign-born
parent, then that person was part of the second generation. If both parents of a native
person were also native, then that person was part of the third-and-higher generation.
Regardless of the data source (Content Test, CPS, or ACS), the person’s nativity status
was derived from the citizenship question, whereas the parents’ nativity statuses were
derived from the PPOB questions.
PPOB recodes: In the analysis, three recodes of the PPOB data were used: 1) parental
broad place of birth; 2) parental world region of birth, and 3) parental selected place of
birth. Each of these recodes was created separately for both father’s POB and mother’s
POB.
Parental broad place of birth (BPOB) was recoded into two distinct categories: whether
the parent was born inside the United States (including Puerto Rico and U.S. Island Areas)
or outside the United States. The parental world region of birth (WROB) recode included
six distinct categories based on the following geographic regions: United States, Puerto
Rico and U.S. Island Areas, Asia, Europe, Latin America and the Caribbean 7, and other
areas. The parental selected place of birth (SPOB) included the 10 countries with the
largest number of responses separately for mother’s POB and father’s POB, as identified
by the distributions derived from the 2009 CPS Annual Social and Economic Supplement
data. It should be noted that the SPOB recode is slightly different for father’s POB than
for mother’s POB. Eight of the selected countries were identical for both father’s and
mother’s POB: Canada, China, El Salvador, India, Italy, Mexico, Philippines, and
Vietnam. Cuba and the Dominican Republic were included in mother’s SPOB, while
Germany and the United Kingdom were included in father’s SPOB. In both variables, all
non-selected places of birth were collapsed into a residual category. Additional
information about the PPOB recodes is available in Appendices F, G, H, and I.
To guide final decision-making on whether or not to include the PPOB questions on the
ACS questionnaire, the statistical results associated with the three recodes were given
different ranks of importance: the “broad” recode was deemed most important, followed
by the “world region” recode and, finally, the “selected” recode. There were two reasons
for this. First, the distribution for PPOB is dominated by the United States. According to
the 2009 CPS, the United States represented more than 75 percent of both father’s and
mother’s POB, followed by all other countries, most representing less than 1 percent
each. A response pattern such as this is likely to exhibit data sparseness issues,
particularly among foreign places of birth that typically have small base populations.
7

Henceforth, the region “Latin America and the Caribbean” will be shortened to “Latin America”.

9

Second, the sample size for each panel of the Content Test was relatively small, much
smaller than that used in the ACS. Considering both the skewed distributions of PPOB
and the small sample sizes in the Content Test, the analysis between treatments of small
country-of-birth groups could yield statistically significant differences that may actually
be sampling artifacts. The three recodes with increasing degrees of geographic focus
were used to mitigate problems associated with interpreting statistical analyses based on
skewed distributions and small populations.
Ancestry recodes: As with the PPOB questions, the first reported ancestry and second
reported ancestry variables each featured several hundred valid response categories, so
the same data sparseness issues presented for PPOB were also a matter of concern for
ancestry. The authors mitigated this problem by constructing two recodes for each of the
two ancestry variables (see Appendices J and K). In the first – called regional ancestry –
the ancestry variables were each recoded into twelve categories based on the geographic
regions from which each ancestry originates. In the second – called selected ancestry –
the ancestry variables were each recoded into a list of ten most populous ancestry
responses, with the remaining responses collapsed into a residual category. This recode
functions similarly to the parental SPOB recode described above; however, in this case,
the list of ten ancestry groups was based on information from a Census 2000 Brief on
ancestry in addition to data from the 2009 ACS (Brittingham and de la Cruz, 2004).
Language spoken at home recode: The question about specific languages other than
English that are spoken at home has hundreds of valid response categories, just as with
the ancestry and PPOB questions. However, the data sparseness problem is more
magnified than in the other topics due to the smaller population universe of this language
question – persons who respond to this item must (1) be at least 5 years or older and (2)
report that they do speak a language other than English at home. The authors mitigated
this problem by constructing a recode – called selected language spoken at home – that
uses a response list for language spoken at home to identify the five languages other than
English most commonly spoken at home and collapsing the remaining languages into a
residual category (see Appendix L). This recode functions similarly to the parental
SPOB and selected ancestry recodes described above; in this case, the list of five
language groups was based on data from the 2009 ACS.
Item missing data rate: The item missing data rate (IMDR) of a given question was the
proportion of (a) eligible respondents (either persons or households) that did not provide
a valid, codable response to (b) the total number of eligible respondents, where eligibility
was determined by the universe definition for the question. Respondents who refused to
answer the question, gave a “don’t know” response, or provided a response that could not
be coded (due to illegibility, inaudibility, or some other reason) were included among
those who had a missing value for that particular question. By keeping the IMDRs as
low as possible, the potential for nonresponse bias can be minimized.
Uncodable response rate: As previously mentioned, uncodable responses are a subset of
the classification for missing values. In fact, the uncodable response rate (URR) for a
given question – which itself is a subset of the IMDR – was defined as the percentage of
10

the total number of eligible respondents that gave a response which could not be coded.
As a supplement to the issue of missing data rates, the analysts were concerned that the
URRs of the PPOB questions might have been significantly higher than the URR of the
existing POB question, or that the URRs of the PPOB questions might be significantly
affected by the question placement.
Multiple ancestry response rate: The two ancestry variables – first and second reported
ancestry – stem from a single question on ancestry. On the mail version of the
questionnaire, two write-in fields are provided with this item, thereby inviting
respondents to submit more than one response to this question. 8 The multiple ancestry
response rate (MARR) was the percentage of the total number of eligible respondents to
the ancestry question who gave two responses.
Response mode recode: As previously stated, the Content Test was administered through
three response modes – mail, CATI, and CAPI. Originally, the authors intended to study
the PPOB data by each of the three modes as part of the supplemental analysis.
However, when data analysis activities were underway, the authors discovered instances
of data sparseness among some of the CATI and CAPI distributions of father’s and
mother’s world region of birth. These empty cells would have required either some
modification of statistical testing procedures designed specifically for tests involving the
empty cells or a different categorization of the data that eliminated the empty cells. The
authors chose the latter option and collapsed the CATI and CAPI modes into a single
non-mail mode category.

3.3.2 Population Universes
The general population universe in each panel of the Content Test was the household
population in the contiguous United States. Many of the relevant variables in this report
– father’s place of birth, mother’s place of birth, place of birth, first reported ancestry,
and second reported ancestry – used this basic universe. Other variables placed more
restrictions on the universe. For instance, the population universe for school enrollment
status was the household population in the contiguous United States aged 3 years or
older, and the population universe for non-English-language-spoken-at-home status was
the household population in the contiguous United States aged 5 years or older. The
universe used for both language spoken at home and English speaking ability was also
restricted to those aged 5 years or older, but a second condition was imposed as well,
wherein universe members must have also reported that they spoke a language other than
English at home.

3.3.3 Analysis Framework and Statistical Testing Procedures
Much of the data analysis in this report followed a common structure:

8

CATI and CAPI respondents could also submit multiple ancestry responses, though they were not
prompted to do so by the interviewers.

11

•
•
•

First, for a given variable, the IMDRs were compared between the Control and
Test panels of the Content Test.
Next, the response distributions of that variable (possibly under one or more
recodes) were compared between the Control and Test panels. These
comparisons were conducted both overall and by individual response categories.
Then, the reliability measures of that variable (which, again, may be recoded)
were compared between the Control and Test panels using Content Follow-up
data in conjunction with the Content Test data for all persons for whom there was
a response in both surveys. These measures included the gross difference rate
(GDR), the index of inconsistency (IoI), and the L-fold index of inconsistency,
which allowed for comparisons of both the overall distribution and by individual
response categories. 9

The statistical analyses used in this report included individual t-tests, simultaneous t-tests,
and chi-square tests. Nearly all of the statistical tests in this report were t-tests that were
designed to determine whether the differences between the Control and Test versions of
estimated rates, percentages, and statistics were statistically significant. Unless otherwise
noted, these tests were two-sided, and all determinations of statistical significance of
differences were made at the α=0.10 level. Note that while differences between estimates
may be statistically significant, readers should be cautioned against making strong
conclusions in instances where the magnitudes of the differences are not substantive or
notable.
Individual t-tests or families of t-tests that involved variables with dichotomous
distributions – including IMDRs, URRs, and MARRs – did not require additional
adjustments. However, the families of t-tests that involved variables with more than two
response categories were considered to be simultaneous with other tests related to that
distribution, and therefore the familywise error rates had to be controlled using a multiple
comparison procedure. The Bonferroni multiple comparison procedure was used for
families of t-tests that pertained to variables with three or four response categories, and
the Bonferroni-Holm procedure was used for families of t-tests that pertained to variables
with five or more response categories.
Only one type of comparison was subjected to a different testing procedure than the ttest. To test whether a response distribution was dependent on the placement of the PPOB
questions, chi-square tests adjusted for the complex sample design with (n-1) degrees of
freedom, where n is the number of response categories in a given variable or recode, were
used. As with the t-tests, all determinations of statistical significance of differences in
overall response distributions were made at the α=0.10 level. Note that the chi-square
statistics were calculated only for variables and recodes with more than two responses;
this is because the chi-square test involving a dichotomous variable provides equivalent
results to the t-test. Therefore, the analysis of response distributions involving PPOB and

9

See Appendix M for definitions and formulae of the reliability measures used in this report.

12

language variables did not feature chi-square tests for the distributions of father’s BPOB,
mother’s BPOB, and non-English-language-spoken-at-home status. 10

4. LIMITATIONS
Control and Test CATI/CAPI workload assignments were not assigned using an
interpenetrated experimental design. That is, interviewers were allowed to administer
interviews for both Control and Test panel cases in the Content Test, in addition to ACS
cases. The potential risk of this approach is the introduction of a cross-contamination or
carry-over effect due to the interviewer administering multiple versions of the same
question item. Interviewers are trained to read the questions verbatim to minimize this
risk, but there still exists the possibility that an interviewer may deviate from the scripted
wording of one question version to another. This could potentially mask a treatment
effect from the data collected; however, this is less of a concern for the PPOB questions
since the two versions had the same wording but different locations on the questionnaire.
The CFU reinterview was not conducted in the same mode of data collection for
households that responded by mail or CAPI in the original interview since CFU
interviews were only administered using a CATI mode of data collection. As a result, the
data quality measures derived from the reinterview may include some bias due to the
differences in mode of data collection.
The Content Test mail mode questionnaires were limited in their ability to collect data on
households of large size; the data of only the first five persons in each household were
collected. This is different from the ACS mail questionnaire, which is designed to handle
larger household sizes. Because of this limitation, comparisons of ACS data to the
Content Test were restricted to the first five household members in both data sources for
mail mode respondents only. 11 The household size limitation was not present for the
CATI and CAPI modes of the Content Test, so no such restrictions on ACS data collected
via CATI/CAPI were necessary.
Respondents needed to provide a telephone number in the original Content Test interview
or the Census Bureau had to be able to find a telephone number for that unit through
reverse address look-up to be included in the CFU interview. As a result, 18.4 percent of
the responding households from the original interview were not eligible for the CFU
reinterview. We did not have the same respondent in the CFU that we had in the original
interview for 9.1 percent of the CFU cases. This means that differences between the
original interview and the CFU for these cases could be due in part to having different
people answering the questions.

10

Likewise, because the L-fold index of inconsistency for a dichotomous variable is equivalent to the
individual index of inconsistency for each of that variable’s responses, the reliability analysis for these
three variables will not feature comparisons of the L-fold statistic between the two treatments.
11
The CPS does not have a mail mode component of data collection, so none of the CPS-based estimates in
this report were subjected to the household size restriction.

13

The Content Test did not include the weighting adjustments for seasonal variations in
ACS response patterns, nonresponse bias, and under-coverage bias. The CFU portion of
the Content Test did include a unit nonresponse adjustment for those Content Test cases
that responded to the Content Test but failed to respond to the CFU. As a result, the
statistics derived from the Content Test data do not provide the same level of inference as
the ACS to the entire population of housing units and persons in the contiguous United
States.
There were also two notable differences between the 2010 ACS questionnaire and the
Content Test questionnaire. To avoid increasing the page size of the Content Test
questionnaire, the second halves of both the school enrollment and residence one year
ago questions were removed to make room for the PPOB questions. 12 While the Content
Test provides data to assess the impact of including the PPOB questions on the quality of
the data from the first halves of the school enrollment and residence one year ago
questions, their impact on the second halves of these questions remains unknown.

5. CRITICAL RESEARCH QUESTIONS AND RESULTS
5.1 Response to the Content Test and Content Follow-Up
Table 1 shows the unit response rates for each of the modes of data collection and all
modes combined (excluding Content Follow-up) by the Control and Test panels. 13 The
comparison between Control and Test shows that respondent participation was similar for
both Control and Test for each of the modes of data collection and all modes combined,
with the exception of the CATI mode. The Test treatment produced a CATI rate of
response that is 3 percentage points lower than that of the Control treatment. The
increase in response due to the Test treatment for the CATI mode of data collection
cannot be explained other than by random occurrence, given that the conditions affecting
unit response were equivalent between the Test and Control groups.

5.2 Are the response distributions of parental place of birth and generational
status (i.e., first, second, and third-or-higher generation) comparable to
existing data sources?
The basic CPS instrument includes questions on both father’s and mother’s place of birth
that are very similar to the PPOB questions used in the Content Test. 14 To determine
whether the PPOB questions appear to work correctly in the ACS, the response
distributions of father’s POB, mother’s POB, and generational status from the Content
12

The second half of the school enrollment questions asked: What grade or level was this person attending?
Respondents were provided check boxes and a write-in box to specify a grade/level. For the residence one
year ago question, the second half asked: Where did this person live one year ago? Respondents were
provided write-in lines to report their address.
13
All tables referenced in this report can be found in Appendix A.
14
The Current Population Survey is collected using only CATI and CAPI modes. The telephone and
personal interviews include questions that ask: 1) In what country was your father born? 2) In what country
was your mother born?

14

Test were compared with corresponding distributions derived from CPS data. Note that
these comparisons could not be statistically tested because the complete set of post-data
collection data processing methods that are standard practice for the production of both
CPS and ACS data, such as edit and imputation procedures, were not applied to the
Content Test data.
Two CPS datasets were chosen for comparison to the Content Test data because of
certain favorable characteristics. The September 2010 CPS basic dataset was
administered during the same time period as the 2010 ACS Content Test. The March
2010 CPS Annual Social and Economic Supplement (ASEC) dataset featured a larger
overall sample size (responses were collected in February, March, and April) and an
oversample of Hispanic respondents, leading to a more robust sample and a greater
likelihood of foreign-born representation in the sample.
Table 2 shows the distributions of generational status and father’s and mother’s WROB.
Overall, the distributions derived from both Content Test panels were very similar to
those derived from both of the CPS datasets. There are some small differences – for
instance, the segment of the population identified as the second generation was about one
percentage point higher in the Content Test distributions than in the CPS distributions –
but given the limited degree of post-processing in the Content Test data, none of these
differences were large enough to warrant concern.
Conclusion: The generational status and father’s and mother’s WROB distributions
derived from the Content Test data were very similar to the corresponding distributions
derived from the CPS data, which suggests that the PPOB questions appeared to function
correctly on the ACS questionnaire.

5.3 Which placement results in a lower missing data rate?
To determine whether the IMDRs of the PPOB questions were affected by the location of
these items on the questionnaire, the IMDRs for father’s and mother’s POB in both
panels of the Content Test were compared. Table 3 shows the IMDRs for both PPOB
questions in the Control and Test panels as well as the statistical significance of the
differences in IMDRs between the panels. The results indicated that, for both father’s
and mother’s POB, the IMDRs were significantly lower for the Control panel (6.9
percent and 6.0 percent, respectively) than for the Test panel (7.4 percent and 6.6 percent,
respectively).
As mentioned in Section 3.3.1, item missing data can take several forms, such as
uncodable responses. Because the PPOB questions yielded write-in data that was later
coded to valid response categories, it was important to know if uncodable responses were
among the PPOB missing data values and – if so – whether the PPOB question placement
had an effect upon the URRs for these questions.
Table 4 shows the estimates of the URRs for father’s POB, mother’s POB, and POB for
both Content Test panels and the differences between the Control and Test URRs. These
15

differences were tested for statistical significance in the same way as the IMDRs. The
URR for father’s POB was higher in the Test panel (0.2 percent) than in the Control
panel (0.1 percent), while the URR differences for both mother’s POB and POB were not
statistically significant.
The URRs from each of the PPOB questions were also compared to the URR from the
POB question within each Content Test panel (see Table 5). Because the intent was to
determine whether the URRs for the PPOB questions were strictly larger than – and not
simply different from – the URR for POB, the comparisons were subjected to one-sided
hypothesis tests, rather than two-sided tests. Within each panel, the comparisons
(father’s POB to POB, and mother’s POB to POB) were made simultaneously, and the
familywise error rate for each pair of comparisons had been controlled using the
Bonferroni multiple comparison method at the α=0.10 level. The URR of mother’s POB
was higher than the URR of POB regardless of treatment. Also, the URR of father’s
POB was higher than the URR of POB only in the Test panel. However, the significant
differences among the URRs were negligible (i.e., no greater than a tenth of a percentage
point).
Conclusion: The Control treatment yielded a lower item missing data rate for both
father’s and mother’s POB (6.9 percent and 6.0 percent, respectively) than the placement
used by the Test panel (7.4 percent and 6.6 percent, respectively). While the URR for
father’s POB was lower for the Test treatment, the magnitude of the difference between
the two panels was negligible.

5.4 Do the two placements have similar or different response distributions?
To determine whether the placement of the PPOB questions had an effect upon the
response distributions of father’s and mother’s POB, the percent distributions of each
question between the two treatments were compared. All three recodes for PPOB were
considered: BPOB, WROB, and SPOB. 15
Table 7 shows the Control and Test response distributions for both PPOB variables and
all three recodes, in addition to the results from the corresponding t-tests. For both
father’s and mother’s POB, all of the differences between the individual response
categories of the Control and Test Panels were not statistically significant, regardless of
which recode was used.
Table 6 shows the relevant chi-square statistics and corresponding test results of father’s
and mother’s POB under the Control and Test panels. The WROB recode did not yield
statistically significant differences for either PPOB variable, but the SPOB recode did
indicate evidence of significant overall differences between the two panels for both
PPOB variables. The discrepancy in test results among the WROB and SPOB recodes
appeared to indicate that the overall differences in PPOB distributions between the two
panels were more prevalent at lower levels of geographic focus. However, because the
individual response categories were not significantly different between the two panels for
15

See Section 3.3 for definitions of these recodes and rationale for their use.

16

the SPOB recodes, it was not apparent whether any particular response categories were
largely responsible for the significance of the differences in the overall response
distributions.
Conclusion: In general, there were no notable significant differences between the Control
and Test panels among the response distributions of father’s and mother’s POB,
regardless of recode used. Thus, the two placements yielded similar distributions of the
PPOB variables.

5.5 Which placement results in more reliable estimates?
To determine which PPOB question placement resulted in more reliable estimates of
father’s and mother’s POB, several reliability statistics were used to compare the Control
and Test panels. As with the distributional analysis, the reliability analysis was
conducted for both PPOB variables using all three recodes for both panels of the Content
Test.
Table 8 shows the GDRs for father’s and mother’s POB using all three recodes for the
two panels in addition to the statistical testing results of the differences between the
Control GDRs and Test GDRs for each response category. The results showed that, for
both PPOB variables, there were no significant differences in the GDRs between the
Control and Test panels for either the BPOB recode or the WROB recode. The results
also indicated that, for the mother’s SPOB recode, there were no significant differences
in GDRs between the two panels. However, for the father’s SPOB recode, the Mexico
category yielded a higher GDR in the Test panel (0.4 percent) than the Control panel (0.2
percent). Likewise, the “Other places” category yielded a higher GDR in the Test panel
(0.8 percent) than the Control panel (0.5 percent). Despite the significance of these two
differences, the magnitudes of the estimates involved – all less than 1 percent – were
negligible.
Table 9 shows the IoIs for father’s and mother’s POB using all three recodes for the two
panels in addition to the statistical testing results of the differences between the Control
IoI and Test IoI for each response category. The results showed that, for both PPOB
variables, there were no significant differences in indices of inconsistency between the
Control and Test panels for all three PPOB recodes.
Table 10 shows the L-fold indices of inconsistency for father’s and mother’s POB using
all three recodes for the two panels in addition to the statistical testing results of the
differences between the Control L-fold indices and Test L-fold indices. 16 The results
showed that, for both PPOB variables, there were no significant differences in L-fold
indices between the Control and Test panels for either the BPOB recode or the WROB
recode. The results also showed that, for the mother’s SPOB recode, the Control and
Test L-fold indices were not significantly different. However, for the father’s SPOB
16

For dichotomous variables such as the father’s and mother’s BPOB recodes, the L-fold index (and
corresponding test of difference in L-fold estimates) is equivalent to the IoI for each response category of
the variable.

17

recode, the Test L-fold index (3.4 percent) was higher than that for the Control panel (2.3
percent). This suggested that, overall, responses to the father’s POB question – when
distributed according to the father’s SPOB recode – were more inconsistent under the
Test panel than under the Control panel. The comparisons between the individual IoIs
for father’s SPOB did not indicate which categories caused the difference in overall
inconsistency to be significant between the Control and Test panels.
Conclusion: In general, there were no notable significant differences in the reliability
measures of father’s and mother’s POB between the two placements used by the Control
and Test panels. Thus, neither placement resulted in more reliable estimates than the
other.

5.6 Does changing the placement of the parental place of birth questions from
before to directly after the ancestry question affect the item missing data rate,
response distribution, or reliability for the ancestry question?
To determine whether the IMDR of the ancestry question was affected by the location of
the PPOB questions on the questionnaire, IMDRs for first reported ancestry in both
panels of the Content Test were compared. 17 Table 3 shows the IMDRs for the ancestry
question under the Control and Test panels as well as the statistical significance of the
differences between the panels. The results indicated that the IMDRs were significantly
lower for the Test panel (13.3 percent) than for the Control panel (14.5 percent).
Table 11 shows the MARRs for the ancestry question under the Control and Test panels
as well as the statistical significance of the differences between the panels. The results
indicated that respondents were more likely to report more than one ancestry under the
Control panel (28.2 percent) than under the Test panel (26.8 percent). So, the Test panel
(in which the PPOB questions were placed after ancestry) resulted not only in lower
IMDRs for ancestry than the Control panel, but also in lower incidence rates of multiple
ancestry responses. The multiple response finding lends support to the idea that placing
the PPOB questions prior to ancestry (as in the Control panel) may aid respondents in
thinking about the backgrounds of the household members, thereby increasing the
amount of information they provide when asked about ancestry. However, further
analysis is necessary to determine whether this additional information would represent
improved ancestry data.
To determine whether the placement of the PPOB questions had an effect upon the
response distributions of first and second ancestry reported, the percent distributions of
each question between the two treatments were compared. Both ancestry recodes were
considered: regional ancestry and selected ancestry. 18 Tables 13 and 14 show the
Control and Test response distributions for both ancestry variables and both recodes, in
addition to the results from the corresponding statistical tests of the differences between
the Control and Test panels. The results indicated that, for both first and second ancestry
17

Due to the voluntary response nature of the second ancestry reported variable, the authors were
concerned only with the item missing data rate of the first ancestry reported.
18
See Section 3.3 for definitions of these recodes and rationale for their use.

18

reported, all of the percentages attributed to the individual response categories were not
significantly different between the two panels, regardless of which recode was used.
Table 12 shows the relevant chi-square statistics and corresponding test results of first
and second ancestry reported for both the Control and Test panels. The selected ancestry
recode was not significantly different for either ancestry variable. By comparison, the
regional ancestry recode was significantly different between the two panels but only for
the first ancestry reported variable. The discrepancy in test results among the regional
and selected ancestry recodes appeared to indicate that the overall differences in ancestry
distributions between the two panels were more prevalent at the higher levels of
geographic focus. However, because the individual response categories were not
significantly different between the two panels for the selected ancestry recodes, it was not
apparent whether any particular response categories were largely responsible for the
significance of the differences in the overall response distributions.
To determine which PPOB question placement resulted in more reliable estimates of first
and second ancestry reported, the differences in the reliability measures (GDRs, IoIs, and
L-fold indices) of the ancestry variables for both Content Test panels were tested for
statistical significance. As with the distributional analysis, the reliability analysis was
conducted for both ancestry variables using both recodes for both panels of the Content
Test. Tables 15 and 16 show the GDRs for first and second ancestry reported using both
recodes for the two panels in addition to the statistical testing results of the differences
between the Control GDR and Test GDR for each response category. The results showed
that, for both ancestry variables, there were no significant differences in GDRs between
the Control and Test panels for either the regional ancestry recode or the selected
ancestry recode.
Tables 17 and 18 show the IoIs for first and second ancestry reported using both recodes
for the two panels in addition to the statistical testing results of the differences between
the Control IoI and Test IoI for each response category. The results showed that, for
second ancestry reported, there were no significant differences in IoIs between the
Control and Test panels for either of the two recodes. The results also showed that, for
first ancestry reported, there were no significant differences between the two panels for
only the regional ancestry recode. The IoIs of all but one of the response categories for
selected first ancestry reported were not significantly different between the two panels. 19
Table 19 shows the L-fold indices for first and second ancestry reported under both
recodes for the two panels in addition to the statistical testing results of the differences
between the Control and Test L-fold indices. The results showed that, for both ancestry
variables, there were no significant differences in L-fold indices of inconsistency between
the Control and Test panels, regardless of the recode used.
Conclusion: In general, there were no notable significant differences in the response
19

The IoI for the American Indian category was higher in the Test panel (59 percent) than in the Control
panel (46 percent). However, this result was likely due to the relatively small population reporting
American Indian ancestry, as IoIs are particularly sensitive for rare populations (Singer and Ennis, 2003).

19

distributions or reliability measures of first or second reported ancestry between the two
placements of PPOB used by the Control and Test panels. However, the placement used
by the Test panel resulted in a lower item missing data rate for first reported ancestry
(13.3 percent) than the placement used by the Control panel (14.5 percent).

5.7 Does the placement of the parental place of birth questions directly before
the school enrollment question affect the item missing data rate, response
distribution, or reliability for the school enrollment question?
To determine whether the IMDR of the school enrollment question was affected by the
location of the parental place of birth questions on the questionnaire, the IMDRs for the
school enrollment variable in both panels of the Content Test were compared. Table 3
shows the IMDRs for the school enrollment question under the Control and Test panels
as well as the statistical significance of the differences between the panels. The results
indicated that the IMDR was significantly lower for the Test panel (4.8 percent) than for
the Control panel (5.4 percent). In other words, when the PPOB questions preceded the
school enrollment question (as in the Control questionnaire), the school enrollment
IMDR was higher than when the PPOB questions were placed after the school enrollment
question (as in the Test questionnaire).
To determine whether the placement of the PPOB questions had an effect upon the
response distributions of school enrollment, the percent distributions between the two
treatments were compared. Table 21 shows the Control and Test response distributions
for the school enrollment variable in addition to the results from the corresponding
statistical tests. The results indicated that all of the percentages attributed to the
individual response categories were not significantly different between the Control and
Test panels.
Table 20 shows the relevant chi-square statistic and test result of the school enrollment
variable under the Control and Test panels. The overall difference in response
distributions between the two panels was statistically significant. However, the
interpretation of these results is difficult as the t-tests did not indicate significant
differences among any of the response categories for the school enrollment variable and
it was not apparent whether any particular response categories were largely responsible
for the significance of the differences in the overall response distributions.
To determine which PPOB question placement resulted in more reliable estimates of
school enrollment, the differences in the reliability measures (GDRs, IoIs, and L-fold
indices) of the school enrollment variables for both Content Test panels were tested for
statistical significance. Table 22 shows the GDRs of the school enrollment variable for
the two panels in addition to the statistical testing results of the differences between the
Control GDR and Test GDR for each response category. The results showed that there
were no significant differences in gross difference rates between the Control and Test
panels for the school enrollment variable.
Table 23 show the IoIs of the school enrollment variable for the two panels in addition to
20

the statistical testing results of the differences between the Control IoI and Test IoI for
each response category. The results showed that there were no significant differences
among the indices of inconsistency between the Control and Test panels for the school
enrollment variable. Table 24 shows the L-fold index of inconsistency for the school
enrollment variable in the two panels in addition to the statistical testing results of the
differences between the Control and Test L-fold indices. The results showed that there
was no significant difference in the L-fold indices of inconsistency for the school
enrollment variable between the Control and Test panels.
Conclusion: In general, there were no notable significant differences in the response
distributions or reliability measures of school enrollment status between the two
placements of PPOB used by the Control and Test panels. However, the placement used
by the Test panel resulted in a lower item missing data rate for school enrollment status
(4.8 percent) than the placement used by the Control panel (5.4 percent).

5.8 Does the placement of the parental place of birth questions directly before
the language spoken questions affect the item missing data rates, response
distributions, or reliability for the language questions?
To determine whether the IMDR of the language questions (non-English-languagespoken-at-home status, language spoken at home, and English speaking ability) were
affected by the location of the PPOB questions on the questionnaire, the IMDRs for the
three language questions in both panels of the Content Test were compared. Table 3
shows the IMDRs for the language questions under the Control and Test panels as well as
the statistical significance of the differences between the panels. The results indicated
that, for the non-English-language-spoken-at-home status question, the IMDR was not
significantly different between the Test panel (4.8 percent) and the Control panel (5.1
percent), while for the language spoken at home and English speaking ability questions,
the IMDRs were lower for the Control panel (4.9 percent and 1.5 percent, respectively)
than for the Test panel (6.9 percent and 2.1 percent, respectively).
To determine whether the placement of the PPOB questions had an effect upon the
response distributions of the language questions, the percent distributions of each
question between the two treatments were compared. 20 Table 21 shows the Control and
Test response distributions for the three language variables, in addition to the results from
the corresponding statistical tests. For all three language variables, all of the percentages
attributed to the individual response categories were not significantly different between
the Control and Test panels.
Table 20 shows the relevant chi-square statistics and corresponding test results of
selected language spoken at home and English speaking ability under the Control and

20

For the distributional and reliability analyses, the language spoken at home variable was recoded using
the selected language spoken at home recode, defined in Section 3.3.1.

21

Test panels. 21 The results indicated that, overall, the response distribution of Englishspeaking ability was not significantly different between the two panels. However, the
overall response distribution of the selected language spoken at home recode was
significantly different between the two panels. However, the t-tests for the selected
language spoken at home recode did not indicate significant differences among any of the
response categories and it was not apparent whether any particular response categories
were largely responsible for the significance of the differences in the overall response
distributions.
To determine which PPOB question placement resulted in more reliable estimates of the
three language questions, the differences in reliability measures (GDRs, IoIs, and L-fold
indices) of the language variables for both Content Test panels were tested for statistical
significance. As with the distributional analysis, the reliability analysis was conducted
for the three language variables for both panels of the Content Test. Table 22 shows the
GDRs for the three language variables under both recodes for the two panels in addition
to the statistical testing results of the differences between the Control GDR and Test
GDR for each response category. The results showed that, for all three language
variables, there were no significant differences in GDRs between the Control and Test
panels.
Table 23 shows the IoIs for the three language variables for the two panels in addition to
the statistical testing results of the differences between the Control IoI and Test IoI for
each response category. The results showed that, for all three language variables, there
were no significant differences in IoIs between the Control and Test panels. Table 24
shows the L-fold indices of inconsistency of the three language variables for the two
panels in addition to the statistical testing results of the differences between the Control
and Test L-fold indices. The results showed that, for all three language variables, there
were no significant differences in L-fold indices between the Control and Test panels.
Conclusion: In general, there were no notable significant differences in the response
distributions or reliability measures of any of the language questions between the two
placements of PPOB used by the Control and Test panels. However, the placement used
by the Control panel resulted in lower IMDRs for non-English-language-spoken-at-home
status and English speaking ability (4.9 percent and 1.5 percent, respectively) than the
placement used by the Test panel (6.9 percent and 2.1 percent, respectively).

5.9 Summary of Critical Analysis
The core selection criteria (see Section 2) for the PPOB questions comprised three major
elements: (1) do the PPOB questions “work” on the ACS questionnaire, providing
reasonable and reliable data; (2) does placement of the PPOB questions on the
questionnaire affect the data produced by the PPOB questions; and (3) does the
placement of the PPOB questions affect the quality of the data produced by the ancestry,
21

Due to its dichotomous nature, the comparison in overall response distributions of the non-Englishlanguage-spoken-at-home status variable between the Control and Test panels was not subjected to a chisquare test.

22

school enrollment, and language questions? The results of the critical analysis (see
Sections 5.2 through 5.8) were used to address the core selection criteria and to determine
how the PPOB question placement affected the quality of data obtained by the PPOB
questions and the proximal questions.
To address Criterion 1, the distributions of father’s WROB, mother’s WROB, and
generational status were compared between the two Content Test panels and the two CPS
data sources (see Section 5.2). The results indicated that response distributions for all
three variables produced by the Content Test data closely resembled those produced by
the existing CPS data, thereby indicating that the PPOB questions appeared to provide
reasonable and reliable data on the ACS questionnaire.
To address Criterion 2, the IMDRs, response distributions, and reliability measures of
several recodes of varying geographic detail for father’s and mother’s POB were
compared between the two Content Test panels (see Sections 5.3 through 5.5). The
results indicated that, in general, the response distributions and reliability measures of the
PPOB questions did not differ significantly between the two panels. However, the
IMDRs for both PPOB questions were lower for the Control panel than for the Test
panel. Overall, the placement of the PPOB questions did not have any notable impact on
the quality of the data produced.
To address Criterion 3, the IMDRs, response distributions, and reliability measures of the
proximal variables were compared between the two Content Test panels (see Sections 5.6
through 5.8). The results indicated that, in general, the response distributions and
reliability measures of the proximal variables did not differ significantly between the two
panels. However, the IMDR comparisons yielded mixed results – IMDRs were lower for
the Control panel among the ancestry, school enrollment status, and non-Englishlanguage-spoken-at-home status variables but lower for the Test panel among the
language spoken at home and English speaking ability variables. Overall, the placement
of the PPOB questions had minimal impact on the surrounding variables.
In summary, the PPOB questions appeared to work properly on the ACS questionnaire
and the question placement had minimum impact upon the data quality of the PPOB
questions and the proximal variables (aside from the IMDRs). While there was no strong
evidence to support one placement over the other, the results of the analyses did suggest
that the placement used by the Control treatment may yield results more favorable for the
PPOB variables.

6. SUPPLEMENTAL RESEARCH QUESTIONS AND RESULTS
6.1 For each mode of data collection, do the two placements have differential
item missing data rates, response distributions, or reliability of the data?
Both panels of the Content Test were administered across three modes of data collection:
mail, CATI, and CAPI. Table 25 shows the weighted distribution of the household
population across mail and non-mail (i.e., CATI and CAPI) response modes in the
23

Control and Test panels. In both panels, there were more mail respondents (57 percent in
each panel) than non-mail respondents (43 percent in each panel). Note that the modal
distribution was not random, due to the mixed-mode survey design.
As previously noted (Section 5.3), both of the PPOB questions had lower IMDRs under
the Control panel than the Test panel. To determine whether this result was consistent
across mode types, the IMDRs for father’s and mother’s POB were compared between
the Control and Test panels while restricting the population universes to 1) mail
respondents and 2) non-mail respondents. Table 26 shows the results of the by-mode
IMDR comparisons. For both father’s and mother’s POB, the Control panel yielded
lower IMDRs (8.0 percent and 6.9 percent, respectively) than the Test panel (8.6 percent
and 8.0 percent, respectively) for mail respondents. However, for both father’s and
mother’s POB, the differences in IMDRs between both panels were not statistically
significant for non-mail respondents. Therefore, it appeared that the significant
difference in IMDRs for the PPOB questions was driven by the mail respondents.
In general, the response distributions of both PPOB questions were not significantly
different between the two panels (see Section 5.4). To determine whether this result was
consistent across mode types, the response distributions for father’s and mother’s POB
were compared between the Control and Test panels while restricting the population
universes by mode as previously described. To avoid potential data sparseness issues, the
PPOB-by-mode analysis was limited to only the WROB recode.
Table 28 shows the response distributions of the PPOB variables by mode for the Control
and Test panels, in addition to the differences in percentages of response categories
between the two panels and the statistical testing results of the individual response
categories. The results indicated that, for both father’s and mother’s WROB, there were
no significant differences between the Control and Test panels among any of the response
categories, regardless of mode type. Table 27 shows the chi-square statistics and test
results for each PPOB variable and mode type that correspond to the comparisons of
overall response distributions between the two panels. The results showed that, for
father’s and mother’s WROB, there were no significant differences in the response
distributions between the Control and Test panels, regardless of mode type.
As previously noted (Section 5.5), both of the PPOB variables did not exhibit notable
differences in reliability measures between the Control and Test panels. To determine
whether this result was consistent across mode types, the reliability measures for father’s
and mother’s POB were compared between the two panels while restricting the
population universes by mode as previously described. Again, the PPOB-by-mode
analysis was limited to only the WROB recode. Tables 30, 32, and 34 show the GDRs,
IoIs, and L-fold indices of father’s and mother’s WROB for each mode type under the
Control and Test panels, the differences in reliability measures between the two panels
for each response category, and the results of the corresponding statistical tests. The
results indicated that, for both PPOB variables, none of these differences were significant
between the two panels for all response categories, regardless of the mode type.

24

Conclusion: There were no notable significant differences among the response
distributions or reliability measures of father’s and mother’s POB between the two
placements of PPOB used by the Control and Test panels, regardless of the mode of data
collection. However, the placement used by the Control panel resulted in lower IMDRs
in the mail mode for both father’s and mother’s POB than the placement used by the Test
panel. When considering only the households whose data was collected by CATI or
CAPI, there were no significant differences in IMDRs for either PPOB question between
the two placements used by the Control and Test panels.

6.2 For each mail response stratum, do the two placements have differential
item missing data rates, response distributions, or reliability of the data?
Compared to the ACS sample design, the samples corresponding to the two panels of the
Content Test had an additional layer of stratification designed to ensure adequate
representation in geographic areas that had a history of either high or low mail response
rates. Table 25 shows the weighted distribution of the household population across mail
response strata in the Control and Test panels. In both panels, there were more
respondents from the high response area (75 percent in each panel) than from the low
response area (25 percent in each panel). Note that the distribution by response stratum
was not random, but purely by design of the sample (which incorporated an oversample
of households in the low response area).
As previously noted (Section 5.3), both of the PPOB questions had lower IMDRs under
the Control panel than the Test panel. To determine whether this result was consistent
across mail response strata, the IMDRs for father’s and mother’s POB were compared
between the Control and Test panels while restricting the population universes to
respondents 1) living in high response areas and 2) living in low response areas. Table
26 shows the results of the by-stratum IMDR comparisons. For father’s POB, the
Control panel yielded lower IMDRs (6.4 percent) than the Test panel (7.2 percent) when
only considering the respondents living in high response areas. For mother’s POB, the
differences in IMDRs between panels were not statistically significant for those living in
high response areas. Also, for both father’s and mother’s POB, the differences in IMDRs
between panels were not statistically significant, regardless of mail response stratum.
In general, the response distributions of both PPOB questions were not significantly
different between the two panels (see Section 5.4). To determine whether this result was
consistent across mail response strata, the response distributions for father’s and mother’s
POB were compared between the Control and Test panels while restricting the population
universes by mail response stratum as previously described. To avoid potential data
sparseness issues, the PPOB-by-stratum analysis was limited to only the WROB recode.
Table 29 shows the response distributions of the PPOB variables by mail response
stratum for the Control and Test panels, in addition to the differences in percentages of
response categories between the two panels and the statistical testing results of the
individual response categories. For both father’s and mother’s WROB, there were no
significant differences between the Control and Test panels among any of the response
categories for respondents who lived in high response areas. However, for both father’s
25

and mother’s WROB, the parents were more likely to be born in Puerto Rico or the U.S.
territories in the Test panel (2.3 percent and 2.2 percent, respectively) than in the Control
panel (1.8 percent and 1.7 percent, respectively) for respondents living in low response
areas. 22 Nonetheless, the magnitude of these differences, coupled with the small number
of respondents in this category for both panels under the low response stratification,
rendered this finding to be negligible.
Table 27 shows the chi-square statistics and test results for each PPOB variable and mail
response stratum that correspond to the comparisons of overall response distributions
between the two panels. For father’s and mother’s WROB, there were no significant
differences in the response distributions between the Control and Test panels, regardless
of response area.
As previously noted (see Section 5.5), both of the PPOB variables did not exhibit notable
differences in reliability measures between the Control and Test panels. To determine
whether this result was consistent across mail response strata, the reliability measures for
father’s and mother’s POB were compared between the two panels while restricting the
population universes by response area as previously described. Again, the PPOB-bystratum analysis was limited to only the WROB recode. Tables 31, 33, and 34 shows the
GDRs, IoIs, and L-fold indices of father’s and mother’s WROB for each mail response
area under the Control and Test panels, the differences in reliability measures between
the two panels for each response category, and the results of the corresponding statistical
tests. The results indicated that, for both PPOB variables, none of these differences were
significant between the two panels for all response categories, regardless of the response
area.
Conclusion: In general, there were no notable significant differences among the response
distributions or reliability measures of father’s and mother’s POB between the two
placements of PPOB used by the Control and Test panels, regardless of the mail response
stratum. However, the placement used by the Control panel resulted in lower IMDRs in
the high response area stratum for both father’s and mother’s POB than the placement
used by the Test panel. When considering only the households in the low response area
stratum, there were no significant differences in IMDRs for either PPOB question
between the two placements used by the Control and Test panels.

6.3 Does either placement elicit respondent or interviewer behaviors that may
contribute to interviewer or respondent error?
The behavior coding and analysis for all topics in the 2010 ACS Content Test was
conducted by the Center for Survey Management (CSM) using the Census Bureau’s new
Computer Audio Recorded Interview (CARI) system. In this system, a sample of
recorded CATI and CAPI interviews were used to assign standardized codes to observed
behaviors among field representatives and respondents for selected questions in both
panels of the Content Test. The occurrence rates of these behaviors were compared
22

The differences in percentages between the Control and Test panels were not statistically significant
among the other five categories for both PPOB variables.

26

between the Control and Test panels using two-sided t-tests with statistical significance
determined at the α=0.10 level. Details about the methodology involved in the CARI
analysis can be found in the CSM report (Goerman and Pascale, 2011).
For the PPOB questions, there were no significant differences in either interviewer or
respondent behavior between the Control and Test panels of the Content Test. For the
ancestry question, however, both interviewer and respondent behavior rates were “better”
under the Test panel than under the Control panel. The analysts also observed that the
behavior patterns for the PPOB questions under both panels were generally as expected.
For example, interviewers tended to stray from the exact text of the mother’s POB
question more often than for father’s POB, likely due to the order of questions (i.e.,
father’s POB precedes mother’s POB in the interview). Also, as the person number of
sampled households increased, the rates of exact readings tended to drop for both PPOB
questions, whereas the rates of response verifications and question skips tended to rise.

6.4 Summary of Supplemental Analysis
The results of the supplemental analysis indicated that the patterns of differences in
IMDRs, response distributions, and reliability measures of the PPOB variables between
the Control and Test panels of the Content Test were generally not dissimilar between
respondents from different mode types or from different mail response strata. Likewise,
there was little evidence of differences in interviewer or respondent behaviors between
the two treatments that would adversely affect the data quality of father’s or mother’s
POB. Overall, the supplemental analysis provided no findings that contradicted the
results of the critical analysis or complicated the selection criteria.

7. ADDITIONAL RESEARCH TOPICS AND RESULTS
This section of the report provides information about additional research that, while
outside the scope of the original analysis plan, addressed follow-up questions asked by
Census Bureau and OMB representatives during a series of debriefing meetings at Census
Bureau headquarters in Suitland, Maryland in Summer 2011. The first topic focused on
the reasonableness of IMDRs and response distributions of the proximal variables
(ancestry, school enrollment, and language) derived from Content Test data when
compared with corresponding rates and estimates derived from ACS data. The second
topic focused on whether or not the proximal variables were affected by response mode
within the Content Test data. In the end, the results of this additional analysis did not
yield results that contradicted the results of the critical analysis. For the sake of
completeness, the details of this additional research are presented below.

7.1 Comparison of Content Test Data to ACS Data for Proximal Variables
As part of the original analysis plan, the response distributions of father’s and mother’s
POB were compared between the two panels of the Content Test and two different CPS
datasets in order to ascertain a general sense of data reasonableness derived from the
PPOB variables on the Content Test questionnaires. During the debriefing meetings, it
27

was inquired whether similar analyses could be conducted to assess the data
reasonableness of the proximal variables in the Content Test – including ancestry, school
enrollment, and language questions, as well as several other questions within the vicinity
of the PPOB items on the questionnaire. The main reason for this additional analysis was
because, in specific terms of the PPOB project, neither of the two panels represented a
true “control” – the questionnaires for both panels included the PPOB questions, and so
the true effect of the presence of PPOB upon the proximal questions could not be
determined since there was no panel in which the PPOB questions were absent. Without
such a panel on the Content Test, the next best alternative would be to compare the
IMDRs and response distributions of proximal variables in the two Content Test panels to
the corresponding variables in the existing ACS data.
The list of proximal variables to be examined included the ancestry, school enrollment,
and language questions and was expanded to include educational attainment, first field of
degree, mobility status, the set of questions on health insurance coverage types, and the
two questions on disability status. 23 IMDRs were produced for all of these variables
using data from both the Control and Test panels of the Content Test in addition to data
from the 2009 ACS. 24 Due to time constraints, response distributions were produced
only for the core proximal variables using these three data sources with the intention to
produce distributions for the other variables if the results warranted additional analysis.
However, because the 2009 ACS data had been fully processed and prepared for public
use, the IMDRs could not be calculated, as all instances of missing data in the raw files
had since been allocated according to Census Bureau edit and imputation procedures.
Instead, allocation rates for the variables under consideration were used; while not
equivalent to the IMDR, the allocation rate is a reasonable analogue for this type of
comparison. Note that the comparisons between the IMDRs and allocation rates could
not be statistically tested because the complete set of post-data collection data processing
methods were not applied to the Content Test data (see Section 5.2). 25
Table 35 shows the allocation rates derived from the 2009 ACS in addition to the IMDRs
derived from the Control and Test panels of the Content Test. The variables are listed in
the order they appear on the ACS and Content Test questionnaires. The ACS allocation
rates and the Content Test IMDRs were, generally speaking, at the same or similar levels,
and the differences between the ACS allocation rates and the IMDRs from either Content
Test panel were no larger a few percentage points. There were two exceptions. First, for
field of degree, the Content Test IMDRs (2.9 percent for Test and 2.5 percent for
Control) were actually lower than the ACS allocation rate (9.0 percent), which can likely
23

The population universe of educational attainment comprised those persons age 3 years or older. The
population universe of mobility status comprised those persons age 1 year or older. All other variables
listed had no further restrictions of their population universes.
24
At the time of analysis, the dataset for the 2010 ACS had not yet been finalized and was unavailable.
Also, to be more comparable with the Content Test data, households that responded to the ACS by mail
were restricted such that only the first five household members were included in the analysis and group
quarters data was excluded from the analysis.
25
In addition to the lack of edit and allocation procedures, the Content Test did not incorporate a Failed
Edit Follow-up (FEFU) operation, which further limits the comparability of Content Test IMDRs to ACS
allocation rates.

28

be attributed to the edit and allocation procedures in the ACS. Second, for first ancestry
reported, the allocation rate could not be calculated because missing data were coded as
“not reported”; however, the percent not reported was similar to the two Content Test
IMDRs. Therefore, it was concluded that the IMDRs of proximal variables in the
Content Test were more or less similar to allocation rates of the same variables in the
ACS.
Tables 37, 38, and 39 show the response distributions derived from the 2009 ACS as well
as the two Content Test panels. 26 In general, none of the variables appeared to have
notably different distributions between the ACS data and either of the Content Test
panels.
Conclusion: For both placements, the results of the additional analysis suggest that the
IMDRs and response distributions for variables following the PPOB questions in the
Content Test compared favorably with those derived from ACS data. Thus, the presence
of the PPOB questions did not adversely affect the quality of the data produced by the
surrounding questions.

7.2 Additional Analysis of Proximal Variables by Mode in Content Test Data
As part of the original research plan, the supplementary questions included an analysis of
the effects of response mode on differences in the IMDRs, response distributions, and
reliability measures of father’s and mother’s POB between the Control and Test panels.
This analysis was subsequently expanded to determine whether the ancestry, school
enrollment, and language questions were affected by response mode. IMDRs by
response mode of first ancestry reported, school enrollment, non-English-languagespoken-at-home status, language spoken at home, and English speaking ability for both
panels of the Content Test were produced. 27 In addition, by-mode response distributions
were also produced for all of these variables as well as second ancestry reported for the
Control and Test panels. Due to time constraints, only the regional ancestry recodes were
examined for the two ancestry variables and additional reliability analysis was not
conducted. The differences between the Control and Test IMDRs and distributions were
tested for statistical significance according to the same procedures described in Sections
5.6 to 5.8.
Table 36 shows the IMDRs for the two panels, the differences in the IMDRs between the
two panels, and the statistical significance of those differences. The mail IMDR for
ancestry was lower under the Test panel (19.0 percent) than under the Control panel (21.1
percent), but the non-mail IMDRs for ancestry were not significantly different between
the two panels. Likewise, the mail IMDR for school enrollment was lower under the Test
panel (6.2 percent) than under the Control panel (7.2 percent), but the non-mail IMDRs
for school enrollment were not significantly different between the two panels. Also, the
mail IMDRs for language spoken at home and English speaking ability were lower under
26

For the sake of brevity, only the regional ancestry recode was used on the first and second ancestry
reported variables for this segment of analysis.
27
IMDRs were not constructed for second ancestry reported; see footnote 17.

29

the Control panel (12.1 percent and 3.3 percent, respectively) than under the Test panel
(17.1 percent and 4.9 percent, respectively), but the non-mail IMDRs for both variables
were not significantly different between the two panels. Finally, the IMDRs for nonEnglish language spoken at home status were not significantly different between the two
panels, regardless of response mode. For each of these variables, the patterns of IMDR
differences for the mail mode populations mimicked those of the total (all modes)
populations such that, when a difference between the Control and Test IMDRs was
significant, that difference appeared to be driven largely by mail mode responses.
Table 40 shows the response distributions of the regional first ancestry recode and the
regional second ancestry recode by mode for both the Control and Test panels, the
differences in estimated percentages of individual response categories between the two
panels, and the statistical significance of those differences. Table 42 shows the chisquare statistics and test results corresponding to each between-treatment comparison. In
general, most of the differences among individual response categories for either of the
ancestry variables and either of the mode types between the two panels were not
significantly different. 28 However, the chi-square analysis indicated that, for both of the
regional ancestry recodes, the overall differences between the Control and Test
distributions were statistically significant for mail responses but not for CATI/CAPI
responses. This finding suggested that the significant differences in the response
distributions of the ancestry variables were largely due to the mail respondents.
Table 41 shows the response distributions of the school enrollment and language
variables by mode for both the Control and Test panels, the differences in estimated
percentages of individual response categories between the two panels, and the statistical
significance of those differences. 29 Table 42 shows the chi-square statistics and test
results corresponding to each by-treatment comparison. For all four variables, there were
no significant differences between the two panels among the individual response
categories, regardless of response mode. Similarly, the chi-square analysis indicated that,
for the school enrollment and English speaking ability variables, the overall differences
between the Control and Test distributions were not statistically significant, regardless of
response mode. However, for the selected language spoken at home recode, the overall
differences between the Control and Test distributions were statistically significant for
mail responses but not for CATI/CAPI responses. This finding suggested that the
significant differences in the response distributions of the ancestry variables were largely
due to the mail respondents.
Conclusion: When the IMDRs and response distributions of the ancestry, school
enrollment, and language questions were differentiated by response mode, the statistical
significance patterns of by-treatment differences corresponding to the mail respondents
28

The Test estimate of the “other, non-regional” response category (14.4 percent) of the region first
ancestry recode was larger than the Control estimate (12.6 percent). However, this result was likely due to
the combined effect of small sample sizes in the Content Test panels and the small population of persons
reporting a non-regional ancestry living in the United States.
29
The selected language spoken at home recode was applied to the language spoken at home variable, as it
was in Section 5.7.

30

were similar to those of all respondents, whereas the differences corresponding to the
CATI/CAPI respondents were generally not statistically significant. This suggested that
the mail respondents were largely responsible for the significant differences in IMDRs
and response distributions of the proximal variables.

7.3 Summary of Additional Analysis
The results of the additional analysis indicated that the IMDRs and response distributions
of the expanded list of questions in close proximity to PPOB in either panel of the
Content Test did not appear to have notable differences from the rates and estimates
derived from the same questions as they appeared on the 2009 ACS questionnaire.
Furthermore, the results showed that the significant differences in IMDRs and response
distributions of the core proximal variables between the two panels for mail-mode
respondents were generally consistent with the total population, while these differences
were not significant among the CATI/CAPI respondents. In sum, the outcome of this
additional research appeared to suggest that: (1) the proximal variables were not
adversely affected by the presence of the PPOB variables; and (2) the significant
differences in IMDRs and response distributions of the proximal variables were largely
driven by mail respondents. While informative, the results of the additional analyses
provided support for and did not contradict the results of the critical research analysis.

31

8. SUMMARY
The second generation is growing rapidly. According to the Current Population Survey
(CPS), in 1996, there were 24.6 million people in the United States who had at least one
foreign-born parent; by 2000, there were 27.3 million; and by 2009, there were 33.0
million. Of these, close to half (45 percent) were less than 18 years of age. 30
Unfortunately, the principal source of information on the second generation, the CPS, is
limited to national-level analysis only, while current data needs are greatest at the subnational level, where immigrants are settling and populations are changing rapidly.
As the destinations of the foreign born have shifted from traditional gateway states such
as California, New York, Texas and Florida to smaller states such as Nevada, North
Carolina, and Georgia, many communities have experienced recent and unprecedented
growth in their immigrant populations. To improve the Census Bureau’s ability to
examine the adaptation and integration of immigrants and their children at the local level,
researchers and data users have petitioned for the inclusion of questions on parental place
of birth (PPOB) to the American Community Survey (ACS) questionnaire. Without
PPOB data, the second generation is indistinguishable from the third-or-higher
generation, which leads to a systematic underestimation of the total impact of recent
immigration. Only a large national sample, like that available in the ACS, can provide
planners and policymakers with the current data they need to access the impact that the
foreign born and their children have on communities, to develop and implement
programs, and to track the experience of smaller immigrant groups.
Questions on PPOB are important because they help to identify sub-groups of the
population, categorized as “first generation” (the foreign born), “second generation” (the
native-born children of at least one foreign-born parent), and “third-or-higher generation”
(native born with no foreign-born parents). This classification allows policymakers and
researchers to examine questions about the adaptation and integration of immigrants and
their descendants over multiple generations. In addition, the PPOB questions are useful
to examine the social and economic characteristics of the children of immigrants because
they clearly define the second generation. To establish whether PPOB would be a viable
addition to the ACS, two PPOB questions (In what country was your father born? In
what country was your mother born?) were added to the 2010 ACS Content Test in two
different placements on the questionnaire.
The results of the analysis indicated that the proposed questions on parental place of birth
appeared to function properly, as their Content Test response distributions were similar to
those derived from PPOB questions on the CPS. When the response distributions and
reliability measures of the PPOB questions and the surrounding questions were compared
between the two treatment panels of the Content Test, the results indicated no notable
significant differences. However, there were differences among the item missing data
rates; the PPOB questions had lower rates in the Control panel, while the surrounding
questions had mixed results (it should be noted that the differences, while significant,
30

Source: U.S. Census Bureau, Current Population Survey, March Supplement, 1996 and 2000, and the
ASEC Supplement, 2009

32

were not large enough to warrant concern about loss in data quality).
Further analysis indicated that the differences in item missing data rates were largely
present among the mail responses and not among the CATI/CAPI responses. Generally,
there were no other notable differences between panels for any of the questions by mode
or by mail response stratum. Also, there were no findings from the behavior analysis that
would suggest an influence upon the data quality of these questions. Furthermore, the
proximal questions in both panels were found to perform similarly to existing ACS data,
thereby suggesting that the data quality of those questions were not adversely affected by
the presence of the PPOB questions. 31
In summary, the supplemental and additional analysis steps did not yield results that
would conflict with the results of the critical analysis. Therefore, the Census Bureau
concluded that the PPOB questions worked well in either location on the questionnaire,
though the location used in the Control panel (between the year of entry and school
enrollment questions) was favored due to the lower item missing data rates for both
questions.
Based on the results of the 2010 ACS Content Test as well as legislative and
programmatic need for the PPOB data, the Census Bureau sent a memorandum to the
Office of Management and Budget (OMB) and the American Community Survey Office
recommending that: 1) two questions on parental place of birth – one for father’s and a
second for mother’s place of birth – be included in the ACS questionnaire (starting in
2013); 2) the format of the questions used be that tested by the 2010 ACS Content Test;
and 3) the placement of the questions used be that tested by the Control version of the
Content Test questionnaire.
The Census Bureau believes there is added value in collecting information about PPOB,
though some may feel that this topic is somewhat duplicative when collected in
connection with existing survey questions on race, Hispanic origin, and ancestry. Adding
the PPOB questions to the questionnaire in 2013 would be done as part of a multi-year
process to further examine the relationship of the data for these topics. The ACS data
would also be evaluated in connection with results from the 2010 Census Alternative
Questionnaire Experiment, and this combined research would be used in determining
recommendations for which questions would remain on the ACS at the conclusion of this
process. The Census Bureau plans to provide various opportunities for public comment
as well as dialogue with groups that are especially interested in these data as we refine the
plans and share results on this cross-topical research.

31

Note that the effect of PPOB question placement upon the secondary part of the school enrollment
question on the ACS could not be evaluated since it was not included on the Content Test questionnaire.

33

References
Brittingham, Angela and G. Patricia de la Cruz. (2004), “Ancestry: 2000”, Census 2000
Brief C2KBR-35, U.S. Census Bureau, Washington, D.C.
Goerman, Patti and Joanne Pascale. (2012, forthcoming), “ACS 2010 Content Test
Behavior Coding Report”, U.S. Census Bureau, Washington, D.C.
Hinsdale, Marjorie, Emily McFarlane, Stacey Weger, Alisú Schoua-Glusberg, and Jeffrey
Kerwin. (2009), “Cognitive Testing of the American Community Survey Content Test
Items”, RTI International, Research Triangle Park, NC.
Singer, Phyllis and Sharon R. Ennis. (2003), “Census 2000 Content Reinterview Survey:
Accuracy of Data for Selected Population and Housing Characteristics as Measured by
Reinterview”, Census 2000 Evaluation B.5, U.S. Census Bureau, Washington, D.C.
Thomas, Kathryn F., Tamara L. Dingbaum, and Henry F. Woltman. (1993), “Content
Reinterview Survey: Accuracy of Data for Selected Population and Housing
Characteristics as Measured by Reinterview”, 1990 Census of Population and Housing
Evaluation and Research Reports, 1990 CPH-E-1, U.S. Census Bureau, Washington,
D.C.

34

Appendix A: Tables
Table 1. Content Test Response Rate Comparisons Between Control and Test
Treatments
Mode
All modes
(CFU excluded)
Mail
CATI
CAPI
CFU

Test Standard
(%) error (%)

Control Standard
(%) error (%)

Test - Standard
Control (%) error (%)

1

Significant

95.7

0.2

95.4

0.2

-0.3

0.3

No

57.7
49.6
91.5
53.5

0.5
1.0
0.5
0.6

58.1
52.6
90.4
54.3

0.5
1.2
0.5
0.5

0.5
3.0
-1.1
0.8

0.7
1.5
0.7
0.7

No
Yes
No
No

Statistical significance of differences is determined at the α=0.10 significance level.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test, September to December 2010.

1

Table 2. Generational Status and Parental World Region of Birth by Selected
Data Sources

Variable/category
Generational status
Total
First generation
Second generation
Third-or-higher generation
Father's world region of birth
Total
United States
Puerto Rico / U.S. territories
Asia
Europe
Latin America
Other areas
Mother's world region of birth
Total
United States
Puerto Rico / U.S. territories
Asia
Europe
Latin America
Other areas

2010 ACS Content Test
Test
Control
treatment
treatment
Estimate
Estimate
(%)
(%)

2010 CPS
March
September
(ASEC)
(basic)
Estimate
Estimate
(%)
(%)

100.0
12.6
12.2
75.3

100.0
12.1
12.3
75.7

100.0
12.4
11.2
76.4

100.0
12.6
11.1
76.3

100.0
75.8
1.0
5.6
4.3
11.9
1.4

100.0
76.5
0.9
5.7
4.6
10.9
1.4

100.0
77.7
0.9
5.0
3.6
11.3
1.4

100.0
77.5
1.0
5.2
3.3
11.5
1.5

100.0
76.4
1.0
5.6
4.2
11.5
1.4

100.0
77.0
0.9
5.7
4.5
10.5
1.4

100.0
78.0
0.9
5.1
3.4
11.2
1.4

100.0
77.7
1.0
5.4
3.3
11.2
1.5

Source: U.S. Census Bureau, 2010 American Community Survey Content Test, September to December
2010; 2010 Current Population Survey Annual Social and Economic Supplement; and September 2010
Current Population Survey.

A-1

Table 3. Item Missing Data Rates of Selected Variables by Treatment

Variable

Test
Estimate
Standard
Number
(%) error (%)
(unweighted)

Control
Test Estimate
Standard Control Standard
Number
1
(%) error (%)
(%) error (%) Significant
(unweighted)

Father's place of birth
48,393
7.4
0.2
48,529
6.9
0.2
0.6
Mother's place of birth
48,393
6.6
0.2
48,529
6.0
0.2
0.6
First ancestry reported
48,393
13.3
0.3
48,529
14.5
0.3
-1.3
School enrollment status
46,812
4.8
0.2
46,812
5.4
0.2
-0.7
Speaks language other
45,652
than English at home
4.8
0.2
45,595
5.1
0.2
-0.2
Language spoken at
11,648
home
6.9
0.4
11,101
4.9
0.3
1.9
English speaking ability
11,648
2.1
0.2
11,101
1.5
0.2
0.6
1
Statistical significance of differences is determined at the α=0.10 significance level.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test, September to December 2010.

0.3
0.3
0.5
0.3

Yes
Yes
Yes
Yes

0.3

No

0.5
0.2

Yes
Yes

Table 4. Uncodable Response Rates of Parental Place of Birth by Treatment

Variable
Father's place of birth
Mother's place of birth
Person's place of birth

Test
Estimate Standard
(%) error (%)
0.2
0.2
0.1
-

Control
Test Estimate Standard Control Standard
1
(%) error (%)
(%) error (%) Significant
0.1
0.1
Yes
0.2
0.1
No
0.1
No

Dash (-) represents or rounds to zero.
1
Statistical significance of differences is determined at the α=0.10 significance level.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test, September to December 2010.

Table 5. Differences in Uncodable Response Rates of Parental Place of Birth
Within Treatment
Treatment/variable
Control
Person's place of birth - father's place of birth
Person's place of birth - mother's place of birth
Test
Person's place of birth - father's place of birth
Person's place of birth - mother's place of birth

Estimate
(%)

Standard
error (%)

-0.1

-

No
Yes

-0.1
-0.1

-

Yes
Yes

1

Significant

Dash (-) represents or rounds to zero.
1
For this family of one-sided hypothesis tests, the familywise error rate has been controlled using the
Bonferroni multiple comparison method at the α=0.10 level.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test, September to December
2010.

Table 6. Chi-Square Tests of Independence for Selected
Recodes of Parental Place of Birth
Variable/recode
Father's place of birth
World region of birth
Selected place of birth
Mother's place of birth
World region of birth
Selected place of birth

Rao-Scott
chi-square

Degrees of
freedom

5.65
16.43

5
10

No
Yes

5.73
16.60

5
10

No
Yes

1

Significant

Statistical significance of differences is determined at the α=0.10 significance level.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test,
September to December 2010.
1

A-2

Table 7. Response Distribution of Selected Recodes of Parental Place of Birth by Treatment
Test
Variable/recode

Standard
Estimate (%) error (%)

Control
Estimate Standard
(%) error (%)

Test Control Standard
1
(%) error (%) Significant

Father's place of birth
Unweighted sample size
44,238
(X)
44,571
(X)
(X)
(X)
(X)
Broad place of birth
Total
100.0
(X)
100.0
(X)
(X)
(X)
(X)
United States2
76.8
0.5
77.4
0.4
-0.6
0.6
No
Not United States
23.2
0.5
22.6
0.4
0.6
0.6
No
World region of birth
Total
100.0
(X)
100.0
(X)
(X)
(X)
(X)
United States
75.8
0.5
76.5
0.4
-0.6
0.6
No
Puerto Rico/U.S. territories
1.0
0.1
0.9
0.1
0.1
0.1
No
Asia
5.6
0.3
5.7
0.3
-0.1
0.4
No
Europe
4.3
0.2
4.6
0.2
-0.3
0.2
No
Latin America
11.9
0.4
10.9
0.3
1.0
0.5
No
Other areas
1.4
0.1
1.4
0.1
0.2
No
Selected place of birth3
Total
100.0
(X)
100.0
(X)
(X)
(X)
(X)
Canada
0.5
0.1
0.6
0.1
-0.1
0.1
No
China
0.8
0.1
1.0
0.1
-0.2
0.1
No
Cuba
0.5
0.1
0.5
0.1
No
Dominican Republic
0.5
0.1
0.5
0.1
0.1
No
El Salvador
0.7
0.1
0.6
0.1
0.1
0.1
No
India
0.9
0.1
1.1
0.1
-0.2
0.2
No
Italy
0.8
0.1
0.7
0.1
0.1
0.1
No
Mexico
7.6
0.3
6.6
0.2
1.0
0.4
No
Philippines
0.7
0.1
0.8
0.1
-0.1
0.1
No
Vietnam
0.7
0.1
0.5
0.1
0.1
0.1
No
Other places
86.5
0.5
87.2
0.3
-0.7
0.5
No
(X) Not applicable.
Dash (-) represents or rounds to zero.
1
For the two-sided hypothesis tests involving father's broad place of birth and mother's broad place of birth, statistical significance of
differences is determined at the α=0.10 significance level. For all other two-sided hypothesis tests in this table, the familywise error
rate has been controlled using the Bonferroni-Holm multiple comparison method at the α=0.10 level.
2
Includes Puerto Rico and U.S. island territories.
3
The countries included in both mother's and father's selected places of birth were based on parental place of birth data from the 2010
Current Population Survey and represent the ten largest groups. There are differences between the distributions: father's place of
birth includes Cuba and the Dominican Republic in its ten largest groups; mother's place of birth includes Germany and the United
Kingdom in its ten largest groups.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test, September to December 2010.

A-3

Table 7. Response Distribution of Selected Recodes of Parental Place of Birth by Treatment
(continued)
Test
Variable/recode

Standard
Estimate (%) error (%)

Control
Estimate Standard
(%) error (%)

Test Control Standard
1
(%) error (%) Significant

Mother's place of birth
Unweighted sample size
44,683
(X)
45,134
(X)
(X)
(X)
(X)
Broad place of birth
Total
100.0
(X)
100.0
(X)
(X)
(X)
(X)
United States2
77.4
0.5
77.9
0.4
-0.5
0.6
No
Not United States
22.6
0.5
22.1
0.4
0.5
0.6
No
World region of birth
Total
100.0
(X)
100.0
(X)
(X)
(X)
(X)
United States
76.4
0.5
77.0
0.4
-0.6
0.6
No
Puerto Rico/U.S. territories
1.0
0.1
0.9
0.1
0.1
No
Asia
5.6
0.2
5.7
0.2
-0.3
0.2
No
Europe
4.2
0.2
4.5
0.2
-0.1
0.4
No
Latin America
11.5
0.4
10.5
0.3
0.9
0.5
No
Other areas
1.4
0.1
1.4
0.1
0.2
No
Selected place of birth3
Total
100.0
(X)
100.0
(X)
(X)
(X)
(X)
Canada
0.5
0.6
0.1
No
China
0.8
0.1
1.0
0.1
-0.2
0.1
No
El Salvador
0.6
0.1
0.6
0.1
0.1
No
Germany
0.6
0.1
0.6
0.1
0.1
No
India
0.9
0.1
1.1
0.1
-0.2
0.2
No
Italy
0.6
0.1
0.5
0.1
0.1
No
Mexico
7.3
0.3
6.3
0.2
1.0
0.4
No
Philippines
0.8
0.1
0.9
0.1
-0.1
0.1
No
United Kingdom
0.7
0.1
0.6
0.1
0.1
0.1
No
Vietnam
0.7
0.1
0.5
0.1
0.2
0.1
No
Other places
86.6
0.4
87.4
0.3
-0.8
0.5
No
(X) Not applicable.
Dash (-) represents or rounds to zero.
1
For the two-sided hypothesis tests involving father's broad place of birth and mother's broad place of birth, statistical significance of
differences is determined at the α=0.10 significance level. For all other two-sided hypothesis tests in this table, the familywise error
rate has been controlled using the Bonferroni-Holm multiple comparison method at the α=0.10 level.
2
Includes Puerto Rico and U.S. island territories.
3
The countries included in both mother's and father's selected places of birth were based on parental place of birth data from the
2010 Current Population Survey and represent the ten largest groups. There are differences between the distributions: father's place
of birth includes Cuba and the Dominican Republic in its ten largest groups; mother's place of birth includes Germany and the United
Kingdom in its ten largest groups.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test, September to December 2010.

A-4

Table 8. Gross Difference Rates for Selected Recodes of Parental Place of Birth by Treatment

Variable/recode/category

Test
Estimate Standard
(%) error (%)

Control
Estimate Standard
(%) error (%)

Test Control Standard
1
(%) error (%) Significant

Father's place of birth
Unweighted sample size
20,784
(X)
20,857
(X)
(X)
(X)
(X)
Broad place of birth
United States2
1.1
0.1
0.9
0.1
0.2
0.2
No
Not United States
1.1
0.1
0.9
0.1
0.2
0.2
No
World region of birth
United States
1.3
0.1
1.0
0.1
0.3
0.2
No
Puerto Rico/U.S. territories
0.2
0.1
0.1
0.1
No
Asia
0.1
0.2
0.1
No
Europe
0.6
0.1
0.5
0.1
0.1
No
Latin America
0.6
0.1
0.3
0.1
0.2
0.1
No
Other areas
0.3
0.1
0.3
0.1
0.1
No
Selected place of birth
Canada
0.1
0.1
No
China
0.1
0.1
No
Cuba
0.1
0.1
No
Dominican Republic
No
El Salvador
No
India
No
Italy
0.1
0.1
No
Mexico
0.4
0.1
0.2
0.2
0.1
Yes
Philippines
No
Vietnam
No
Other places
0.8
0.1
0.5
0.1
0.3
0.1
Yes
(X) Not applicable.
Dash (-) represents or rounds to zero.
1
For the two-sided hypothesis tests involving father's broad place of birth and mother's broad place of birth, statistical
significance of differences is determined at the α=0.10 significance level. For all other two-sided hypothesis tests in this table,
the familywise error rate has been controlled using the Bonferroni-Holm multiple comparison method at the α=0.10 level.
2
Includes Puerto Rico and U.S. island territories.
3
The countries included in both mother's and father's selected places of birth were based on parental place of birth data from
the 2010 Current Population Survey and represent the ten largest groups. There are differences between the distributions:
father's place of birth includes Cuba and the Dominican Republic in its ten largest groups; mother's place of birth includes
Germany and the United Kingdom in its ten largest groups.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test, September to December 2010.

A-5

Table 8. Gross Difference Rates for Selected Recodes of Parental Place of Birth by Treatment
(continued)

Variable/recode/category

Test
Estimate Standard
(%) error (%)

Control
Estimate Standard
(%) error (%)

Test Control Standard
1
(%) error (%) Significant

Mother's place of birth
Unweighted sample size
21,303
(X)
21,369
(X)
(X)
(X)
(X)
Broad place of birth
United States2
1.4
0.1
1.4
0.1
0.1
0.2
No
Not United States
1.4
0.1
1.4
0.1
0.1
0.2
No
World region of birth
United States
1.5
0.1
1.4
0.1
0.1
0.2
No
Puerto Rico/U.S. territories
0.1
0.1
0.1
No
Asia
0.3
0.1
0.4
0.1
-0.1
0.1
No
Europe
0.6
0.1
0.7
0.1
-0.1
0.1
No
Latin America
0.8
0.1
0.6
0.1
0.2
0.1
No
Other areas
0.3
0.1
0.4
0.1
-0.1
0.1
No
Selected place of birth
Canada
0.1
0.1
No
China
0.1
0.1
0.1
No
El Salvador
No
Germany
0.1
0.1
0.1
No
India
No
Italy
0.1
0.1
No
Mexico
0.4
0.1
0.2
0.2
0.1
No
Philippines
0.1
No
United Kingdom
0.1
0.1
No
Vietnam
No
Other places
0.9
0.1
0.8
0.1
0.1
0.1
No
(X) Not applicable.
Dash (-) represents or rounds to zero.
1
For the two-sided hypothesis tests involving father's broad place of birth and mother's broad place of birth, statistical
significance of differences is determined at the α=0.10 significance level. For all other two-sided hypothesis tests in this table,
the familywise error rate has been controlled using the Bonferroni-Holm multiple comparison method at the α=0.10 level.
2
Includes Puerto Rico and U.S. island territories.
3
The countries included in both mother's and father's selected places of birth were based on parental place of birth data from
the 2010 Current Population Survey and represent the ten largest groups. There are differences between the distributions:
father's place of birth includes Cuba and the Dominican Republic in its ten largest groups; mother's place of birth includes
Germany and the United Kingdom in its ten largest groups.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test, September to December 2010.

A-6

Table 9. Indices of Inconsistency for Selected Recodes of Parental Place of Birth by
Treatment

Variable/recode/category
Father's place of birth
Unweighted sample size
Broad place of birth
2
United States
Not United States
World region of birth
United States
Puerto Rico/U.S. territories
Asia
Europe
Latin America
Other areas
3
Selected place of birth
Canada
China
Cuba
Dominican Republic
El Salvador
India
Italy
Mexico
Philippines
Vietnam
Other places

Test
Estimate Standard
(%) error (%)

Control
Estimate Standard
(%) error (%)

Test Control Standard
1
(%) error (%) Significant

20,784

(X)

20,857

(X)

(X)

(X)

(X)

3.3
3.3

0.4
0.4

2.8
2.8

0.3
0.3

0.5
0.5

0.5
0.5

No
No

3.5
10.8
1.5
6.8
2.8
9.9

0.4
2.6
0.4
1.0
0.5
2.1

2.9
5.6
2.0
5.7
1.8
11.2

0.3
2.1
0.5
0.8
0.3
2.3

0.6
5.2
-0.5
1.1
1.0
-1.3

0.5
3.4
0.6
1.3
0.6
3.0

No
No
No
No
No
No

6.9
4.6
7.6
2.8
4.3
1.3
6.3
3.0
0.9
4.0
3.4

3.4
2.1
4.8
1.3
1.7
0.9
2.2
0.5
0.7
2.5
0.5

5.1
4.5
2.9
2.1
0.4
1.4
5.8
1.7
1.3
1.1
2.3

1.7
1.9
1.7
1.1
0.3
1.0
2.0
0.4
0.8
0.9
0.3

1.9
0.1
4.6
0.7
3.9
-0.1
0.5
1.3
-0.5
2.9
1.1

3.9
2.8
5.3
1.5
1.7
1.3
2.9
0.6
1.0
2.6
0.5

No
No
No
No
No
No
No
No
No
No
No

(X) Not applicable.
Dash (-) represents or rounds to zero.
1
For the two-sided hypothesis tests involving father's broad place of birth and mother's broad place of birth, statistical
significance of differences is determined at the α=0.10 significance level. For all other two-sided hypothesis tests in this
table, the familywise error rate has been controlled using the Bonferroni-Holm multiple comparison method at the α=0.10
level.
2
Includes Puerto Rico and U.S. island territories.
3

The countries included in both mother's and father's selected places of birth were based on parental place of birth data
from the 2010 Current Population Survey and represent the ten largest groups. There are differences between the
distributions: father's place of birth includes Cuba and the Dominican Republic in its ten largest groups; mother's place of
birth includes Germany and the United Kingdom in its ten largest groups.

Note: The index of inconsistency can be used to determine the degree of inconsistency among a specific response category
between the original Content Test interview and the Content Follow-up interview. Generally, the degree of inconsistency for
a response category is low if the index of inconsistency is less than 20 percent; moderate if the index of inconsistency is at
least 20 percent but less than 50 percent; and high if the index of inconsistency is 50 percent or higher.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test, September to December 2010.

A-7

Table 9. Indices of Inconsistency for Selected Recodes of Parental Place of Birth by
Treatment (continued)

Variable/recode/category
Mother's place of birth
Unweighted sample size
Broad place of birth
2
United States
Not United States
World region of birth
United States
Puerto Rico/U.S. territories
Asia
Europe
Latin America
Other areas
3
Selected place of birth
Canada
China
El Salvador
Germany
India
Italy
Mexico
Philippines
United Kingdom
Vietnam
Other places

Test
Estimate Standard
(%) error (%)

Control
Estimate Standard
(%) error (%)

Test Control Standard
1
(%) error (%) Significant

21,303

(X)

21,369

(X)

(X)

(X)

(X)

4.1
4.1

0.3
0.3

4.1
4.1

0.4
0.4

0.1
0.1

0.5
0.5

No
No

4.3
8.3
2.6
6.8
4.2
11.0

0.3
2.5
0.6
0.9
0.5
2.2

4.2
8.1
3.9
7.3
3.5
12.8

0.4
2.3
0.8
1.1
0.4
2.3

0.3
-1.3
-0.6
0.7
-1.8

0.5
3.1
1.1
1.4
0.6
3.1

No
No
No
No
No
No

8.0
4.8
3.0
10.3
1.6
6.6
2.7
2.5
4.5
4.1
3.8

2.1
1.8
1.4
2.7
1.0
2.3
0.5
1.9
1.4
2.3
0.4

6.1
6.4
2.3
8.6
1.2
7.6
1.5
4.4
7.1
5.5
3.5

1.5
3.3
1.0
2.4
1.1
2.7
0.2
2.1
2.2
2.2
0.4

1.8
-1.6
0.7
1.6
0.3
-0.9
1.3
-1.9
-2.6
-1.4
0.3

2.6
3.8
1.5
3.5
1.5
3.5
0.6
2.9
2.4
3.1
0.6

No
No
No
No
No
No
No
No
No
No
No

(X) Not applicable.
Dash (-) represents or rounds to zero.
1
For the two-sided hypothesis tests involving father's broad place of birth and mother's broad place of birth, statistical
significance of differences is determined at the α=0.10 significance level. For all other two-sided hypothesis tests in this
table, the familywise error rate has been controlled using the Bonferroni-Holm multiple comparison method at the α=0.10
level.
2
Includes Puerto Rico and U.S. island territories.
3

The countries included in both mother's and father's selected places of birth were based on parental place of birth data
from the 2010 Current Population Survey and represent the ten largest groups. There are differences between the
distributions: father's place of birth includes Cuba and the Dominican Republic in its ten largest groups; mother's place of
birth includes Germany and the United Kingdom in its ten largest groups.

Note: The index of inconsistency can be used to determine the degree of inconsistency among a specific response category
between the original Content Test interview and the Content Follow-up interview. Generally, the degree of inconsistency for
a response category is low if the index of inconsistency is less than 20 percent; moderate if the index of inconsistency is at
least 20 percent but less than 50 percent; and high if the index of inconsistency is 50 percent or higher.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test, September to December 2010.

A-8

Table 10. L-Fold Indices of Inconsistency for Selected Recodes of Parental Place of Birth
by Treatment

Variable/recode
Father's place of birth
Broad place of birth
World region of birth
Selected place of birth
Mother's place of birth
Broad place of birth
World region of birth
Selected place of birth

Test
Estimate Standard
(%) error (%)

Control
Test Estimate Standard Control Standard
1
(%) error (%)
(%) error (%) Significant

3.3
3.8
3.4

0.4
0.4
0.4

2.8
3.3
2.3

0.3
0.3
0.3

0.5
0.6
1.2

0.5
0.5
0.5

No
No
Yes

4.1
4.6
3.7

0.3
0.3
0.4

4.1
4.8
3.5

0.4
0.4
0.4

0.1
-0.2
0.3

0.5
0.5
0.6

No
No
No

Statistical significance of differences is determined at the α=0.10 significance level.
Note: The L-fold index of inconsistency is a weighted average of the indices of inconsistency for every response category
that pertains to a variable. It can be used to determine the overall degree of inconsistency of a variable between the
original Content Test interview and the Content Follow-up interview. Generally, the overall degree of inconsistency for a
variable is low if the L-fold index of inconsistency is less than 20 percent; moderate if the L-fold index of inconsistency is
at least 20 percent but less than 50 percent; and high if the L-fold index of inconsistency is 50 percent or higher.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test, September to December 2010.
1

Table 11. Rate of Multiple Ancestries Reported by Treatment
Treatment/variable
Test
Reported multiple ancestries
Control
Reported multiple ancestries
Difference
Test – Control

Estimate
(%)

Standard
error (%)

26.8

0.4

(X)

28.2

0.5

(X)

-1.4

0.6

Yes

1

Significant

(X) Not applicable.
1
Statistical significance of differences is determined at the α=0.10 significance level.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test,
September to December 2010.

Table 12. Chi-Square Tests of Independence for Regional and
Selected Ancestry Recodes
Variable/recode
First ancestry reported
Regional ancestry groups
Selected ancestry groups
Second ancestry reported
Regional ancestry groups
Selected ancestry groups

Rao-Scott
chi-square

Degrees of
freedom

21.50
10.56

11
10

Yes
No

12.07
7.60

11
10

No
No

Statistical significance of differences is determined at the α=0.10 significance level.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test,
September to December 2010.
1

A-9

1

Significant

Table 13. Response Distribution of Region of Ancestry Recode by Treatment
Test
Estimate Standard
(%) error (%)

Variable/recode
Region of first ancestry reported
Unweighted sample size
41,005
Total
100.0
African
8.1
Asian
5.8
Eastern European
4.9
Northern European
18.1
Southern European
6.1
Western European
15.0
Caribbean
2.9
Central American
11.2
South American
0.8
North American
12.8
Oceanian
0.1
Other, non-regional
14.2
Region of second ancestry reported
Unweighted sample size
9,388
Total
100.0
African
1.3
Asian
2.0
Eastern European
8.6
Northern European
37.9
Southern European
7.2
Western European
25.5
Caribbean
1.1
Central American
2.3
South American
0.4
North American
9.1
Oceanian
0.3
Other, non-regional
4.3

Control
Test Estimate Standard Control Standard
1
(%) error (%)
(%) error (%) Significant

(X)
(X)
0.2
0.3
0.2
0.4
0.2
0.4
0.2
0.4
0.1
0.4
0.3

40,387
100.0
8.3
5.8
5.0
19.2
6.4
15.4
2.5
10.2
1.2
12.3
0.1
13.5

(X)
(X)
0.3
0.3
0.2
0.4
0.2
0.4
0.1
0.3
0.1
0.3
0.1
0.4

(X)
(X)
-0.2
0.1
-0.1
-1.1
-0.3
-0.4
0.4
0.9
-0.4
0.5
0.6

(X)
(X)
0.4
0.4
0.2
0.6
0.3
0.5
0.2
0.5
0.1
0.5
0.1
0.5

(X)
(X)
No
No
No
No
No
No
No
No
No
No
No
No

(X)
(X)
0.2
0.3
0.4
0.9
0.4
0.7
0.2
0.2
0.1
0.5
0.1
0.3

9,883
100.0
1.4
2.5
8.5
37.2
7.3
25.5
1.3
1.9
0.7
9.8
0.2
3.8

(X)
(X)
0.2
0.3
0.5
0.8
0.4
0.7
0.2
0.2
0.2
0.6
0.3

(X)
(X)
-0.5
0.1
0.7
-0.1
-0.2
0.4
-0.3
-0.8
0.1
0.6

(X)
(X)
0.3
0.4
0.6
1.0
0.6
1.0
0.3
0.3
0.2
0.8
0.1
0.4

(X)
(X)
No
No
No
No
No
No
No
No
No
No
No
No

(X) Not applicable.
Dash (-) represents or rounds to zero.
1
For these families of two-sided hypothesis tests, the familywise error rate has been controlled using the Bonferroni-Holm
multiple comparison method at the α=0.10 level.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test, September to December 2010.

A-10

Table 14. Response Distribution of Selected Ancestries by Treatment

Variable/recode
First ancestry reported
Unweighted sample size
Total
African American
American
American Indian
English
French
German
Irish
Italian
Mexican
Polish
Other ancestry groups2
Second ancestry reported
Unweighted sample size
Total
African American
American
American Indian
English
French
German
Irish
Italian
Mexican
Polish
2
Other ancestry groups

Test
Estimate Standard
(%) error (%)

Control
Estimate Standard
(%) error (%)

Test Control
(%)

Standard
1
error (%) Significant

41,005
100.0
8.4
9.5
2.5
5.9
1.6
12.1
6.9
4.5
9.9
2.4
36.3

(X)
(X)
0.2
0.3
0.2
0.2
0.1
0.3
0.3
0.2
0.4
0.2
0.5

40,387
100.0
8.6
8.8
2.7
6.2
1.5
12.5
7.4
4.8
9.1
2.3
36.2

(X)
(X)
0.3
0.3
0.2
0.2
0.1
0.3
0.2
0.2
0.3
0.2
0.5

(X)
(X)
-0.2
0.7
-0.2
-0.3
-0.4
-0.5
-0.3
0.9
0.2

(X)
(X)
0.4
0.5
0.2
0.3
0.2
0.4
0.4
0.3
0.5
0.2
0.7

(X)
(X)
No
No
No
No
No
No
No
No
No
No
No

9,388
100.0
1.6
3.1
5.1
10.5
4.4
17.4
16.5
5.2
1.7
4.7
29.9

(X)
(X)
0.2
0.3
0.5
0.5
0.4
0.5
0.7
0.4
0.2
0.4
0.9

9,883
100.0
1.7
3.3
5.7
10.6
4.1
17.5
15.8
4.9
1.2
4.0
31.3

(X)
(X)
0.2
0.4
0.5
0.6
0.3
0.6
0.7
0.4
0.2
0.3
0.8

(X)
(X)
-0.2
-0.6
-0.1
0.3
0.7
0.3
0.5
0.7
-1.4

(X)
(X)
0.3
0.5
0.6
0.7
0.6
0.8
0.9
0.6
0.3
0.5
1.2

(X)
(X)
No
No
No
No
No
No
No
No
No
No
No

(X) Not applicable.
Dash (-) represents or rounds to zero.
1
For these families of two-sided hypothesis tests, the familywise error rate has been controlled using the Bonferroni-Holm
multiple comparison method at the α=0.10 level.
2
Includes ancestry not reported.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test, September to December 2010.

A-11

Table 15. Gross Difference Rates for Region of Ancestry Recode by Treatment
Test
Estimate Standard
(%) error (%)

Variable/recode
Region of first ancestry reported
Unweighted sample size
18,890
African
2.8
Asian
1.6
Eastern European
2.7
Northern European
12.2
Southern European
2.8
Western European
10.3
Caribbean
1.0
Central American
2.1
South American
0.3
North American
12.3
Oceanian
0.1
Other, non-regional
15.7
Region of second ancestry reported
Unweighted sample size
3,318
African
0.7
Asian
1.5
Eastern European
6.8
Northern European
26.6
Southern European
6.5
Western European
24.6
Caribbean
0.8
Central American
1.1
South American
0.2
North American
6.7
Oceanian
0.4
Other, non-regional
3.6

Control
Estimate Standard
(%) error (%)

Test Control Standard
1
(%) error (%) Significant

(X)
0.3
0.2
0.2
0.4
0.2
0.4
0.1
0.2
0.1
0.5
0.5

18,524
2.4
1.4
2.7
12.1
3.3
10.5
0.9
2.6
0.5
11.3
15.6

(X)
0.2
0.2
0.2
0.5
0.2
0.4
0.1
0.3
0.1
0.5
0.6

(X)
0.4
0.2
0.1
-0.5
-0.2
0.1
-0.4
-0.1
0.9
-

(X)
0.4
0.3
0.3
0.6
0.3
0.6
0.2
0.4
0.1
0.7
0.8

(X)
No
No
No
No
No
No
No
No
No
No
No
No

(X)
0.2
0.4
0.8
1.2
0.6
1.0
0.2
0.4
0.1
0.8
0.2
0.6

3,477
0.9
1.8
7.4
24.6
6.1
23.8
0.7
1.0
0.7
7.4
0.1
3.5

(X)
0.2
0.4
0.6
1.2
0.7
1.2
0.2
0.3
0.3
0.8
0.1
0.5

(X)
-0.2
-0.3
-0.6
1.9
0.4
0.8
0.1
0.1
-0.6
-0.7
0.3
0.1

(X)
0.3
0.6
1.0
1.4
0.9
1.4
0.3
0.4
0.3
1.0
0.2
0.8

(X)
No
No
No
No
No
No
No
No
No
No
No
No

(X) Not applicable.
Dash (-) represents or rounds to zero.
1
For these families of two-sided hypothesis tests, the familywise error rate has been controlled using the Bonferroni-Holm
multiple comparison method at the α = 0.10 level.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test, September to December 2010.

A-12

Table 16. Gross Difference Rates for Selected Ancestries by Treatment

Variable/recode
First ancestry reported
Unweighted sample size
African American
American
American Indian
English
French
German
Irish
Italian
Mexican
Polish
2
Other ancestry groups
Second ancestry reported
Unweighted sample size
African American
American
American Indian
English
French
German
Irish
Italian
Mexican
Polish
2
Other ancestry groups

Test
Estimate Standard
(%) error (%)

Control
Estimate Standard
(%) error (%)

Test Control Standard
1
(%) error (%) Significant

18,890
1.9
9.7
2.7
6.1
1.7
8.4
5.7
1.8
1.9
1.4
18.8

(X)
0.2
0.5
0.2
0.3
0.2
0.4
0.3
0.2
0.2
0.1
0.5

18,524
1.7
8.8
2.6
6.8
1.4
8.7
5.4
2.1
2.2
1.4
19.4

(X)
0.2
0.4
0.2
0.5
0.1
0.4
0.3
0.2
0.3
0.2
0.6

(X)
0.2
0.9
0.1
-0.7
0.3
-0.3
0.3
-0.3
-0.3
-0.6

(X)
0.3
0.7
0.3
0.5
0.2
0.5
0.4
0.3
0.3
0.2
0.7

(X)
No
No
No
No
No
No
No
No
No
No
No

3,318
0.9
1.8
4.2
11.1
4.8
21.0
16.7
5.5
0.9
4.8
19.0

(X)
0.2
0.3
0.7
0.7
0.6
1.0
1.1
0.6
0.4
0.6
1.0

3,477
1.0
2.4
4.2
11.9
4.6
19.2
15.9
5.0
0.7
4.6
22.0

(X)
0.3
0.5
0.6
0.9
0.5
0.9
1.0
0.6
0.3
0.6
1.0

(X)
-0.1
-0.7
-0.8
0.2
1.8
0.9
0.6
0.2
0.2
-2.9

(X)
0.3
0.6
0.8
1.1
0.8
1.3
1.3
0.9
0.4
0.8
1.4

(X)
No
No
No
No
No
No
No
No
No
No
No

(X) Not applicable.
Dash (-) represents or rounds to zero.
1
For these families of two-sided hypothesis tests, the familywise error rate has been controlled using the Bonferroni-Holm
multiple comparison method at the α = 0.10 level.
2
Includes ancestry not reported.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test, September to December 2010.

A-13

Table 17. Indices of Inconsistency for Region of Ancestry Recode by Treatment
Test
Estimate Standard
(%) error (%)

Variable/recode
Region of first ancestry reported
Unweighted sample size
18,890
African
23.4
Asian
15.2
Eastern European
28.4
Northern European
36.8
Southern European
24.6
Western European
34.7
Caribbean
21.0
Central American
10.9
South American
21.3
North American
64.8
Oceanian
45.5
Other, non-regional
68.1
Region of second ancestry reported
Unweighted sample size
3,318
African
56.6
Asian
53.7
Eastern European
44.7
Northern European
54.2
Southern European
48.1
Western European
60.5
Caribbean
57.6
Central American
38.5
South American
47.1
North American
59.2
Oceanian
83.6
Other, non-regional
65.9

Control
Estimate Standard
(%) error (%)

Test Control Standard
1
(%) error (%) Significant

(X)
2.1
1.9
1.8
1.1
1.5
1.2
2.4
1.0
3.7
2.1
14.6
1.9

18,524
20.5
14.6
28.2
35.6
26.9
35.0
20.9
14.7
21.3
61.5
20.2
65.4

(X)
2.1
2.1
1.9
1.4
1.7
1.3
3.2
1.5
3.9
2.1
10.2
1.7

(X)
3.0
0.6
0.2
1.2
-2.3
-0.2
0.1
-3.9
3.3
25.2
2.7

(X)
2.8
2.7
2.7
1.6
2.3
1.7
3.9
1.9
5.7
2.8
17.8
2.4

(X)
No
No
No
No
No
No
No
No
No
No
No
No

(X)
10.9
10.3
3.7
2.5
3.8
2.1
10.9
11.8
18.7
4.2
14.4
8.1

3,477
63.7
56.2
45.4
50.4
47.1
62.2
34.8
47.6
39.5
48.3
39.2
69.8

(X)
10.0
10.3
3.5
2.3
5.2
2.5
10.8
9.4
6.7
5.4
32.6
6.8

(X)
-7.1
-2.6
-0.7
3.8
1.0
-1.7
22.8
-9.1
7.6
10.9
44.4
-3.9

(X)
14.0
15.8
4.8
2.8
6.7
3.2
14.7
13.6
20.3
6.5
34.6
11.0

(X)
No
No
No
No
No
No
No
No
No
No
No
No

(X) Not applicable.
Dash (-) represents or rounds to zero.
1
For these families of two-sided hypothesis tests, the familywise error rate has been controlled using the Bonferroni-Holm
multiple comparison method at the α = 0.10 level.
Note: The index of inconsistency can be used to determine the degree of inconsistency among a specific response
category between the original Content Test interview and the Content Follow-up interview. Generally, the degree of
inconsistency for a response category is low if the index of inconsistency is less than 20 percent; moderate if the index of
inconsistency is at least 20 percent but less than 50 percent; and high if the index of inconsistency is 50 percent or higher.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test, September to December 2010.

A-14

Table 18. Indices of Inconsistency for Selected Ancestries by Treatment

Variable/recode
First ancestry reported
Unweighted sample size
African American
American
American Indian
English
French
German
Irish
Italian
Mexican
Polish
2
Other ancestry groups
Second ancestry reported
Unweighted sample size
African American
American
American Indian
English
French
German
Irish
Italian
Mexican
Polish
2
Other ancestry groups

Test
Estimate Standard
(%) error (%)

Control
Estimate Standard
(%) error (%)

Test Control Standard
1
(%) error (%) Significant

18,890
14.9
70.1
59.0
47.4
45.1
33.8
42.8
21.5
10.5
31.6
40.7

(X)
1.7
2.5
3.0
1.8
3.8
1.4
1.8
1.6
1.0
2.1
1.0

18,524
13.9
73.2
46.0
48.3
45.7
34.1
38.5
22.4
13.9
29.2
41.7

(X)
1.5
2.5
3.6
2.6
3.2
1.5
1.8
1.7
1.5
2.2
1.2

(X)
1.0
-3.1
13.0
-0.9
-0.6
-0.3
4.4
-1.0
-3.5
2.4
-1.1

(X)
2.2
3.4
4.5
3.1
4.9
1.9
2.5
2.5
1.9
2.9
1.4

(X)
No
No
Yes
No
No
No
No
No
No
No
No

3,318
61.8
60.3
56.3
55.1
57.2
66.3
54.7
51.2
46.1
51.1
47.3

(X)
11.3
9.9
5.3
3.9
5.9
2.5
2.8
5.4
17.2
4.1
2.3

3,477
61.9
67.4
41.2
54.4
59.2
64.9
54.3
52.1
82.0
58.0
52.2

(X)
10.8
10.1
6.6
2.7
4.6
2.6
3.3
6.6
11.1
5.2
2.3

(X)
-0.1
-7.1
15.1
0.7
-2.0
1.4
0.4
-0.9
-35.9
-7.0
-4.9

(X)
14.5
13.6
8.1
4.9
7.8
3.6
4.1
8.6
18.0
6.5
3.0

(X)
No
No
No
No
No
No
No
No
No
No
No

(X) Not applicable.
1
For these families of two-sided hypothesis tests, the familywise error rate has been controlled using the BonferroniHolm multiple comparison method at the α = 0.10 level
2
Includes ancestry not reported.
Note: The index of inconsistency can be used to determine the degree of inconsistency among a specific response
category between the original Content Test interview and the Content Follow-up interview. Generally, the degree of
inconsistency for a response category is low if the index of inconsistency is less than 20 percent; moderate if the index
of inconsistency is at least 20 percent but less than 50 percent; and high if the index of inconsistency is 50 percent or
higher.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test, September to December 2010.

Table 19. L-Fold Indices of Inconsistency for Ancestry Recodes by Treatment

Variable/recode
First ancestry reported
Regional ancestry groups
Selected ancestry groups
Second ancestry reported
Regional ancestry groups
Selected ancestry groups

Test
Estimate Standard
(%) error (%)

Control
Estimate Standard
(%) error (%)

Test Control Standard
1
(%) error (%) Significant

36.7
36.8

0.9
0.8

36.5
37.1

0.9
0.8

0.2
-0.3

1.0
1.1

No
No

55.0
55.0

2.4
1.5

52.9
55.6

2.3
1.6

2.1
-0.7

1.9
1.8

No
No

Statistical significance of differences is determined at the α=0.10 significance level.
Note: The L-fold index of inconsistency is a weighted average of the indices of inconsistency for every response category
that pertains to a variable. It can be used to determine the overall degree of inconsistency of a variable between the
original Content Test interview and the Content Follow-up interview. Generally, the overall degree of inconsistency for a
variable is low if the L-fold index of inconsistency is less than 20 percent; moderate if the L-fold index of inconsistency is
at least 20 percent but less than 50 percent; and high if the L-fold index of inconsistency is 50 percent or higher.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test, September to December 2010.
1

A-15

Table 20. Chi-Square Tests of Independence for School
Enrollment Status, Language Spoken at Home, and English
Speaking Ability
Variable/recode

Rao-Scott
chi-square

Degrees of
freedom

6.27

2

Yes

10.88

5

Yes

1.62

3

No

School enrollment status
Selected language spoken at
home
English speaking ability

1

Significant

Statistical significance of differences is determined at the α=0.10 confidence level.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test,
September to December 2010.
1

Table 21. Response Distribution for School Enrollment Status, Language Spoken at Home, and
English Speaking Ability by Treatment

Variable/recode
School enrollment status
Unweighted sample size
Total
Did not attend school
Attended private school
Attended public school
Speaks language other than English
at home
Unweighted sample size
Total
Yes
No
Selected language spoken at home
Unweighted sample size
Total
Chinese
French
Spanish
Tagalog
Vietnamese
Other non-English languages
English speaking ability
Unweighted sample size
Total
Speaks English very well
Speaks English well
Does not speak English well
Does not speak English at all

Test
Estimate Standard
(%) error (%)

Control
Estimate Standard
(%) error (%)

Test Control Standard
1
(%) error (%) Significant

43,995
100.0
72.1
3.9
24.0

(X)
(X)
0.3
0.1
0.3

43,675
100.0
72.8
4.1
23.0

(X)
(X)
0.3
0.1
0.3

(X)
(X)
-0.7
-0.2
1.0

(X)
(X)
0.5
0.2
0.5

(X)
(X)
No
No
No

42,899
100.0
21.4
78.6

(X)
(X)
0.5
0.5

42,732
100.0
20.5
79.5

(X)
(X)
0.3
0.3

(X)
(X)
0.9
-0.9

(X)
(X)
0.6
0.6

(X)
(X)
No
No

10,880
100.0
2.5
1.7
61.3
2.3
2.6
29.7

(X)
(X)
0.4
0.2
1.2
0.3
0.4
1.2

10,569
100.0
3.5
2.0
59.0
2.6
1.7
31.1

(X)
(X)
0.4
0.2
1.1
0.4
0.3
1.1

(X)
(X)
-1.0
-0.4
2.3
-0.3
0.9
-1.4

(X)
(X)
0.5
0.3
1.7
0.4
0.4
1.7

(X)
(X)
No
No
No
No
No
No

11,430
100.0
58.3
20.1
15.1
6.4

(X)
(X)
0.8
0.6
0.5
0.3

10,950
100.0
59.6
19.4
14.7
6.3

(X)
(X)
0.9
0.6
0.5
0.4

(X)
(X)
-1.3
0.7
0.4
0.1

(X)
(X)
1.2
0.9
0.7
0.4

(X)
(X)
No
No
No
No

(X) Not applicable.
1
For the two-sided hypothesis tests involving the 'speaks language other than English at home' question, statistical significance of
differences is determined at the α=0.10 significance level. For the family of two-sided hypothesis tests involving school enrollment
status, the familywise error rate has been controlled using the Bonferroni multiple comparison method at the α=0.10 level. For all
other families of two-sided hypothesis tests in this table, the familywise error rate has been controlled using the Bonferroni-Holms
multiple comparison method at the α=0.10 level.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test, September to December 2010.

A-16

Table 22. Gross Difference Rates for School Enrollment Status, Language Spoken at
Home, and English Speaking Ability by Treatment

Variable/recode
School enrollment status
Unweighted sample size
Did not attend school
Attended private school
Attended public school
Speaks language other than
English at home
Unweighted sample size
Yes
No
Selected language spoken at
home
Unweighted sample size
Chinese
French
Spanish
Tagalog
Vietnamese
Other non-English languages
English speaking ability
Unweighted sample size
Speaks English very well
Speaks English well
Does not speak English well
Does not speak English at all

Test
Estimate Standard
(%) error (%)

Control
Estimate Standard
(%) error (%)

Test Control Standard
1
(%) error (%) Significant

20,875
3.2
1.8
3.6

(X)
0.2
0.2
0.2

20,660
3.1
2.0
3.2

(X)
0.2
0.1
0.2

(X)
0.1
-0.2
0.4

(X)
0.3
0.2
0.3

(X)
No
No
No

20,422
4.5
4.5

(X)
0.2
0.2

20,252
4.5
4.5

(X)
0.3
0.3

(X)
-

(X)
0.4
0.4

(X)
No
No

4,061
1.4
0.7
0.6
0.1
0.1
2.6

(X)
0.4
0.4
0.2
0.1
0.6

3,950
1.0
0.6
0.8
0.1
1.9

(X)
0.3
0.3
0.2
0.4

(X)
0.4
0.1
-0.2
0.1
0.7

(X)
0.5
0.5
0.3
0.1
0.1
0.7

(X)
No
No
No
No
No
No

4,079
19.7
22.4
11.2
5.3

(X)
1.1
1.1
0.7
0.5

3,977
16.9
20.3
12.0
4.8

(X)
1.0
0.9
0.8
0.5

(X)
2.9
2.1
-0.8
0.5

(X)
1.4
1.3
1.1
0.7

(X)
No
No
No
No

(X) Not applicable.
Dash (-) represents or rounds to zero.
1
For the two-sided hypothesis tests involving the 'speaks language other than English at home' question, statistical
significance of differences is determined at the α=0.10 significance level. For the family of two-sided hypothesis tests
involving school enrollment status, the familywise error rate has been controlled using the Bonferroni multiple comparison
method at the α=0.10 level. For all other families of two-sided hypothesis tests in this table, the familywise error rate has
been controlled using the Bonferroni-Holms multiple comparison method at the α=0.10 level.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test, September to December 2010.

A-17

Table 23. Indices of Inconsistency for School Enrollment Status, Language Spoken at Home,
and English Speaking Ability by Treatment

Variable/recode
School enrollment status
Unweighted sample size
Did not attend school
Attended private school
Attended public school
Speaks language other than
English at home
Unweighted sample size
Yes
No
Selected language spoken at
home
Unweighted sample size
Chinese
French
Spanish
Tagalog
Vietnamese
Other non-English languages
English speaking ability
Unweighted sample size
Speaks English very well
Speaks English well
Does not speak English well
Does not speak English at all

Test
Estimate Standard
(%) error (%)

Control
Estimate Standard
(%) error (%)

Test Control Standard
1
(%) error (%) Significant

20,875
8.3
24.1
10.6

(X)
0.5
1.8
0.6

20,660
8.1
26.3
9.6

(X)
0.5
1.8
0.5

(X)
0.2
-2.2
1.0

(X)
0.7
2.6
0.8

(X)
No
No
No

20,422
14.6
14.6

(X)
0.7
0.7

20,252
15.3
15.3

(X)
0.9
0.9

(X)
-0.7
0.7

(X)
1.1
1.1

(X)
No
No

4,061
22.6
23.8
1.3
2.9
3.0
6.6

(X)
6.0
12.4
0.3
1.4
1.9
1.5

3,950
17.7
15.5
1.6
1.3
0.4
4.6

(X)
5.0
5.9
0.5
0.7
0.4
1.0

(X)
4.8
8.3
-0.3
1.6
2.6
2.0

(X)
7.8
13.8
0.6
1.5
1.9
1.8

(X)
No
No
No
No
No
No

4,079
39.6
62.2
43.9
36.3

(X)
2.1
2.8
2.8
2.7

3,977
34.0
58.0
48.2
35.3

(X)
1.9
2.4
2.8
3.2

(X)
5.5
4.1
-4.3
1.0

(X)
2.8
3.6
3.9
4.4

(X)
No
No
No
No

(X) Not applicable.
1
For the two-sided hypothesis tests involving the 'speaks language other than English at home' question, statistical
significance of differences is determined at the α=0.10 significance level. For the family of two-sided hypothesis tests
involving school enrollment status, the familywise error rate has been controlled using the Bonferroni multiple comparison
method at the α=0.10 level. For all other families of two-sided hypothesis tests in this table, the familywise error rate has
been controlled using the Bonferroni-Holms multiple comparison method at the α=0.10 level.
Note: The index of inconsistency can be used to determine the degree of inconsistency among a specific response category
between the original Content Test interview and the Content Follow-up interview. Generally, the degree of inconsistency for a
response category is low if the index of inconsistency is less than 20 percent; moderate if the index of inconsistency is at least
20 percent but less than 50 percent; and high if the index of inconsistency is 50 percent or higher.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test, September to December 2010.

A-18

Table 24. L-Fold Index of Inconsistency for School Enrollment Status, Language Spoken at
Home, and English Speaking Ability by Treatment

Variable/recode
School enrollment status
Speaks language other
than English at home
Selected language
spoken at home
English speaking ability

Test
Estimate Standard
(%) error (%)
10.7
0.6

Control
Estimate Standard
(%) error (%)
10.5
0.5

Test Control Standard
1
(%) error (%) Significant
0.2
0.8
No

14.6

0.7

15.3

0.9

-0.7

1.1

No

5.5
46.5

1.2
1.9

4.2
43.9

0.9
1.9

1.3
2.6

1.4
2.4

No
No

Statistical significance of differences is determined at the α=0.10 significance level.
Note: The L-fold index of inconsistency is a weighted average of the indices of inconsistency for every response category
that pertains to a variable. It can be used to determine the overall degree of inconsistency of a variable between the
original Content Test interview and the Content Follow-up interview. Generally, the overall degree of inconsistency for a
variable is low if the L-fold index of inconsistency is less than 20 percent; moderate if the L-fold index of inconsistency is at
least 20 percent but less than 50 percent; and high if the L-fold index of inconsistency is 50 percent or higher.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test, September to December 2010.
1

Table 25. Population in Responding Households by Mode
and by Mail Response Stratum

Variable
Total
Mode
Mail response
CATI/CAPI response
Stratum
High response area
Low response area

Test
Estimate Standard
(%)
error (%)
100.0
(X)

Control
Estimate
Standard
(%)
error (%)
100.0
(X)

57.4
42.6

0.5
0.5

57.0
43.0

0.5
0.5

74.9
25.1

0.2
0.2

74.8
25.2

0.2
0.2

(X) Non-applicable.
Note: Numbers in thousands.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test,
September to December 2010.

Table 26. Item Missing Data Rates of Parental Place of Birth Variables by Mode and by Mail
Response Stratum

Variable

Test
Estimate
Standard
Number
(%) error (%)
(unweighted)

Control
Test Estimate
Standard
Standard SigniControl
Number
1
(%) error (%)
(%) error (%) ficant
(unweighted)

Father's place of birth
Mail response
29,738
8.6
0.3
29,889
8.0
0.3
0.7
CATI/CAPI response
18,655
6.1
0.4
18,640
5.6
0.4
0.5
High response area
19,662
7.2
0.3
19,870
6.4
0.3
0.8
Low response area
28,731
7.9
0.2
28,659
8.0
0.3
-0.1
Mother's place of birth
Mail response
29,738
8.0
0.3
29,889
6.9
0.2
1.1
CATI/CAPI response
18,655
5.0
0.4
18,640
5.0
0.4
0.0
High response area
19,662
6.5
0.3
19,870
5.7
0.3
0.7
Low response area
28,731
6.9
0.2
28,659
6.7
0.2
0.1
1
Statistical significance of differences is determined at the α=0.10 significance level.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test, September to December 2010.

A-19

0.4
0.6
0.5
0.3

Yes
No
Yes
No

0.4
0.6
0.4
0.3

Yes
No
No
No

Table 27. Chi-Square Tests of Independence for Parental World
Region of Birth by Mode and by Mail Response Stratum
Variable/recode
Father's world region of birth
Mail response
CATI/CAPI response
High response area
Low response area
Mother's world region of birth
Mail response
CATI/CAPI response
High response area
Low response area

Rao-Scott
chi-square

Degrees of
freedom

7.68
5.97
7.12
6.30

5
5
5
5

No
No
No
No

7.75
6.77

5
5
5
5

No
No
No
No

6.58
8.73

Statistical significance of differences is determined at the α=0.10 significance level.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test,
September to December 2010.
1

A-20

1

Significant

Table 28. Response Distribution of Parental World Region of Birth by Mode

Variable/recode
Father's world region of birth
Mail response
Unweighted sample size
Total
United States
Puerto Rico / U.S. territories
Asia
Europe
Latin America
Other areas
CATI/CAPI response
Unweighted sample size
Total
United States
Puerto Rico / U.S. territories
Asia
Europe
Latin America
Other areas
Mother's world region of birth
Mail response
Unweighted sample size
Total
United States
Puerto Rico / U.S. territories
Asia
Europe
Latin America
Other areas
CATI/CAPI response
Unweighted sample size
Total
United States
Puerto Rico / U.S. territories
Asia
Europe
Latin America
Other areas

Test
Estimate Standard
(%) error (%)

Control
Estimate Standard
(%) error (%)

Test Control Standard Signi1
(%) error (%) ficant

26,641
100.0
82.4
0.5
5.7
5.2
5.0
1.3

(X)
(X)
0.5
0.1
0.3
0.2
0.2
0.1

26,956
100.0
81.4
0.6
6.1
5.6
4.8
1.4

(X)
(X)
0.4
0.1
0.3
0.2
0.2
0.1

(X)
(X)
0.9
-0.1
-0.4
-0.4
0.2
-0.2

(X)
(X)
0.6
0.1
0.4
0.3
0.3
0.2

(X)
(X)
No
No
No
No
No
No

17,597
100.0
68.7
1.6
5.5
3.3
19.4
1.6

(X)
(X)
0.9
0.2
0.5
0.2
0.7
0.2

17,615
100.0
71.0
1.3
5.2
3.4
17.7
1.3

(X)
(X)
0.7
0.2
0.4
0.3
0.6
0.2

(X)
(X)
-2.3
0.2
0.3
-0.2
1.7
0.2

(X)
(X)
1.1
0.2
0.7
0.4
1.0
0.3

(X)
(X)
No
No
No
No
No
No

26,901
100.0
82.5
0.4
5.8
5.1
4.8
1.4

(X)
(X)
0.4
0.3
0.2
0.2
0.1

27,377
100.0
81.4
0.6
6.2
5.4
4.8
1.5

(X)
(X)
0.4
0.1
0.3
0.2
0.2
0.1

(X)
(X)
1.1
-0.2
-0.4
-0.4
-0.1

(X)
(X)
0.6
0.1
0.4
0.3
0.3
0.2

(X)
(X)
No
No
No
No
No
No

17,782
100.0
69.8
1.5
5.3
3.3
18.7
1.4

(X)
(X)
0.9
0.2
0.5
0.3
0.7
0.2

17,757
100.0
72.1
1.3
5.1
3.5
16.8
1.2

(X)
(X)
0.7
0.2
0.4
0.3
0.6
0.1

(X)
(X)
-2.3
0.3
0.2
-0.2
1.9
0.2

(X)
(X)
1.1
0.3
0.6
0.4
0.9
0.2

(X)
(X)
No
No
No
No
No
No

(X) Not applicable.
Dash (-) represents or rounds to zero.
1
For these families of two-sided hypothesis tests, the familywise error rate has been controlled using the Bonferroni-Holm
multiple comparison method at the α=0.10 level.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test, September to December 2010.

A-21

Table 29. Response Distribution of Parental World Region of Birth by Mail Response
Stratum

Variable/recode
Father's world region of birth
High response area
Unweighted sample size
Total
United States
Puerto Rico / U.S. territories
Asia
Europe
Latin America
Other areas
Low response area
Unweighted sample size
Total
United States
Puerto Rico / U.S. territories
Asia
Europe
Latin America
Other areas
Mother's world region of birth
High response area
Unweighted sample size
Total
United States
Puerto Rico / U.S. territories
Asia
Europe
Latin America
Other areas
Low response area
Unweighted sample size
Total
United States
Puerto Rico / U.S. territories
Asia
Europe
Latin America
Other areas

Test
Estimate Standard
(%) error (%)

Control
Estimate Standard
(%) error (%)

Test Control Standard Signi1
(%) error (%) ficant

18,232
100.0
81.8
0.5
5.7
4.8
5.9
1.3

(X)
(X)
0.6
0.1
0.4
0.2
0.4
0.2

18,587
100.0
82.4
0.6
5.8
5.1
4.7
1.3

(X)
(X)
0.5
0.1
0.3
0.2
0.4
0.1

(X)
(X)
-0.5
-0.1
-0.2
-0.4
1.1
-

(X)
(X)
0.8
0.2
0.5
0.3
0.6
0.2

(X)
(X)
No
No
No
No
No
No

26,006
100.0
59.6
2.3
5.4
2.9
28.2
1.6

(X)
(X)
0.6
0.2
0.3
0.1
0.6
0.1

25,984
100.0
60.0
1.8
5.2
3.0
28.2
1.7

(X)
(X)
0.6
0.1
0.3
0.1
0.6
0.2

(X)
(X)
-0.4
0.5
0.1
-0.2
-0.1

(X)
(X)
0.8
0.2
0.4
0.2
0.8
0.2

(X)
(X)
No
Yes
No
No
No
No

18,364
100.0
82.2
0.5
5.7
4.8
5.5
1.3

(X)
(X)
0.6
0.1
0.3
0.2
0.4
0.2

18,738
100.0
82.7
0.7
5.8
5.0
4.5
1.3

(X)
(X)
0.5
0.1
0.3
0.2
0.3
0.1

(X)
(X)
-0.5
-0.1
-0.2
-0.2
1.0
0.1

(X)
(X)
0.8
0.2
0.5
0.3
0.6
0.2

(X)
(X)
No
No
No
No
No
No

26,319
100.0
61.0
2.2
5.4
2.6
27.4
1.5

(X)
(X)
0.5
0.2
0.3
0.1
0.6
0.1

26,396
100.0
61.2
1.7
5.4
3.0
27.1
1.6

(X)
(X)
0.6
0.1
0.3
0.2
0.6
0.1

(X)
(X)
-0.3
0.5
-0.4
0.3
-0.1

(X)
(X)
0.8
0.2
0.4
0.2
0.7
0.2

(X)
(X)
No
Yes
No
No
No
No

(X) Not applicable.
Dash (-) represents or rounds to zero.
1
For these families of two-sided hypothesis tests, the familywise error rate has been controlled using the BonferroniHolm multiple comparison method at the α=0.10 level.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test, September to December 2010.

A-22

Table 30. Gross Difference Rates for Parental World Region of Birth by Mode

Variable/recode/category
Father's world region of birth
Mail response
Unweighted sample size
United States
Puerto Rico / U.S. territories
Asia
Europe
Latin America
Other areas
CATI/CAPI response
Unweighted sample size
United States
Puerto Rico / U.S. territories
Asia
Europe
Latin America
Other areas
Mother's world region of birth
Mail response
Unweighted sample size
United States
Puerto Rico / U.S. territories
Asia
Europe
Latin America
Other areas
CATI/CAPI response
Unweighted sample size
United States
Puerto Rico / U.S. territories
Asia
Europe
Latin America
Other areas

Test
Estimate Standard
(%) error (%)

Control
Estimate Standard
(%) error (%)

Test Control Standard
(%) error (%)

Signi1
ficant

13,673
0.9
0.1
0.5
0.2
0.2

(X)
0.1
0.1
-

13,794
0.7
0.1
0.4
0.2
0.2

(X)
0.1
0.1
0.1

(X)
0.2
0.1
-

(X)
0.1
0.1
0.1
0.1

(X)
No
No
No
No
No
No

7,111
1.7
0.3
0.2
0.6
1.0
0.4

(X)
0.3
0.1
0.1
0.2
0.2
0.1

7,063
1.3
0.2
0.2
0.7
0.5
0.5

(X)
0.2
0.1
0.1
0.2
0.1
0.1

(X)
0.4
0.2
-0.1
0.0
0.5
0.0

(X)
0.4
0.1
0.1
0.2
0.2
0.2

(X)
No
No
No
No
No
No

14,003
1.4
0.2
0.7
0.6
0.2

(X)
0.1
0.1
0.1
0.1
-

14,153
1.2
0.1
0.2
0.6
0.5
0.3

(X)
0.1
0.1
0.1
0.1

(X)
0.1
0.1
0.1
-0.1

(X)
0.2
0.1
0.1
0.1
0.1

(X)
No
No
No
No
No
No

7,300
1.6
0.2
0.3
0.5
1.0
0.4

(X)
0.2
0.1
0.1
0.1
0.2
0.1

7,216
1.7
0.2
0.5
0.8
0.7
0.4

(X)
0.2
0.1
0.1
0.2
0.1
0.1

(X)
-0.2
-0.4
0.3
0.0

(X)
0.3
0.1
0.2
0.2
0.2
0.2

(X)
No
No
No
No
No
No

Dash (-) represents or rounds to zero.
1
For these families of two-sided hypothesis tests, the familywise error rate has been controlled using the Bonferroni-Holm
multiple comparison method at the α=0.10 level.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test, September to December 2010.

A-23

Table 31. Gross Difference Rates for Parental World Region of Birth by Mail Response
Stratum

Variable/recode/category
Father's world region of birth
High response area
Unweighted sample size
United States
Puerto Rico / U.S. territories
Asia
Europe
Latin America
Other areas
Low response area
Unweighted sample size
United States
Puerto Rico / U.S. territories
Asia
Europe
Latin America
Other areas
Mother's world region of birth
High response area
Unweighted sample size
United States
Puerto Rico / U.S. territories
Asia
Europe
Latin America
Other areas
Low response area
Unweighted sample size
United States
Puerto Rico / U.S. territories
Asia
Europe
Latin America
Other areas

Test
Estimate Standard
(%) error (%)

Control
Estimate Standard
(%) error (%)

Test Control Standard
(%) error (%)

Signi1
ficant

9,431
1.2
0.1
0.1
0.6
0.5
0.3

(X)
0.2
0.1
0.1
0.1
0.1

9,633
1.0
0.1
0.2
0.6
0.2
0.2

(X)
0.1
0.1
0.1
0.1
0.1

(X)
0.3
0.3
0.1

(X)
0.2
0.1
0.1
0.2
0.1
0.1

(X)
No
No
No
No
No
No

11,353
1.4
0.4
0.2
0.5
0.8
0.3

(X)
0.2
0.1
0.1
0.1
0.2
0.1

11,224
1.1
0.1
0.3
0.4
0.7
0.6

(X)
0.1
0.1
0.1
0.1
0.1

(X)
0.3
0.2
-0.1
0.2
0.2
-0.3

(X)
0.2
0.1
0.1
0.1
0.2
0.1

(X)
No
No
No
No
No
No

9,595
1.3
0.1
0.3
0.6
0.5
0.2

(X)
0.1
0.1
0.1
0.1
0.1

9,806
1.4
0.1
0.4
0.7
0.5
0.3

(X)
0.2
0.1
0.1
0.1
0.1
0.1

(X)
-0.1
-0.1
-0.1
0.1
-0.1

(X)
0.2
0.1
0.1
0.2
0.1
0.1

(X)
No
No
No
No
No
No

11,708
2.2
0.2
0.2
0.5
1.6
0.5

(X)
0.3
0.1
0.1
0.1
0.2
0.1

11,563
1.7
0.3
0.3
0.6
1.1
0.6

(X)
0.1
0.1
0.1
0.1
0.2
0.1

(X)
0.5
-0.1
-0.1
-0.1
0.5
-0.1

(X)
0.3
0.1
0.1
0.1
0.3
0.2

(X)
No
No
No
No
No
No

Dash (-) represents or rounds to zero.
1
For these families of two-sided hypothesis tests, the familywise error rate has been controlled using the Bonferroni-Holm
multiple comparison method at the α=0.10 level.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test, September to December 2010.

A-24

Table 32. Indices of Inconsistency for Parental World Region of Birth by Mode

Variable/recode/category
Father's world region of birth
Mail response
Unweighted sample size
United States
Puerto Rico / U.S. territories
Asia
Europe
Latin America
Other areas
CATI/CAPI response
Unweighted sample size
United States
Puerto Rico / U.S. territories
Asia
Europe
Latin America
Other areas
Mother's world region of birth
Mail response
Unweighted sample size
United States
Puerto Rico / U.S. territories
Asia
Europe
Latin America
Other areas
CATI/CAPI response
Unweighted sample size
United States
Puerto Rico / U.S. territories
Asia
Europe
Latin America
Other places

Test
Estimate Standard
(%) error (%)

Control
Estimate Standard
(%) error (%)

Test Control Standard
(%) error (%)

Signi1
ficant

13,673
3.3
4.9
1.4
5.2
1.9
7.5

(X)
0.3
2.5
0.4
0.7
0.4
1.8

13,794
2.4
3.3
1.3
3.9
1.8
6.9

(X)
0.3
2.1
0.3
0.7
0.4
2.2

(X)
0.9
1.5
0.2
1.3
0.6

(X)
0.4
3.2
0.5
0.8
0.6
2.8

(X)
No
No
No
No
No
No

7,111
3.9
12.5
1.5
9.6
3.3
11.5

(X)
0.6
3.3
0.6
2.5
0.6
3.4

7,063
3.4
6.7
3.1
8.7
1.9
16.2

(X)
0.6
2.9
1.2
2.1
0.4
4.6

(X)
0.5
5.8
-1.6
0.9
1.4
-4.7

(X)
0.9
4.4
1.4
3.2
0.8
5.5

(X)
No
No
No
No
No
No

14,003
5.0
7.0
2.1
6.7
7.3
8.5

(X)
0.4
3.4
0.5
0.8
0.9
1.6

14,153
4.2
5.8
2.0
5.2
5.8
9.5

(X)
0.5
2.7
0.5
0.7
0.9
2.1

(X)
0.8
1.2
0.1
1.5
1.5
-1.0

(X)
0.6
4.1
0.7
1.1
1.1
2.5

(X)
No
No
No
No
No
No

7,300
3.9
8.8
3.1
6.9
3.5
13.7

(X)
0.5
3.3
1.1
2.1
0.6
4.0

7,216
4.4
9.2
7.2
10.6
2.8
18.0

(X)
0.6
3.4
2.2
2.7
0.5
4.6

(X)
-0.4
-0.4
-4.1
-3.7
0.7
-4.3

(X)
0.8
4.4
2.5
3.3
0.8
6.1

(X)
No
No
No
No
No
No

Dash (-) represents or rounds to zero.
1
For these families of two-sided hypothesis tests, the familywise error rate has been controlled using the Bonferroni-Holm
multiple comparison method at the α=0.10 level.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test, September to December 2010.

A-25

Table 33. Indices of Inconsistency for Parental World Region of Birth by Mail Response
Stratum

Variable/recode/category
Father's world region of birth
High response area
Unweighted sample size
United States
Puerto Rico / U.S. territories
Asia
Europe
Latin America
Other areas
Low response area
Unweighted sample size
United States
Puerto Rico / U.S. territories
Asia
Europe
Latin America
Other areas
Mother's world region of birth
High response area
Unweighted sample size
United States
Puerto Rico / U.S. territories
Asia
Europe
Latin America
Other areas
Low response area
Unweighted sample size
United States
Puerto Rico / U.S. territories
Asia
Europe
Latin America
Other areas

Test
Estimate Standard
(%) error (%)

Control
Estimate Standard
(%) error (%)

Test Control Standard
(%) error (%)

Signi1
ficant

9,431
4.1
13.5
1.2
6.4
4.3
10.0

(X)
0.6
6.6
0.5
1.2
1.1
2.6

9,633
3.5
6.4
1.5
5.7
2.6
8.6

(X)
0.5
3.2
0.5
0.9
0.8
2.7

(X)
0.7
7.0
-0.3
0.7
1.7
1.4

(X)
0.8
7.5
0.7
1.5
1.3
3.8

(X)
No
No
No
No
No
No

11,353
2.9
9.3
2.3
9.2
2.1
9.4

(X)
0.4
2.1
0.7
1.6
0.5
2.6

11,224
2.3
4.5
3.8
6.2
1.7
17.3

(X)
0.3
1.6
1.2
1.2
0.3
4.5

(X)
0.6
4.8
-1.6
3.0
0.4
-7.9

(X)
0.5
2.8
1.5
2.0
0.5
5.4

(X)
No
No
No
No
No
No

9,595
4.3
11.6
2.7
6.2
5.3
8.3

(X)
0.5
6.2
0.8
1.1
1.0
2.2

9,806
4.9
7.3
4.0
6.8
5.9
11.2

(X)
0.6
3.7
1.0
1.3
1.2
2.7

(X)
-0.6
4.3
-1.3
-0.6
-0.6
-3.0

(X)
0.8
6.8
1.3
1.7
1.6
3.3

(X)
No
No
No
No
No
No

11,708
4.7
6.3
2.3
9.7
4.0
19.8

(X)
0.6
1.8
0.6
1.7
0.6
5.0

11,563
3.7
9.3
3.7
10.1
2.8
16.5

(X)
0.3
1.9
1.0
1.6
0.4
3.9

(X)
1.0
-2.9
-1.5
-0.4
1.3
3.3

(X)
0.7
2.7
1.3
2.3
0.7
6.5

(X)
No
No
No
No
No
No

1

For these families of two-sided hypothesis tests, the familywise error rate has been controlled using the Bonferroni-Holm
multiple comparison method at the α=0.10 level.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test, September to December 2010.

A-26

Table 34. L-Fold Indices of Inconsistency for Parental World Region of Birth by Mode
and by Mail Response Stratum

Variable/recode
Father's world region of birth
Mail response
CATI/CAPI response
High response area
Low response area
Mother's world region of birth
Mail response
CATI/CAPI response
High response area
Low response area

Test
Estimate Standard
(%) error (%)

Control
Estimate Standard
(%) error (%)

Test Control Standard
(%) error (%)

Signi1
ficant

3.3
4.4
4.4
3.3

0.3
0.6
0.6
0.4

2.6
3.9
3.7
2.9

0.3
0.6
0.5
0.4

0.7
0.5
0.7
0.4

0.4
0.8
0.8
0.6

No
No
No
No

5.3
4.3
4.7
4.9

0.4
0.5
0.5
0.5

4.5
5.2
5.6
4.2

0.5
0.7
0.6
0.4

0.8
-0.9
-0.8
0.6

0.6
0.8
0.8
0.6

No
No
No
No

Statistical significance of differences is determined at the α=0.10 level.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test, September to December 2010.
1

Table 35. Allocation Rates for Selected Variables from the 2009 American
Community Survey Compared with Item Missing Data Rates from the 2010 ACS
Content Test

Variable
School enrollment status
Educational attainment
First field of degree
Ancestry
First ancestry reported
Percent not reported, first ancestry reported
Speaks language other than English at home
Language spoken at home
English speaking ability
Residence one year ago
Health insurance:
Insurance through employer/union
Insurance purchased directly
Medicare
Medicaid
TRICARE
VA
Indian Health Service
Disability:
Difficulty hearing
Difficulty seeing

2009 ACS
allocation rate
(including
assigned
and imputed)
2.9
4.4
9.0
0.4
10.5
2.7
4.7
3.1
3.6

2010 ACS
Content Test
item missing
data rate
Test Control
4.8
5.4
6.8
7.0
2.9
2.5
13.3
14.5
4.8
5.1
6.9
4.9
2.1
1.5
5.4
5.4

9.6
23.8
21.3
25.9
24.9
25.1
25.6

12.4
24.4
21.9
25.6
26.3
26.6
27.1

12.4
25.2
22.6
26.2
27.0
27.3
27.7

2.3
2.5

4.4
4.6

4.7
4.9

Source: U.S. Census Bureau, 2009 American Community Survey and the 2010 American Community
Survey Content Test, September to December 2010.

A-27

Table 36. Item Missing Data Rates of Proximal Variables by Mode

Variable
First ancestry reported
Mail response
CATI/CAPI response
School enrollment
Mail response
CATI/CAPI response
2
Educational attainment
Mail response
CATI/CAPI response
Speaks language other than English at home
Mail response
CATI/CAPI response
Language spoken at home
Mail response
CATI/CAPI response
English speaking ability
Mail response
CATI/CAPI response

Test
Estimate
Standard
Number
(%) error (%)
(unweighted)

Control
Estimate
Standard
Number
(%) error (%)
(unweighted)

Test Control Standard
(%) error (%)

Signi1
ficant

29,889
18,640

19.0
6.8

0.4
0.4

29,738
18,655

21.1
7.1

0.4
0.4

-2.0
-0.3

0.6
0.6

Yes
No

28,958
17,854

6.2
3.1

0.2
0.4

29,061
17,751

7.2
3.4

0.2
0.3

-1.0
-0.3

0.3
0.5

Yes
No

28,958
17,854

7.9
5.4

0.2
0.4

29,061
17,751

8.2
5.5

0.2
0.3

-0.3
-0.1

0.3
0.6

No
No

28,373
17,854

6.3
3.1

0.2
0.4

29,061
17,751

6.6
3.2

0.2
0.3

-0.3
-0.2

0.3
0.4

No
No

5,580
6,068

17.1
0.1

1.0
0.1

5,262
5,839

12.1
0.1

0.6
0.0

5.0
0.0

1.2
0.1

Yes
No

5,580
6,068

4.9
0.3

0.4
0.2

5,262
5,839

3.3
0.2

0.3
0.1

1.6
0.1

0.5
0.2

Yes
No

Statistical significance of differences is determined at the α=0.10 significance level.
Though educational attainment is not part of the official analysis plan, it is included here by request of critical reviewers.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test, September to December 2010.
1
2

A-28

Table 37. Region of Ancestry Recode by Mode and Data Source

Total population
Region of first ancestry reported
African
Asian
Eastern European
Northern European
Southern European
Western European
Caribbean
Central American
South American
North American
Oceanian
Other, non-regional

ACS
2009
100.0
10.9
6.0
5.1
21.5
6.7
16.3
3.1
11.4
1.0
10.4
0.3
7.3

Region of second ancestry reported
African
Asian
Eastern European
Northern European
Southern European
Western European
Caribbean
Central American
South American
North American
Oceanian
Other, non-regional

100.0
1.0
2.1
8.6
41.2
7.1
28.7
1.2
1.6
0.4
7.2
0.2
0.7

All modes
ACS
2010
Content
Test
(test)
100.0
8.1
5.8
4.9
18.1
6.1
15.0
2.9
11.2
0.8
12.8
0.1
14.2
100.0
1.3
2.0
8.6
37.9
7.2
25.5
1.1
2.3
0.4
9.1
0.3
4.3

ACS
2010
Content
Test
(control)
100.0
8.3
5.8
5.0
19.2
6.4
15.4
2.5
10.2
1.2
12.3
0.1
13.5

ACS
2009
100.0
6.5
6.3
6.5
26.2
7.7
19.4
1.7
4.7
0.7
11.4
0.2
8.7

100.0
1.4
2.5
8.5
37.2
7.3
25.5
1.3
1.9
0.7
9.8
0.2
3.8

100.0
0.5
1.8
9.6
42.9
6.6
30.3
0.7
0.9
0.3
5.6
0.2
0.6

Mail mode
ACS
2010
Content
Test
(test)
100.0
5.3
6.3
6.3
21.3
7.2
16.9
1.2
4.4
0.6
15.9
0.1
14.4
100.0
1.1
2.1
9.0
39.2
6.1
26.8
0.5
1.7
0.6
7.9
0.3
4.7

ACS
2010
Content
Test
(control)
100.0
5.1
6.6
6.1
22.7
7.5
17.7
1.3
4.0
0.7
15.5
0.1
12.6
100.0
1.2
2.2
10.1
39.1
6.2
27.4
0.7
1.0
0.4
7.7
0.1
3.9

CATI/CAPI modes
ACS
ACS
2010
2010
Content
Content
ACS
Test
Test
2009
(test)
(control)
100.0
100.0
100.0
15.2
10.9
11.4
5.8
5.4
5.0
3.7
3.4
3.9
16.4
14.9
15.8
5.7
5.0
5.3
12.9
13.2
13.3
4.6
4.6
3.7
18.9
17.7
16.1
1.5
1.0
1.7
9.4
9.7
9.2
0.4
0.2
0.2
5.7
13.9
14.4
100.0
1.8
2.5
7.0
38.6
8.0
26.2
1.9
2.6
0.6
9.8
0.3
0.9

Note: First and second ancestry reported for 2009 ACS data do not include persons for whom an ancestry group was not reported.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test, September to December 2010; and 2009 American Community Survey.

A-29

100.0
1.7
1.9
8.0
36.3
8.4
24.0
1.9
3.0
0.1
10.5
0.3
3.9

100.0
1.6
2.9
6.6
34.9
8.6
23.3
2.1
2.9
1.0
12.3
0.3
3.6

Table 38. Selected Ancestries by Mode and Data Source

Total population
First ancestry reported
African American
American
American Indian
English
French
German
Irish
Italian
Mexican
Polish
Other ancestry groups

ACS 2009
100.0
10.9
6.8
2.7
7.1
2.0
12.8
8.3
5.0
10.0
2.4
32.0

Second ancestry reported
African American
American
American Indian
English
French
German
Irish
Italian
Mexican
Polish
Other ancestry groups

100.0
0.8
0.0
6.2
11.7
5.0
19.6
17.6
5.3
1.0
4.2
28.4

All modes
ACS 2010
ACS 2010
Content Test Content Test
(test)
(control)
100.0
100.0
8.4
8.6
9.5
8.8
2.5
2.7
5.9
6.2
1.6
1.5
12.1
12.5
6.9
7.4
4.5
4.8
9.9
9.1
2.4
2.3
36.3
36.2
100.0
1.6
3.1
5.1
10.5
4.4
17.4
16.5
5.2
1.7
4.7
29.9

100.0
1.7
3.3
5.7
10.6
4.1
17.5
15.8
4.9
1.2
4.0
31.3

ACS 2009
100.0
6.7
8.2
2.0
9.1
2.2
15.2
9.3
6.1
4.3
3.2
33.9
100.0
0.4
0.0
4.5
12.6
5.1
20.8
17.2
5.0
0.7
4.7
29.1

Mail mode
ACS 2010
ACS 2010
Content Test Content Test
(test)
(control)
100.0
100.0
5.9
5.4
13.3
12.5
1.4
1.8
7.4
7.8
1.4
1.6
13.8
14.3
7.6
8.2
5.8
6.0
4.1
3.7
3.2
3.2
36.2
35.5
100.0
1.2
2.7
3.9
11.4
4.3
18.5
15.9
4.4
1.4
4.7
31.6

100.0
1.3
2.6
4.3
10.9
4.4
18.8
16.0
4.1
0.7
4.5
32.5

CATI/CAPI modes
ACS 2010
ACS 2010
Content Test Content Test
ACS 2009
(test)
(control)
100.0
100.0
100.0
15.1
10.9
11.6
5.3
5.7
5.3
3.5
3.5
3.5
4.9
4.5
4.6
1.8
1.7
1.5
10.1
10.5
10.7
7.3
6.3
6.6
3.9
3.3
3.7
16.5
15.7
14.3
1.7
1.6
1.5
30.1
36.5
36.8
100.0
1.5
0.0
9.0
10.5
4.8
17.7
18.3
5.8
1.6
3.6
27.2

Note: First and second ancestry for 2009 ACS data do not include persons for whom an ancestry group was not reported.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test, September to December 2010; and 2009 American Community Survey.

A-30

100.0
2.2
3.5
6.5
9.4
4.4
16.3
17.3
6.2
2.0
4.6
27.7

100.0
2.1
4.1
7.3
10.2
3.8
16.0
15.7
5.9
1.7
3.3
29.8

Table 39. Education and Language Spoken at Home Variables by Mode and Data Source

Selected populations
Population age 3 and older
School enrollment status
Did not attend school
Attended private school
Attended public school
Educational attainment
Less than high school diploma
High school diploma or equivalent
Some college
Bachelor's degree
Graduate degree

All modes
ACS 2010
Content
Test
ACS 2009
(test)

ACS 2010
Content
Test
(control)

Mail mode
ACS 2010
Content
Test
ACS 2009
(test)

ACS 2010
Content
Test
(control)

CATI/CAPI modes
ACS 2010 ACS 2010
Content
Content
Test
Test
ACS 2009
(test)
(control)

100.0
72.7
4.7
22.7

100.0
72.1
3.9
24.0

100.0
72.8
4.1
23.0

100.0
75.8
4.8
19.4

100.0
76.8
3.9
19.2

100.0
77.1
4.3
18.7

100.0
69.6
3.9
26.5

100.0
66.8
3.8
29.3

100.0
68.1
4.0
27.9

100.0
32.3
22.9
24.6
13.0
7.2

100.0
31.0
21.4
25.1
14.6
8.0

100.0
30.4
21.8
25.5
14.2
8.1

100.0
26.1
22.4
25.2
16.3
10.1

100.0
23.5
21.9
25.8
17.5
11.3

100.0
23.2
22.0
26.3
17.1
11.4

100.0
39.8
23.0
23.1
9.9
4.2

100.0
39.4
21.0
24.3
11.3
4.3

100.0
38.5
21.6
24.6
11.0
4.5

100.0
20.5
79.5

100.0
13.8
86.2

100.0
16.0
84.0

100.0
15.5
84.5

100.0
27.3
72.7

100.0
27.4
72.6

100.0
26.1
73.9

100.0
3.5
2.0
59.0
2.6
1.7
31.1

100.0
5.1
3.8
46.0
4.2
3.0
37.9

100.0
5.3
2.4
39.6
3.8
5.1
43.9

100.0
6.4
3.7
39.2
4.6
2.9
43.2

100.0
1.5
1.3
71.8
1.7
1.7
22.0

100.0
1.0
1.3
73.1
1.6
1.2
21.9

100.0
1.8
1.1
70.7
1.5
0.9
24.0

100.0
53.4
19.9
17.9
8.9

100.0
55.6
18.2
17.4
8.8

Population age 5 and over
Speaks language other than English
at home
100.0
100.0
Yes
19.9
21.4
No
80.1
78.6
d
Population age 5 and over that speaks language other than English at home
Selected language
spoken at home
100.0
100.0
Chinese
2.9
2.5
French
2.2
1.7
Spanish
62.2
61.3
Tagalog
2.6
2.3
Vietnamese
2.2
2.6
Other non-English languages
27.9
29.7

English speaking ability
100.0
100.0
100.0
100.0
100.0
100.0
100.0
Speaks English very well
57.0
58.3
59.6
63.4
66.3
65.8
52.9
Speaks English well
20.0
20.1
19.4
21.7
20.5
21.2
18.5
Does not speak English well
15.9
15.1
14.7
11.9
10.7
10.5
18.5
Does not speak English at all
7.5
6.4
6.3
3.1
2.4
2.5
10.1
Source: U.S. Census Bureau, 2010 American Community Survey Content Test, September to December 2010; and 2009 American Community Survey.

A-31

Table 40. Response Distributions of Regional Ancestry Recode by Mode

Variable and mode type
Region of first reported ancestry
Mail responses
African
Asian
Eastern European
Northern European
Southern European
Western European
Caribbean
Central American
South American
North American
Oceanian
Other, non-regional
CATI/CAPI responses
African
Asian
Eastern European
Northern European
Southern European
Western European
Caribbean
Central American
South American
North American
Oceanian
Other, non-regional

Test
Estimate Standard
(%) error (%)

Control
Estimate Standard
(%) error (%)

Test Control

Standard
error (%)

Significant

100.0
5.3
6.3
6.3
21.3
7.2
16.9
1.2
4.4
0.6
15.9
0.1
14.4

(X)
0.2
0.3
0.3
0.5
0.3
0.4
0.1
0.2
0.1
0.4
0.0
0.4

100.0
5.1
6.6
6.1
22.7
7.5
17.7
1.3
4.0
0.7
15.5
0.1
12.6

(X)
0.2
0.3
0.3
0.5
0.3
0.4
0.1
0.2
0.1
0.4
0.0
0.3

(X)
0.2
-0.2
0.2
-1.4
-0.3
-0.8
-0.1
0.4
-0.1
0.4
0.0
1.8

(X)
0.3
0.4
0.3
0.7
0.4
0.6
0.1
0.3
0.1
0.7
0.0
0.5

(X)
No
No
No
No
No
No
No
No
No
No
No
Yes

100.0
10.8
5.4
3.4
14.9
5.0
13.2
4.6
17.7
1.0
9.7
0.2
13.9

(X)
0.4
0.5
0.3
0.6
0.4
0.6
0.3
0.7
0.1
0.6
0.1
0.6

100.0
11.4
5.0
3.9
15.8
5.3
13.3
3.7
16.1
1.7
9.2
0.2
14.4

(X)
0.6
0.4
0.3
0.7
0.4
0.5
0.2
0.6
0.3
0.5
0.1
0.7

(X)
-0.6
0.4
-0.5
-0.9
-0.3
-0.1
0.9
1.6
-0.6
0.5
0.0
-0.5

(X)
0.7
0.7
0.4
0.9
0.5
0.8
0.4
0.9
0.3
0.8
0.1
1.0

(X)
No
No
No
No
No
No
No
No
No
No
No
No

Statistical significance of differences is determined at the α=0.10 significance level.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test, September to December 2010.
1

A-32

Table 40. Response Distributions of Regional Ancestry Recode by Mode (continued)

Variable and mode type
Region of second reported ancestry
Mail responses
African
Asian
Eastern European
Northern European
Southern European
Western European
Caribbean
Central American
South American
North American
Oceanian
Other, non-regional
CATI/CAPI responses
African
Asian
Eastern European
Northern European
Southern European
Western European
Caribbean
Central American
South American
North American
Oceanian
Other, non-regional

Test
Estimate Standard
(%) error (%)

Control
Estimate Standard
(%) error (%)

Test - Standard
Control error (%)

Significant

100.0
1.1
2.1
9.0
39.2
6.1
26.8
0.5
1.7
0.6
7.8
0.3
4.7

(X)
0.2
0.3
0.5
0.9
0.4
0.8
0.1
0.3
0.1
0.5
0.1
0.4

100.0
1.2
2.2
10.1
39.1
6.2
27.4
0.7
1.0
0.4
7.7
0.1
3.9

(X)
0.2
0.3
0.6
0.7
0.4
0.7
0.1
0.2
0.1
0.5
0.0
0.3

(X)
-0.1
-0.1
-1.1
0.1
-0.1
-0.6
-0.2
0.7
0.2
0.2
0.2
0.8

(X)
0.2
0.4
0.7
1.1
0.5
1.1
0.1
0.3
0.2
0.8
0.1
0.5

(X)
No
No
No
No
No
No
No
No
No
No
No
No

100.0
1.7
1.9
8.0
36.3
8.4
24.0
1.9
3.0
0.1
10.5
0.3
3.9

(X)
0.4
0.4
0.8
1.6
0.8
1.2
0.5
0.4
0.0
0.9
0.1
0.6

100.0
1.6
2.9
6.6
34.9
8.6
23.3
2.1
2.9
1.0
12.3
0.2
3.6

(X)
0.4
0.5
0.8
1.5
0.9
1.2
0.5
0.4
0.4
1.1
0.1
0.5

(X)
0.1
-1.0
1.5
1.4
-0.1
0.6
-0.2
0.1
-0.9
-1.8
0.0
0.3

(X)
0.5
0.7
1.2
2.1
1.1
1.7
0.7
0.6
0.4
1.4
0.2
0.7

(X)
No
No
No
No
No
No
No
No
No
No
No
No

Statistical significance of differences is determined at the α=0.10 significance level.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test, September to December 2010.
1

A-33

Table 41. Response Distributions of School Enrollment and Language Variables by Mode

Variable and mode type
School enrollment status
Mail responses
Did not attend school
Attended public school
Attended private school

Test
Estimate Standard
(%) error (%)

Control
Estimate Standard
(%) error (%)

Test - Standard
Control error (%)

Significant

100.0
76.8
19.2
3.9

(X)
0.3
0.3
0.2

100.0
77.0
18.7
4.3

(X)
0.3
0.3
0.2

(X)
-0.2
0.5
-0.3

(X)
0.4
0.4
0.2

(X)
No
No
No

100.0
66.8
29.3
3.8

(X)
0.6
0.6
0.2

100.0
68.1
27.9
4.0

(X)
0.5
0.5
0.2

(X)
-1.3
1.5
-0.2

(X)
0.8
0.8
0.3

(X)
No
No
No

Non-English language spoken at home status
Mail responses
100.0
Yes
16.0
No
84.0

(X)
0.4
0.4

100.0
15.5
84.5

(X)
0.4
0.4

(X)
0.5
-0.5

(X)
0.5
0.5

(X)
No
No

100.0
27.4
72.6

(X)
0.8
0.8

100.0
26.1
73.9

(X)
0.6
0.6

(X)
1.3
-1.3

(X)
1.1
1.1

(X)
No
No

Specific language spoken at home
Mail responses
Spanish
Chinese
Tagalog
French
Vietnamese
Other non-English language

100.0
39.6
5.3
3.8
2.4
5.1
43.9

(X)
1.5
0.7
0.5
0.3
0.7
1.6

100.0
39.2
6.4
4.6
3.7
2.9
43.2

(X)
1.4
0.8
0.7
0.5
0.5
1.5

(X)
0.4
-1.2
-0.8
-1.3
2.2
0.7

(X)
2.2
1.0
0.8
0.6
0.8
2.3

(X)
No
No
No
No
No
No

CATI/CAPI responses
Spanish
Chinese
Tagalog
French
Vietnamese
Other non-English language

100.0
73.1
1.0
1.5
1.3
1.2
21.9

(X)
1.5
0.4
0.4
0.3
0.3
1.5

100.0
70.7
1.8
1.5
1.1
0.9
24.0

(X)
1.3
0.5
0.4
0.2
0.3
1.4

(X)
2.3
-0.8
0.1
0.2
0.3
-2.0

(X)
2.0
0.6
0.6
0.4
0.5
2.1

(X)
No
No
No
No
No
No

English-speaking ability
Mail responses
Speaks English very well
Speaks English well
Does not speak English well
Does not speak English at all

100.0
66.3
20.5
10.7
2.4

(X)
1.1
0.7
0.6
0.3

100.0
65.7
21.2
10.5
2.5

(X)
0.8
0.7
0.6
0.3

(X)
0.6
-0.7
0.2
-0.1

(X)
1.2
0.8
0.8
0.4

(X)
No
No
No
No

CATI/CAPI responses
Speaks English very well
Speaks English well
Does not speak English well
Does not speak English at all

100.0
53.3
19.9
17.9
8.9

(X)
1.2
1.0
0.8
0.5

100.0
55.6
18.2
17.4
8.8

(X)
1.3
0.8
0.8
0.7

(X)
-2.3
1.7
0.4
0.1

(X)
1.8
1.5
1.1
0.7

(X)
No
No
No
No

CATI/CAPI responses
Did not attend school
Attended public school
Attended private school

CATI/CAPI responses
Yes
No

Statistical significance of differences is determined at the α=0.10 significance level.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test, September to December 2010.
1

A-34

Table 42. Chi-Square Tests of Independence for Regional Ancestry, School
Enrollment, and Language Variables
Variable/recode
Region of first ancestry reported
Mail response
CATI/CAPI response
Region of second ancestry reported
Mail response
CATI/CAPI response
School enrollment status
Mail response
CATI/CAPI response
Selected language spoken at home
Mail response
CATI/CAPI response
English speaking ability
Mail response
CATI/CAPI response

Rao-Scott
chi-square

Degrees of
freedom

21.29
14.48

11
11

Yes
No

18.08
16.82

11
11

Yes
No

3.07
4.03

2
2

No
No

0.02
0.59

5
5

Yes
No

0.84
2.71

3
3

No
No

1

Significant

Statistical significance of differences is determined at the α=0.10 significance level.
Source: U.S. Census Bureau, 2010 American Community Survey Content Test, September to December 2010.
1

A-35

Appendix B: Images of the Mail Versions of the Control and Test
Questions

Figure B-1. Control Version of the Questionnaire Page Featuring Parental Place of Birth

B-1

Figure B-2. Test Version of the Questionnaire Page Featuring Parental Place of Birth

B-2

Appendix C: CATI and CAPI Versions of the Control and Test
Questions
Note that the CATI/CAPI question text is programmed to change based upon the subject
of the question. When the respondent is asked about her own father’s place of birth, for
instance, the interviewer would read the question as: “In what country was your
FATHER born?” However, when the respondent is asked the same question in regards to
her husband, named John, the interviewer would read the question as: “In what country
was John’s FATHER born?”

Father’s Place of Birth

In what country was [your/’s] FATHER born? Tell me
the name of the country, or Puerto Rico, Guam, etc.

Mother’s Place of Birth

In what country was [your/’s] MOTHER born? Tell me
the name of the country, or Puerto Rico, Guam, etc.

C-1

Appendix D: Spanish Language Versions of the Control and Test
Questions (CATI/CAPI Modes)
The mail version of the Content Test questionnaire was administered only in the English
language; however, the CATI/CAPI versions featured a Spanish-language alternative, the
text of which appears below. Note that, on the Spanish-language version of the
questionnaire, the PPOB questions are located in the same position and order as on the
English-language version. Also, as with the English-language version, the CATI/CAPI
question text is programmed to change based upon the subject of the question. When the
respondent is asked about his own mother’s place of birth, for instance, the interviewer
would read the question as: “En qué país nació su MADRE?” However, when the
respondent is asked the same question in regards to his wife, named Rosa, the interviewer
would read the question as: “En qué país nació el MADRE de Rosa?”

Father’s Place of Birth

En qué país nació [su PADRE / el PADRE de ]? Dígame el
nombre del país o Puerto Rico, Guam, etc.

Mother’s Place of Birth

En qué país nació [su MADRE / el MADRE de ]? Dígame el
nombre del país o Puerto Rico, Guam, etc.

D-1

Appendix E: Process Flowchart of the Content Follow-Up Survey Reinterview

E-1

Appendix F: Parental Place of Birth Crosswalk for World Region of
Birth Recode in the American Community Survey
The coding scheme for father’s and mother’s place of birth (FPOB and MPOB) in the
Content Test was identical to that used for place of birth (POB) in the American
Community Survey. The parental place of birth variables were recoded into father’s and
mother’s world region of birth (FWROB and MWROB) with the following crosswalk:

Response for FPOB / MPOB
Any of the 50 US states or the District of Columbia (001 – 056); United States
unspecified (057)
American Samoa (060); Guam (066); Johnston Atoll (067); Northern Marianas
(069); Midway Islands (071); Puerto Rico (072); Navassa Island (076); US Virgin
Islands (078); Wake Island (079); Baker Island (081); Howland Island (084); Jarvis
Island (086); Kingman Reef (089); Palmyra Atoll (095); US outlying area (096)
Albania (100); Andorra (101); Austria (102); Belgium (103); Bulgaria (104);
Czechoslovakia (105); Denmark (106); Faroe Islands (107); Finland (108); France
(109); Germany (110); Gibralter (115); Greece (116); Hungary (117); Iceland
(118); Ireland (119); Italy (120); Jan Meyan (121); Liechtenstein (122);
Luxembourg (123); Malta (124); Monaco (125); Netherlands (126); Norway (127);
Poland (128); Portugal (129); Azores Islands (130); Madeira Islands (131);
Romania (132); San Marino (133); Spain (134); Svalbard (135); Sweden (136);
Switzerland (137); United Kingdom (138); England (139); Scotland (140); Wales
(141); Northern Ireland (142); Guernsey (143); Jersey (144); Isle of Man (145);
Vatican City (146); Yugoslavia (147); Czech Republic (148); Slovakia (149);
Bosnia and Herzegovina (150); Croatia (151); Macedonia (152); Slovenia (153);
Serbia (154); Estonia (155); Latvia (156); Lithuania (157); Belarus (160); Moldova
(162); Russia (163); Ukraine (164); USSR (165); Europe (166); Kosovo (167);
Montenegro (168)
Armenia (158); Azerbaijan (159); Georgia (161); Afghanistan (200); Bahrain (201);
Bangladesh (202); Bhutan (203); Brunei (204); Myanmar (205); Cambodia (206);
China (207); Cyprus (208); Hong Kong (209); India (210); Indonesia (211); Iran
(212); Iraq (213); Israel (214); Japan (215); Jordan (216); Korea (217); Kazakhstan
(218); Kyrgyzstan (219); South Korea (220); North Korea (221); Kuwait (222);
Laos (223); Lebanon (224); Macau (225); Malaysia (226); Maldives (227);
Mongolia (228); Nepal (229); Oman (230); Pakistan (231); Paracel Islands (232);
Philippines (233); Qatar (234); Saudi Arabia (235); Singapore (236); Spratley
Islands (237); Sri Lanka (238); Syria (239); Taiwan (240); Tajikistan (241);
Thailand (242); Turkey (243); Turkmenistan (244); United Arab Emirates (245);
Uzbekistan (246); Vietnam (247); Yemen (248); Asia (249); East Timor (250)
Mexico (303); Belize (310); Costa Rica (311); El Salvador (312); Guatemala (313);
Honduras (314); Nicaragua (315); Panama (316); Central America (317); Anguilla
(320); Antigua and Barbuda (321); Aruba (322); Bahamas (323); Barbados (324);
British Virgin Islands (325); Cayman Islands (326); Cuba (327); Dominica (328);
Dominican Republic (329); Grenada (330); Guadeloupe (331); Haiti (332); Jamaica
(333); Martinique (334); Montserrat (335); Netherlands Antilles (336); St
Barthelemy (337); St Kitts-Nevis (338); St Lucia (339); St Vincent and the
Grenadines (340); Trinidad and Tobago (341); Turks and Caicos Islands (342);
West Indies (343); Argentina (360); Bolivia (361); Brazil (362); Chile (363);
Colombia (364); Ecuador (365); Falkland Islands (366); French Guiana (367);

F-1

Coded value for
FWROB /
MWROB
United States (1)
Puerto Rico or US
territory (2)

Europe (3)

Asia (4)

Latin America (5)

Response for FPOB / MPOB
Guyana (368); Paraguay (369); Peru (370); Suriname (371); Uruguay (372);
Venezuela (373); South America (374)
Bermuda (300); Canada (301); Greenland (302); St Pierre and Miquelon (304);
North America (305); Algeria (400); Angola (401); Benin (402); Botswana (403);
British Indian Ocean Territory (404); Burkina Faso (405); Burundi (406);
Cameroon (407); Cape Verde (408); Central African Republic (409); Chad (410);
Comoros (411); Congo (412); Djibouti (413); Egypt (414); Equatorial Guinea
(415); Ethiopia (416); Eritrea (417); Europa Island (418); Gabon (419); Gambia
(420); Ghana (421); Glorioso Islands (422); Guinea (423); Guinea-Bissau (424);
Ivory Coast (425); Juan de Nova Island (426); Kenya (427); Lesotho (428); Liberia
(429); Libya (430); Madagascar (431); Malawi (432); Mali (433); Mauritania (434);
Mayotte (435); Morocco (436); Mozambique (437); Namibia (438); Niger (439);
Nigeria (440); Reunion (441); Rwanda (442); Sao Tome and Principe (443);
Senegal (444); Mauritius (445); Seychelles (446); Sierra Leone (447); Somalia
(448); South Africa (449); St Helena (450); Sudan (451); Swaziland (452);
Tanzania (453); Togo (454); Tromelin Island (455); Tunisia (456); Uganda (457);
Western Sahara (458); Democratic Republic of Congo (459); Zimbabwe (461);
Africa (462); Australia (501); Christmas Island, Indian Ocean (502); Cook Islands
(505); Coral Sea Islands (506); Heard and McDonald Islands (507); Fiji (508);
French Polynesia (509); Kiribati (510); Marshall Islands (511); Micronesia (512);
Nauru (513); New Caledonia (514); New Zealand (515); Niue (516); Norfolk Island
(517); Palau (518); Papua New Guinea (519); Pitcairn Islands (520); Solomon
Islands (521); Tokelau (522); Tonga (523); Tuvalu (524); Vanuatu (525); Wallis
and Futuna Islands (526); Samoa (527); Oceania (528); at sea (554); abroad (555)

F-2

Coded value for
FWROB /
MWROB

Other Areas (6)

Appendix G: Parental Place of Birth Crosswalk for World Region of
Birth Recode in the Current Population Survey
The coding scheme for father’s and mother’s place of birth (PEFNTVTY and
PEMNTVTY) was identical to that used for person’s place of birth (PENATVTY) in the
Current Population Survey. The CPS parental place of birth variables were recoded into
father’s and mother’s world region of birth (FWROB and MWROB) according to the
following crosswalk:

Response for PEFNTVTY / PEMNTVTY
United States (057)
American Samoa (060); Guam (066); Johnston Atoll (067); Northern Marianas
(069); Midway Islands (071); Puerto Rico (072); Navassa Island (076); US Virgin
Islands (078); Wake Island (079); Baker Island (081); Howland Island (084); Jarvis
Island (086); Kingman Reef (089); Palmyra Atoll (095); US outlying area (096)
Albania (100); Andorra (101); Austria-Hungary (063); Austria (102); Belgium
(103); Bulgaria (104); Czechoslovakia (105); Denmark (106); Faroe Islands (107);
Finland (108); France (109); Corsica (111); Lorraine (113); Alsace Lorraine (176);
Germany (110); Bavaria (114); Frankfurt (177); Berlin (178); Bremen (184);
Dutchland (185); Hamburg (186); Hanover (187); Hesse (188); Lubeck (251);
Prussia (255); Saxony (260); Wurzberg (277); Gibralter (115); Greece (116); Crete
(278); Hungary (117); Iceland (118); Ireland (119); Dublin (279); Eire (280);
Galway (281); Irish Republic (284); Cork (297); Republic of Ireland (464); Italy
(120); Trieste (299); Sicily (466); Jan Meyan (121); Liechtenstein (122);
Luxembourg (123); Malta (124); Monaco (125); Netherlands (126); Rotterdam
(306); Holland (307); Amsterdam (566); Norway (127); Poland (128); Danzig
(576); East Prussia (578); Portugal (129); Azores Islands (130); Madeira Islands
(131); Romania (132); Rumania (579); Transylvania (580); San Marino (133);
Spain (134); Canary Islands (291); Espana (293); Majorca (294); Mallorca (295);
Svalbard (135); Sweden (136); Switzerland (137); Zurich (170); United Kingdom
(138); England (139); Scotland (140); Wales (141); Northern Ireland (142);
Guernsey (143); Jersey (144); Isle of Man (145); Britain (171); British Isles (172);
Channel Islands (173); Great Britain (174); UK (175); South Wales (179); Belfast
(181); Londonderry (182); N. Ireland (183); Vatican City (146); Yugoslavia (147);
Jugoslavia (189); Federal Republic of Yugoslavia (586); Czech Republic (148);
Bohemia (191); Slovakia (149); Slovak Republic (193); Slavonia (588); Bosnia and
Herzegovina (150); Bosnia (195); Herzegovina (197); Zadar (098); Croatia (151);
Pelagosa (199); Macedonia (152); Slovenia (153); Serbia (154); Estonia (155);
Latvia (156); Lithuania (157); Byelorussia (075); White Russia (080); Belarus
(160); Byelarus (198); Moldova (162); Bessarabia (563); Moldavia (564); Russia
(163); Kaliningrad (581); Russian Federation (583); Siberia (585); Ukraine (164);
Crimea (613); Ukrainia (614); Soviet Union (624); Union of Soviet Socialist
Republics (625); Europe (166); Kosovo (167); Montenegro (168)
Armenia (158); Yerevan (190); Azerbaijan (159); Georgia (161); Afghanistan
(200); Bahrain (201); East Pakistan (196); Bangladesh (202); Bhutan (203); Brunei
(204); Cambodia (206); Kampuchea (398); Kwantung (088); Mainland China
(090); People’s Republic of China (091); Red China (092); Shanghai (093); Tibet
(094); China (207); Manchuria (271); Cyprus (208); Republic of Cyprus (252);
Hong Kong (209); British Hong Kong (253); China Hong Kong (254); India (210);
Assam (256); Delhi (257); Goa (258); Hyderabad (259); Portuguese India (262);

G-1

Coded value for
FWROB/
MWROB
United States (1)
Puerto Rico or US
territory (2)

Europe (3)

Asia (4)

Response for PEFNTVTY / PEMNTVTY
Punjab, India (263); Pajasthan (264); Sikkim (265); West Bengal (266); Indonesia
(211); Borneo (267); Dutch East Indies (268); Dutch Indonesia (269); Dutch New
Guinea (270); Irian (272); Java (273); Netherlands East Indies (274); Sumatra
(275); West Timor (276); Iran (212); Iraq (213); Israel (214); Japan (215); Okinawa
(282); Jordan (216); Arab Palestine (283); Korea (217); Seoul (286); Kazakhstan
(218); Kyrgyzstan (219); South Korea (220); Republic of Korea (289); Rok (290);
North Korea (221); Democratic People’s Republic of Korea (292); Kuwait (222);
Laos (223); Loas (296); Lebanon (224); Macau (225); Macao (298); Malaysia
(226); Labuan (287); Sarawak (288); Maldives (227); Mongolia (228); Nepal (229);
Oman (230); Pakistan (231); West Pakistan (285); Punjab (308); Punjab, Pakistan
(309); Paracel Islands (232); Philippines (233); Manila (572); Republic of
Philippines (573); Filipines (575); Qatar (234); Arabia (194); Saudi Arabia (235);
Singapore (236); Spratley Islands (237); Sri Lanka (238); Ceylon (590); Syria
(239); Latakia (593); Syrian Arab Republic (598); Taiwan (240); Formosa (601);
Republic of China (603); ROC (604); Taiwan ROC (605); Tajikistan (241);
Tadzhik (606); Thailand (242); Turkey (243); Turkmenistan (244); United Arab
Emirates (245); Abu Dhabi (615); Dubai (617); Sharjah (621); Uzbekistan (246);
Vietnam (247); Da-Lat (344); Da-Nang (345); Gia-Dinh (346); Ha-Dong (347);
Hai-Phong (348); Hanoi (349); Khanh-Hung (350); Long-Xuyen (351); My-Tho
(352); Nam-Dinh (353); Nha-Trang (354); North Vietnam (355); Phan-Theit (356);
Quang-Long (357); Qui-Nhon (358); Rach-Gia (359); Cam-Pha (627); Cam-Ranh
(630); Can-Tho (633); Saigon (634); South Vietnam (640); Thanh-Hoa (648); TuyHoa (649); Vinh-Long (650); Vung-Tau (651); Yemen (248); Aden (671); South
Yemen (672); Yar (674); Yemen Arab Republic (675); Asia Minor (062); Asia
(249); Middle East (375); Palestine (376); Southeast Asia (377); West Bank (378);
Gaza Strip (379); East Timor (350); Timor (380)
Mexico (303); all Mexican states (399-401, 412, 468, 470, 474, 476, 477, 479-482,
485, 488, 494, 495, 498, 532-534, 537, 538, 540, 541, 545-548, 556, 561); British
Honduras (082); Belize (310); Costa Rica (311); Salvador (168); San Salvador
(169); El Salvador (312); Guatemala (313); Honduras (314); Nicaragua (315);
Panama (316); Balboa (567); Canal Zone (568); Cristobal (569); Panama Canal
Zone (570); Republic of Panama (571); Central America (317); Anguilla BWI
(319); Anguilla (320); Barbuda (180); Antigua WI (318); Antigua and Barbuda
(321); Antigua (694); Aruba Netherlands (059); Aruba DWI (192); Aruba (322);
Bahamas UK (065); Grand Bahama (074); Bahamas (323); Barbados (324);
Anegada (112); British Virgin Islands (325); British VI (463); Tortola (465);
Cayman Islands (326); Grand Cayman (467); Cuba (327); Havana (469); Dominica
(328); Dominica British West Indies (471); Dominica WI (472); Dominican
Republic (329); Dom Rep (473); Republica Dominicana (475); Grenada (330);
Guadeloupe (331); St Martin (478); Haiti (332); Jamaica (333); Martinique (334);
Montserrat (335); Netherlands Antilles (336); Bonaire (483); Curacao (484); St
Eustatius (486); St Maarten (487); St Barthelemy (337); St Barts (389); St KittsNevis (338); Nevis (490); St Christopher (491); St Christopher-Nevis (492); St
Kitts (493); St Lucia (339); St Vincent and the Grenadines (340); Grenadines (496);
St Vincent (497); The Grenadines (499); Trinidad and Tobago (341); Tobago (500);
Trinidad (607); Turks and Caicos Islands (342); Caicos Islands (503); Grand Turk
(504); Turks Islands (609); West Indies (343); Antilles (654); British West Indies
(659); British WI (661); BWI (663); Caribbean (667); Latin America (668);
Leeward Islands (669); Windward Islands (670); Argentina (360); Bolivia (361);
Brazil (362); Brasil (384); Easter Island (085); Chile (363); Colombia (364); San
Andres (679); Tortoise Islands (099); Ecuador (365); Galapagos Islands (381);

G-2

Coded value for
FWROB/
MWROB

Latin America (5)

Response for PEFNTVTY / PEMNTVTY
Falkland Islands (366); French Guiana (367); Guyana (368); British Guiana (529);
British Guyana (530); Guiana (531); Paraguay (369); Peru (370); Suriname (371);
Dutch Guiana (535); Surinam (536); Uruguay (372); Venezuela (373); Monagas
(539); South America (374); Burma (565)
Bermuda (300); Canada (301); all Canadian provinces (382-397); Greenland (302);
St Pierre and Miquelon (304); North America (305); Algiers (542); Algeria (543);
Benin (402); Botswana (403); British Indian Ocean Territory (404); Burkina Faso
(405); Upper Volta (549); Burundi (406); Cameroon (407); Cape Verde (408);
Central African Republic (409); Chad (410); Anojouan (097): Comoros (411);
Great Comore (557); Djibouti (413); Jibuti (562); Egypt (414); Equatorial Guinea
(415); Ethiopia (416); Eritrea (417); Europa Island (418); Gabon (419); Ghana
(421); Glorioso Islands (422); Guinea (423); Guinea-Bissau (424); Ivory Coast
(425); Cote D’Ivoire (574); Juan de Nova Island (426); Kenya (427); British East
Africa (577); Lesotho (428); Liberia (429); Libya (430); Tripoli (582); Madagascar
(431); Malagasy Republic (584); Malawi (432); Mali (433); Mauritania (434);
Mayotte (435); Morocco (436); French Morocco (589); Tangier (591);
Mozambique (437); Manica (592); Namibia (438); Niger (439); Nigeria (440);
Reunion (441); Rwanda (442); Sao Tome and Principe (443); Principe Island (599);
Sao Tome (600); Senegal (444); Dakar (602); Mauritius (445); Seychelles (446);
Sierra Leone (447); Somalia (448); South Africa (449); Republic of South Africa
(608); Transvaal (610); Union of South Africa (611); St Helena (450); Ascension
Island (612); Sudan (451); Swaziland (452); Tanzania (453); Tanganyika (616);
Zanzibar (618); Togo (454); Togoland (620); Tromelin Island (455); Tunisia (456);
Tunis (622); Uganda (457); Western Sahara (458); Democratic Republic of Congo
(459); Belgian Congo (626); Brazzaville (558); Congo (559); People’s Republic of
Congo (560); Kinshasa (628); Zaire (629); Zambia (460); Zimbabwe (461);
Rhodesia (631); Southern Rhodesia (632); Africa (462); Central Africa (635);
Eastern Africa (636); North Africa (637); Southern Africa (638); West Africa
(639); Australia (501); all Australian territories (641-647); Christmas Island, Indian
Ocean (502); Christmas Island (658); Cook Islands (505); Coral Sea Islands (506);
Heard and McDonald Islands (507); Fiji (508); Koro Island (653); French Polynesia
(509); Tahiti (655); Kiribati (510); Canton and Enderbury Islands (656); Canton
Island (657); Marshall Islands (511); Kwajalein (660); Micronesia (512); Federated
States of Micronesia (662); Ponape (664); Truk (665); Yap (666); Nauru (513);
New Caledonia (514); New Zealand (515); Niue (516); Norfolk Island (517); Palau
(518); Papua New Guinea (519); New Guinea (673); Pitcairn Islands (520);
Solomon Islands (521); Tokelau (522); Union Islands (678); Tonga (523); Tuvalu
(524); Vanuatu (525); New Hebrides (681); Wallis and Futuna Islands (526);
Samoa (527); Western Samoa (685); Oceania (528); Antarctica (686); French
Southern and Antarctic (687); Melanesia (688); Polynesia (690); Angola (544); at
sea (554); high seas (692); international waters (693); abroad (555); foreign country
(695); overseas (696)

G-3

Coded value for
FWROB/
MWROB

Other Areas (6)

Appendix H: Parental Place of Birth Crosswalk for Broad Place of
Birth Recode in the American Community Survey
The minimum amount of information about parental place of birth that is necessary to
ascertain a person’s generation status (aside from that person’s citizenship status) is
whether or not each of the parents was born in the United States, Puerto Rico, or a U.S.
territory. Hence, a broad recode for both parental place of birth variables allowed for the
assessment of the variables’ ability to produce estimates of generational status if the
original variables had proven to be unreliable at finer levels of detail. The parental place
of birth variables were recoded into father’s and mother’s broad place of birth (FBPOB
and MBPOB) with the following crosswalk:

Response for FPOB / MPOB
Any of the 50 US states or the District of Columbia (001 – 056);
United States unspecified (057); American Samoa (060); Guam
(066); Johnston Atoll (067); Northern Marianas (069); Midway
Islands (071); Puerto Rico (072); Navassa Island (076); US Virgin
Islands (078); Wake Island (079); Baker Island (081); Howland
Island (084); Jarvis Island (086); Kingman Reef (089); Palmyra Atoll
(095); US outlying area (096)
All other non-missing responses (100 – 555)

H-1

Coded value for FBPOB /
MBPOB
Born in the United States (1)

Born outside the United States (2)

Appendix I: Parental Place of Birth Crosswalks for the Selected Place
of Birth Recodes in the American Community Survey
One problem with analyzing PPOB data lies with the large number of potential response
categories. In the ACS, the place of birth variable must be collapsed for data products
because there are too few cases for a number of countries to yield reliable information at
that level of detail. Due to the smaller sample size used in the Content Test, the data
sparseness issue was of even greater concern with regards to the PPOB analysis. The
world region recodes (FWROB and MWROB) and the broad place of birth recodes
(FBPOB and MBPOB) addressed the issue well, but it came at the expense of a loss in
geographic detail. Ideally, the single-country responses should be analyzed for each
place of birth that has sufficient cell size to yield robust estimates – to this end, the
selected country of birth recodes (FSPOB and MSPOB) were used.
To construct these recodes, the ten countries reported as the most frequent father’s and,
separately, mother’s places of birth were identified using the March 2010 CPS Basic
dataset. The father’s ten places of birth each represented a separate category in FSPOB,
and likewise for mother’s ten places of birth and MSPOB. For each recode, the countries
not identified among the ten places of birth were collapsed into a residual category; thus,
FSPOB and MSPOB had 11 categories each. Note that first, the list of ten countries
derived from father’s place of birth (PEFNTVTY) did not have to match the list derived
from mother’s place of birth (PEMNTVTY), and second, the lists did not account for
sampling variability and were not intended to represent true rank order.

Response for FPOB
Canada (301)
China (207), Hong
Kong (209), Paracel
Islands (232)
Cuba (327)
Dominican Republic
(329)
El Salvador (312)
India (210)
Italy (120)
Mexico (303)
Philippines (233)

Vietnam (247)
All other places of
birth not listed above
(001-559)

Coded value for
FSPOB
Canada (1)
China (2)

Cuba (3)
Dominican Republic
(4)
El Salvador (5)
India (6)
Italy (7)
Mexico (8)
Philippines (9)

Vietnam (10)
Other places (11)

Response for MPOB
Canada (301)
China (207), Hong
Kong (209), Paracel
Islands (232)
El Salvador (312)
Germany (110)
India (210)
Italy (120)
Mexico (303)
Philippines (233)
United Kingdom (138),
England (139),
Scotland (140), Wales
(141), Northern Ireland
(142), Guernsey (143),
Jersey (144), Isle of
Man (145)
Vietnam (247)
All other places of birth
not listed above (001559)

I-1

Coded value for
MSPOB
Canada (1)
China (2)

El Salvador (3)
Germany (4)
India (5)
Italy (6)
Mexico (7)
Philippines (8)
United Kingdom (9)

Vietnam (10)
Other places (11)

Appendix J: Ancestry Crosswalk for Regional Ancestry Recode in the
American Community Survey
The first and second ancestry (ANC1 and ANC2) variables were recoded into first and
second regional ancestry (RANC1 and RANC2) with the following crosswalk:

Response for ANC1 / ANC2
Angolan (500); Benin (502); Botswana (504); Burundian (506); Cameroon (508); Cape
Verdean (510); Central African Republic (512); Chadian (513); Congolese (515);
Congo Brazzaville (516); Djibouti (519); Equatorial Guinea (520); Corsico Islander
(521); Ethiopian (522); Eritrean (523); Gabonese (525); Gambian (527); Ghanaian
(529); Guinean (530); Guinea Bissau (531); Ivory Coast (532); Kenyan (534); Lesotho
(538); Liberian (541); Madagascan (543); Malawian (545); Malian (546); Mauritanian
(547); Mozambican (549); Namibian (550); Niger (551); Nigerian (553); Fulani (554);
Hausa (555); Ibo (556); Tiv (557); Yoruba (558); Rwandan (561); Senegalese (564);
Sierra Leonean (566); Somalian (568); Swaziland (569); South African (570); Union
of South Africa (571); Afrikaner (572); Natalian (573); Zulu (574); Sudanese (576);
Dinka (577); Nuer (578); Fur (579); Baggara (580); Tanzanian (582); Tanganyikan
(583); Zanzibar Islander (584); Togo (586); Ugandan (588); Upper Voltan (589); Volta
(590); Zairian (591); Zambian (592); Zimbabwean (593); African Islands (594);
Mauritian (595); Central African (596); Eastern African (597); Western African (598);
African (599); Algerian (400); Egyptian (402); Libyan (404); Moroccan (406); Ifni
(407); Tunisian (408); North African (411); Alhucemas (412); Berber (413); Rio de
Oro (414); Bahraini (415); Afro American (900); Afro (901); African American (902);
Afghan (600); Baluchistan (601); Pathan (602); Bangladeshi (603); Bhutanese (607);
Nepali (609); Asian Indian (615); Kashmir (616); Bengali (618); East Indian (620);
Andaman Islander (622); Andhra Pradesh (624); Assamese (626); Goanese (628);
Gujarati (630); Karnatakan (632); Keralan (634); Madhya Pradesh (636);
Maharashtran (638); Madras (640); Mysore (642); Nagaland (644); Orissa (646);
Pondicherry (648); Punjab (650); Rajasthan (652); Sikkim (654); Tamil Nadu (656);
Uttar Pradesh (658); East Indies (675); Pakistani (680); Sri Lankan (690); Singhalese
(691); Veddah (692); Maldivian (695); Burmese (700); Shan (702); Cambodian (703);
Khmer (704); Chinese (706); Cantonese (707); Manchurian (708); Mandarin (709);
Mongolian (712); Tibetan (714); Hong Kong (716); Macao (718); Filipino (720);
Indonesian (730); Borneo (732); Java (734); Sumatra (736); Japanese (740); Issei
(741); Nisei (742); Sansei (743); Yonsei (744); Gonsei (745); Ryukyu Islander (746);
Okinawan (748); Korean (750); Laotian (765); Meo (766); Hmong (768); Malaysian
(770); North Borneo (771); Singaporean (774); Thai (776); Black Thai (777); Western
Lao (778); Taiwanese (782); Formosan (783); Vietnamese (785); Katu (786); Ma
(787); Mnong (788); Montagnard (790); Indo Chinese (792); Eurasian (793);
Amerasian (794); Asian (795); Iranian (416); Iraqi (417); Israeli (419); Jordanian
(421); Transjordan (422); Kuwaiti (423); Lebanese (425); Saudi Arabian (427); Syrian
(429); Armenian (431); Turkish (434); Yemeni (435); Omani (436); Muscat (437);
Trucial States (438); Qatar (439); Bedouin (441); Kurdish (442); Kuria Muria Islander
(444); Palestinian (465); Gaza Strip (466); West Bank (467); South Yemen (470);
Aden (471); United Arab Emirates (480); Assyrian (483); Chaldean (484); Syriac
(485); Mideast (490);
Albanian (100); Azerbaijani (101); Belorussian (102); Bulgarian (103); Carpatho
Rusyn (104); Carpathian (105); Rusyn (106); Ruthenian (107); Cossack (108);
Croatian (109); Czech (111); Bohemian (112); Moravian (113); Czechoslovakian
(114); Estonian (115); Livonian (116); Finno Ugrian (117); Mordovian (118); Voytak
(119); Gruziia (120); German from Russia (122); Volga (123); Rom (124); Hungarian
(125); Magyar (126); Kalmyk (127); Macedonian (130); Montenegrin (131); North

J-1

Coded value for
RANC1 /
RANC2
African (1)

Asian (2)

Eastern European
(3)

Response for ANC1 / ANC2
Caucasian (132); North Caucasian Turkic (133); Ossetian (140); Polish (142);
Kashubian (143); Romanian (144); Bessarabian (145); Moldavian (146); Wallachian
(147); Russian (148); Muscovite (150); Serbian (152); Slovak (153); Slovene (154);
Sorbian/Wend (155); Soviet Turkic (156); Bashkir (157); Chuvash (158); Gagauz
(159); Mesknetian (160); Tuvinian (161); Yakut (163); Soviet Union (164); Tatar
(165); Soviet Central Asia (167); Turkestani (168); Uzbeg (169); Georgia Cis (170);
Ukrainian (171); Lemko (172); Bioko (173); Husel (174); Windish (175);
Herzegovinian (177); Tajik (180); Eastern European (190); Bukovina (191); Silesian
(193); Central European (181);
British (011); British Isles (012); Channel Islander (013); Gibraltar (014); English
(022); Scottish (088); Welsh (097); Danish (020); Finnish (024); Scotch Irish (087);
Latvian (128); Lithuanian (129); Norwegian (082); Scandinavian (098); Irish Scotch
(094); Northern European (183); Alsatian (001); Andorran (002); Tirol (004); Faroe
Islander (023); Karelian (025); Breton (028); Frisian (029); Friulian (030); Ladin
(031); Icelander (049); Irish (050); Lapp (075); Liechtensteiner (076); Luxemburger
(077); Maltese (078); Manx (079); Monegasque (080); North Irish (081); Occitan
(083); Swedish (089); Aland Islander (090); Romansch (095); Suisse Romane (096);
Celtic (099)
Greek (046); Cretan (047); Cyclades (048); Italian (051); Trieste (052); Abruzzi (053);
Apulian (054); Basilicata (055); Calabrian (056); Amalfin (057); Emilia Romagna
(058); Rome (059); Ligurian (060); Lombardian (061); Marche (062); Molise (063);
Neapolitan (064); Piedmontese (065); Puglia (066); Sardinian (067); Sicilian (068);
Tuscany (069); Trentino (070); Umbrian (071); Valle Daost (072); Venetian (073); San
Marino (074); Portuguese (084); Azores Islander (085); Madeira Islander (086);
Spaniard (200); Andalusian (201); Asturian (202); Castillian (203); Yugoslavian (176);
Southern European (185); Cypriot (017); Greek Cypriot (018); Turkish Cypriot (019);
Spanish Basque (007); Catalonian (204); Balearic Islander (205); Gallego (206);
Valencian (207); Spanish (291); Spanish American (295); Canary Islander (208);
Galician (196);
Austrian (003); Dutch (021); French (026); Lorraine (027); German (032); Bavaria
(033); Berlin (034); Hamburg (035); Hannover (036); Hessian (037); Lubecker (038);
Pomeranian (039); Prussian (040); Saxon (041); Sudetenlander (042); Westphalian
(043); East German (044); West German (045); Swiss (091); Suisse (092); Switzer
(093); Western European (187); Basque (005); French Basque (006); Belgian (008);
Flemish (009); Walloon (010); Cornish (015); Corsican (016);
Bahamian (300); Barbadian (301); Belizean (302); Bermudan (303); Cayman Islander
(304); Jamaican (308); Trinidadian Tobagonian (314); Trinidadian (315); Tobagonian
(316); US Virgin Islander (317); St. Croix Islander (318); St. John Islander (319); St.
Thomas Islander (320); British Virgin Islander (321); British West Indies (322); Turks
and Caicos Islander (323); Anguilla Islander (324); Antigua and Barbuda (325);
Montserrat Islander (326); Kitts/Nevis Islander (327); Dominica Islander (328);
Grenadian (329); Vincent-Grenadine Islander (330); St. Lucia Islander (331); French
West Indies (332); Guadeloupe Islander (333); Cayenne (334); West Indian (335);
Haitian (336); Puerto Rican (261); Cuban (271); Dominican (275);
Mexican (210); Mexican American (211); Mexican (212); Chicano (213); La Raza
(214); Mexican American Indian (215); Mexican State (218); Mexican Indian (219);
Costa Rican (221); Guatemalan (222); Honduran (223); Nicaraguan (224); Panamanian
(225); Salvadoran (226); Central American (227); Canal Zone (229); Central American
Indian (913); Californio (292); Tejano (293); Nuevo Mexicano (294);
Argentinean (231); Bolivian (232); Chilean (233); Colombian (234); Ecuadorian (235);
Paraguayan (236); Peruvian (237); Uruguayan (238); Venezuelan (239); Criollo (248);
South American (249); Brazilian (360); South American Indian (914); Guyanese (370);
San Andres (365); Providencia (375); Surinam (380);

J-2

Coded value for
RANC1 /
RANC2

Northern
European (4)

Southern
European (5)

Western
European (6)

Caribbean (7)

Central
American (8)

South American
(9)

Response for ANC1 / ANC2
American (939); United States (940); Alabama (942); Arizona (943); Arkansas (944);
California (945); Colorado (946); Connecticut (947); District of Columbia (948);
Delaware (949); Florida (950); Idaho (951); Illinois (952); Indiana (953); Iowa (954);
Kansas (955); Kentucky (956); Louisiana (957); Maine (958); Maryland (959);
Massachusetts (960); Michigan (961); Minnesota (962); Mississippi (963); Missouri
(964); Montana (965); Nebraska (966); Nevada (967); New Hampshire (968); New
Jersey (969); New Mexico (970); New York (971); North Carolina (972); North
Dakota (973); Ohio (974); Oklahoma (976); Oregon (977); Pennsylvania (978); Rhode
Island (979); South Carolina (980); South Dakota (981); Tennessee (982); Texas (983);
Utah (984); Vermont (985); Virginia (986); Washington (987); West Virginia (988);
Wisconsin (989); Wyoming (990); Georgia (991); Southerner (993); North American
(994); Native American (917); Indian (918); Cherokee (919); American Indian (920);
Canadian (931); Newfoundland (933); Nova Scotia (934); French Canadian (935);
Aleut (921); Eskimo (922); Inuit (923); Anglo (925); Appalachian (927); Aryan (928);
Greenlander (930); Creole (907);
Australian (800); Tasmanian (801); Australian Aborigine (802); New Zealander (803);
Polynesian (808); Kapingamarangan (809); Maori (810); Hawaiian (811); Part
Hawaiian (813); Samoan (814); Tongan (815); Tokelauan (816); Cook Islander (817);
Tahitian (818); Niuean (819); Micronesian (820); Guamanian (821); Chamorro
Islander (822); Saipanese (823); Palauan (824); Marshallese (825); Kosraean (826);
Ponapean (827); Trukese (828); Yapese (829); Carolinian (830); Kiribatese (831);
Nauruan (832); Tarawa Islander (833); Tinian Islander (834); Melanesian (840); Fijian
(841); New Guinean (843); Papuan (844); Solomon Islander (845); New Caledonian
(846); Vanuatuan (847); Pacific Islander (850); Pacific (860); Chamolinian (862)
Arab (495); Arabic (496); Black (903); Negro (904); Nonwhite (905); Colored (906);
Acadian (936); Cajun (937); Dutch West Indies (310); Aruba Islander (311); St.
Maarten Islander (312); Pennsylvania German (929); Slavic (178); Slavonian (179);
White (924); Mixture (995); Uncodable Entries (996); Other Responses (998); Not
reported (999); Mulatto (908); Latin (251); Latino (252); Hispanic (290); Germanic
(194); European (195); Latin American (250);

J-3

Coded value for
RANC1 /
RANC2
North American
(10)

Oceanian (11)

Other, nonregional (12)

Appendix K: Ancestry Crosswalk for Selected Ancestry Recode in the
American Community Survey
As with the parental place of birth variables, the large number of potential responses to
the ancestry variables presented a data sparseness issue that could have complicated the
data analysis of individual ancestry groups. To counter this problem, recodes were built
from first and second ancestry (ANC1 and ANC2) into selected first and second ancestry
(SANC1 and SANC2) that focused on the most populous ancestry groups. First, ANC1
and ANC2 were recoded using a crosswalk (see below) that was based on the 15 largest
ancestry groups from Census 2000 (Brittingham and de la Cruz, 2004). Next, of these 15
groups, the ten largest were used to create recodes for SANC1 and SANC2 (See
Appendix C for a detailed explanation of the process underlying this recode method – the
main difference was that the 2009 ACS data was the source for identifying the ten
ancestry groups, rather than the March 2010 CPS Basic data). In the crosswalk below,
the ancestry recode groups marked by asterisks were not among the ten most populous
groups and therefore were coded into the residual category, “Other ancestry groups.”

Response for ANC1 / ANC2
Afro American (900); Afro (901); African American (902); Black (903);
Negro (904); Nonwhite (905); Colored (906); Creole (907); Mulatto
(908)
America (939); United States (940); Alabama (942); Arizona (943);
Arkansas (944); California (945); Colorado (946); Connecticut (947);
District of Columbia (948); Delaware (949); Florida (950); Idaho (951);
Illinois (952); Indiana (953); Iowa (954); Kansas (955); Kentucky (956);
Louisiana (957); Maine (958); Maryland (959); Massachusetts (960);
Michigan (961); Minnesota (962); Mississippi (963); Missouri (964);
Montana (965); Nebraska (966); Nevada (967); New Hampshire (968);
New Jersey (969); New Mexico (970); New York (971); North Carolina
(972); North Dakota (973); Ohio (974); Oklahoma (976); Oregon (977);
Pennsylvania (978); Rhode Island (979); South Carolina (980); South
Dakota (981); Tennessee (982); Texas (983); Utah (984); Vermont
(985); Virginia (986); Washington (987); West Virginia (988);
Wisconsin (989); Wyoming (990); Georgia (991); Southerner (993);
North American (994)
Native American (917); Indian (918); Cherokee (919); American Indian
(920)
British (011); British Isles (012); Channel Islander (013); Gibralter
(014); English (022)
Corsican (016); French (026); Lorraine (027); Breton (028); Occitan
(083)
German (032); Bavaria (033); Berlin (034); Hamburg (035); Hannover
(036); Hessian (037); Lubecker (038); Pomeranian (039); Prussian
(040); Saxon (041); Sudetenlander (042); Westphalian (043); East
German (044); West German (045)
Irish (050); Manx (079); North Irish (081)
Italian (051); Trieste (052); Abruzzi (053); Apulian (054); Basilicata
(055); Calabrian (056); Amalfin (057); Emilia Romagna (058); Rome

K-1

Coded value for interim
ancestry recodes
African American (1)

American (2)

American Indian (3)
English (4)
French (5)
German (6)

Irish (7)
Italian (8)

Response for ANC1 / ANC2
(059); Ligurian (060); Lombardian (061); Marche (062); Molise (063);
Neapolitan (064); Piedmontese (065); Puglia (066); Sardinian (067);
Sicilian (068); Tuscany (069); Trentino (070); Umbrian (071); Valle
Daost (072); Venetian (073); San Marino (074)
Mexican (210); Mexican American (211); Mexicano (212); Chicano
(213); La Raza (214); Mexican American Indian (215); Mexican State
(218); Mexican Indian (219)
Polish (142); Kashubian (143)
All other ancestry groups not listed above (001 – 999)
Scottish (088)
Dutch (021)
Norwegian (082)
Scotch Irish (087); Irish Scotch (094)
Swedish (089)

K-2

Coded value for interim
ancestry recodes

Mexican (9)

Polish (10)
Other ancestry groups (11)
Scottish (*)
Dutch (*)
Norwegian (*)
Scotch-Irish (*)
Swedish (*)

Appendix L: Language Crosswalk for Selected Language Spoken at
Home in the American Community Survey
As with the parental place of birth and ancestry questions, the large number of potential
responses to the language spoken at home question presented a data sparseness issue that
complicated the data analysis of individual language-spoken groups. To mitigate this
issue, the language spoken at home (LAN) variable was recoded into selected language
spoken at home (SLAN) representing the five largest language-spoken-at-home groups
based on the 2009 ACS data. See Appendix I for a detailed explanation of the process
underlying this recode method. The SLAN recode was constructed using the following
crosswalk:

Coded value for
SLAN

Response for LAN
Chinese (708)
French (620)
Spanish (625)
Tagalog (742)
Vietnamese (728)
All other non-English languages not listed above (001-999)

L-1

Chinese (1)
French (2)
Spanish (3)
Tagalog (4)
Vietnamese (5)
Other non-English languages (6)

Appendix M: Reliability Measures
In this report, Census Bureau analysts used data from both panels of the Content Test in
conjunction with corresponding data from the Content Follow-up survey to produce three
types of statistics that measure response reliability. For each variable examined, the
universe for calculating reliability measures was restricted to households which
participated in both surveys and also met the universe restrictions for the variable in
question. Given that a variable has i response categories, the response behaviors of
persons in the reliability universe for that variable can be organized into i different
response matrices, one for each response category of the variable (Figure M-1).

Figure M-1. Content Test / Content Follow-up response matrix for the ith response category.

CFU response
(reinterview)
Yes
No
Total

Yes

Content test response
No

Total

For response category i in a given variable, if a person’s response fell into category i for
both the Content Test and the CFU, then that person was included among the cell marked
. However, if the person’s response fell into category i for the CFU but not for the
Content Test, then that person was included among those in ; vice versa, in . Finally,
if the person’s response did not fall into category i for either survey, then that person was
included among those in . After the respondents had been distributed within the
response matrices, then the following statistics could be constructed from the components
of the matrices.

Gross Difference Rate (GDR): A measure of the gross rate of disagreement between the
content test and reinterview. The GDR was calculated for each response category of a
given variable. Displayed as a percentage, the GDR of the ith response category is
defined as:

.

M-1

Index of Inconsistency (IoI): An estimate of the magnitude of response variability for a
given item. The IoI was calculated for each response category of a given variable.
Displayed as a percentage, the IoI of the ith response category is approximately defined
as:

.

L-fold Index of Inconsistency: A weighted average of all IoI estimates corresponding to
the response categories of a given variable. The L-fold index is a measure of the overall
magnitude of response variability for the entire variable. Displayed as a percentage, the
L-fold index is defined as:

,

where the summations are done across all i response categories. It should be noted that,
when there are only two response categories, the L-fold index is equivalent to the IoIs for
either response category (which are also equal to each other); hence, L-fold calculations
and the corresponding statistical tests were not performed for variables such as father’s
and mother’s broad place of birth.
Additional information about the reliability measures used in this report can be found in
the 1990 Census of Population and Housing Evaluation and Research Report entitled
Content Reinterview Survey: Accuracy of Data for Selected Population and Housing
Characteristics as Measured by Reinterview (Bureau of the Census, 1990).

M-2

Appendix N: Information Page
Test Design
A single set of questions in two placements within the detailed person
section. One will be after the place of birth/citizenship series and before the
Treatments
school enrollment question and the other will be after the ancestry question
and before the language spoken question. (See page 4.)
35,000 households per treatment (70,000 total)
Sample Size
Similar to production ACS with an additional level of stratification into high
Sample Design
and low mail response areas.
Mail, CATI, and CAPI, with a CATI content follow-up (CFU) of all
households. CATI and CAPI interviews will be recorded using ComputerModes
Assisted Recorded Interviewing (CARI) technology.
Same schedule as the production September panel: mailout in late August,
CATI in October, CAPI in November. CFU goes from mid-September to
Time Frame
mid-December.

Research Questions & Evaluation Measures
No.
1

2
3
4

Research Questions
Are the response distributions of parental
place of birth and generational status (i.e.,
first, second, and third or higher
generation) comparable to existing data
sources?

Which placement results in a lower
missing data rate?
Do the two placements have similar or
different response distributions?
Which placement results in more reliable
estimates?

N-1

Evaluation Measures
Compare the response distributions for
the two placements to the distributions
derived from the parental place of birth
data in the Current Population Survey.
Formal statistical comparisons cannot be
made between the Content Test data and
other sources since the Content Test data
will not have been edited or imputed, nor
will there be adjustments for nonresponse
or raking to known population totals.
Compare the item missing data rates
between the two placements.
Compare the response distributions
between the two placements.
Using data from the Content Test and the
Content Follow-up (CFU), compare the
indices of inconsistency, and the L-fold
index of inconsistency between the two
placements.

No.
5

6

7

8

9

10

Research Questions
Does changing the placement of the
parental place of birth questions from
before to directly after the ancestry
question affect the item missing data
rates, response distributions, or reliability
for the ancestry question?
Does the placement of the parental place
of birth questions directly before the
school enrollment question affect the item
missing data rate, response distribution, or
reliability for the school enrollment
question?
Only the first part of the school
enrollment question set will be included
in the Content Test.
Does the placement of the parental place
of birth questions directly before the
language spoken questions affect the item
missing data rates, response distributions,
or reliability for the language spoken
questions?
For each mode of data collection, do the
two placements have differential item
missing data rates, response distributions,
or reliability of the data?

For each mail response stratum, do the
two placements have differential item
missing data rates, response distributions,
or reliability of the data?
Does either placement elicit respondent or
interviewer behaviors that may contribute
to interviewer or respondent error?

N-2

Evaluation Measures
Compare the item missing data rates,
response distributions, and reliability
measures for ancestry between the two
placements. Compare the percent of
responses that report more than one
ancestry between the two placements.
Compare the item missing data rates,
response distributions, and reliability
measures between the two placements.

Compare the item missing data rates,
response distributions, and reliability
measures between the two placements.

For each mode (mail, CATI, CAPI),
compare the item missing data rates,
response distributions, and reliability
measures between the two placements.
Comparisons across modes of data
collection cannot be made since
measurable differences cannot be
attributed strictly to the mode of data
collection. Observed differences across
modes may also be due to mode specific
respondent characteristics and
reinterview mode effects (CFU only).
For each mail response stratum (high and
low), compare the item missing data rates,
response distributions and reliability
measures between the two placements.
Compare the behavior coding results
derived from the CARI recordings
between the two placements.

Selection Criteria (In order of priority)
Research
Question
1
2-4
5-7

Criteria
The parental place of birth response distributions should be comparable to the
distributions from CPS ASEC to be considered acceptable for the ACS.
The item missing data rates, response distributions, and reliability measures will
be considered together when determining which placement performed better.
The response distributions, item missing data rates, and reliability measures for
ancestry, school enrollment and language spoken will be considered together
when determining which question placement performed better.

Supplemental Information
Research
Question
8-10

Criteria
Not part of the selection criteria. These data are presented to give additional
information regarding how the question placement performed.

N-3