April 2011 ACS Internet Test- Attitudes and Behavior Study Follow-up

March 7, 2012
AMERICAN COMMUNITY SURVEY RESEARCH AND EVALUATION REPORT MEMORANDUM SERIES
#ACS12-RER-16
Research and Methodology Directorate Research Report Series
(Survey Methodology #2012-03)
MEMORANDUM FOR

ACS Research and Evaluation Steering Committee

From:

Peter Miller /Signed/
Chief, Center for Survey Measurement

Prepared by:

Elizabeth Nichols
Center for Survey Measurement

Subject:

The April 2011 American Community Survey Internet Test:
Attitudes and Behavior Study Follow up

Attached is the final American Community Survey Research and Evaluation report “The April
2011 American Community Survey (ACS) Internet Test: Attitudes and Behavior Study Follow
up.” This report summarizes the findings from a follow up survey of mail responders, Internet
responders and nonresponders in the experimental treatments of the April 2011 ACS Internet
Test. The purpose Attitudes and Behavior Study was to determine whether anything in the
mailing materials or messages motivated some to choose the Internet to report to the ACS and
others to choose the paper form. This follow up also investigated why some in the Internet
Test did not respond at all.
If you have any questions about this report, please contact Elizabeth Nichols (301-763-1724) or
Peter Miller (301-763-9593)
Attachment
cc:
ACS Research and Evaluation Team

Debbie Griffin
ACSO
Todd Hughes
Tony Tersine
Jennifer Tancreto
DSSD
Steve Hefter
Mary Davis
Rachel Horwitz
Mary Frances Zelenak
Brenna Matthews
Michelle Ruiter
Scott Boggess
HHES
Colleen Hughes
POP
Anne Ross
Janice Valdisera
Nancy Bates
DIR
Kathy Ashenfelter
CSM
Jenny Childs

American Community Survey Research and Evaluation Program
March 7, 2012

The April 2011 American
Community Survey Internet
Test:
Attitudes and Behavior Study
Follow up
FINAL REPORT

Elizabeth Nichols
Center for Survey Measurement

Intentionally Blank

TABLE OF CONTENTS
EXECUTIVE SUMMARY...................................................................................................................... v
Test Objective ............................................................................................................................... v
Methodology ................................................................................................................................ v
Research Questions and Results .................................................................................................. v
1. BACKGROUND ............................................................................................................................. 1
1.1 Motivation for the April 2011 ACS Internet Test Attitudes and Behavior Study .................. 1
1.2 Previous Testing .................................................................................................................... 1
2. METHODOLOGY ........................................................................................................................... 2
2.1 ACS Internet Test Experimental Treatments ......................................................................... 2
2.2 ACS Internet Follow-up Interview Design.............................................................................. 4
2.3 Sample Selection and Response Rates .................................................................................. 6
2.4 ABS Questionnaire Design ..................................................................................................... 7
2.5 Analysis Methods .................................................................................................................. 8
3. LIMITATIONS................................................................................................................................ 8
3.1 No Push Regular mail respondents were selected for the ABS. ............................................ 8
3.2 Nonresponse group for all four notification strategies is reduced for analysis in this report.
..................................................................................................................................................... 8
3.3 Low response rate for nonrespondent group ....................................................................... 9
3.4 Coding of responses .............................................................................................................. 9
3.5 ABS Questionnaire Responses ............................................................................................... 9
4. RESULTS ....................................................................................................................................... 9
4.1 What is the respondent’s motivation for responding to the ACS and what are the reasons
for not responding to the ACS? ................................................................................................. 10
4.1.1 What are the reasons for choosing the Internet survey to answer the ACS? ............. 10
4.1.2 What are the reasons for choosing the paper form to answer the ACS? .................... 12
4.1.3 What are the reasons for not responding to the ACS?................................................ 14
4.2 Was the mode choice message effectively communicated in the different notification
strategies? ................................................................................................................................. 18
4.2.1 Did respondents recall the mailing materials containing the mode choice message? 18
4.2.2 What messages in the letter were salient to respondents? ........................................ 20
4.2.3 Did respondents mention the “green” messages in the Push strategies? .................. 23
4.2.4 Did respondents recall the mode choice message? .................................................... 23

i

4.3 How did Internet respondents, mail respondents and nonrespondents interact with the
online form? .............................................................................................................................. 25
4.3.1 Did Internet respondents complete the ACS at home, work, or some other place? .. 25
4.3.2 Did Internet respondents use the paper form as a guide and did they fill out the
paper form? .......................................................................................................................... 26
4.3.3 Were there usability problems with the online ACS?.................................................. 26
4.3.4 Did mail respondents or nonrespondents in the Push strategies consider answering
the ACS using the Internet survey more so than the mail and nonrespondents in the Choice
strategies?............................................................................................................................. 27
4.3.5 Did mail respondents or nonrespondents try to access the Internet form or start the
Internet form? If they did so, were there usability problems with the form or login task? 27
4.4 What are the attitudes toward completing a form online? ............................................. 28
4.4.1 What is the mode preference for nonrespondents, mail control, and mail Internet
Test respondents who did not realize there was an online survey option?......................... 28
4.4.2 Would respondents have completed the ACS if their first mode choice was not
available? .............................................................................................................................. 29
4.5 What are the characteristics of Internet respondents, mail respondents and
nonrespondents? ...................................................................................................................... 30
4.5.1 What percent of mail respondents, Internet respondents and nonrespondents use
the Internet, and therefore have Internet access? .............................................................. 30
4.5.2 How much, and for what purposes, do the mail respondents, Internet respondents,
and nonrespondents use the Internet? ................................................................................ 32
4.5.3 What are the demographic characteristics of ABS Internet responders, mail
responders and nonresponders? .......................................................................................... 33
5. SUMMARY ................................................................................................................................. 34
6. FUTURE RESEARCH .................................................................................................................... 36
Acknowledgements ....................................................................................................................... 37
References ..................................................................................................................................... 37
Appendix A April 2011 ACS Internet Test Questionnaire Cover .................................................. A-1
Appendix B April 2011 ACS Internet Test Letter for the Initial Mailing Package......................... B-1
Appendix C April 2011 ACS Internet Test Replacement Letter ................................................... C-1
Appendix D April 2011 ACS Internet Test Internet Instruction Card ........................................... D-1
Appendix E ABS Questions for Internet Respondents ................................................................. E-1
Appendix F ABS Questions for Mail Respondents ....................................................................... F-1
Appendix G ABS Questions for Nonrespondents ........................................................................ G-1
Appendix H ABS Questions for Internet Use – For everyone in the ABS..................................... H-1
Appendix I Demographic Questions for Nonrespondents ........................................................... I-1

ii

LIST OF TABLES
Table 1. Percent of 457 ACS Internet respondents who did not know they could have chosen to
report by paper by notification strategy .................................................................................................... 10
Table 2. Percent of ACS Internet respondents who reported these reasons for deciding to
complete the ACS on the Internet by notification strategy ........................................................................ 11
Table 3. Percent of 335 ACS mail respondents who did not know they could have chosen to
report via the Internet by notification strategy .......................................................................................... 12
Table 4. Percent of ACS mail respondents who reported these reasons for deciding to complete
the ACS using the paper form instead of via Internet by notification strategy ......................................... 13
Table 5. Percent of nonrespondents who reported these reasons for not responding to the April
ACS by notification strategy ........................................................................................................................ 15
Table 6. Percent of ABS respondents who opened the envelope and who recalled each piece of
mailing materials by response group (Internet, Mail and Nonrespondents (NR)) and notification
strategy ........................................................................................................................................................ 19
Table 7. Percent of ABS respondents who read the letter and who recalled these messages in
the letter by response group (Internet, Mail and Nonrespondents (NR)) and notification strategy.......... 21
Table 8. Percent of respondents who reported knowing about both Internet and paper
reporting modes by response group and notification strategy ................................................................. 24
Table 9. Where Internet responders completed the online ACS ............................................................... 25
Table 10. Percent of ABS mail respondents and nonrespondents who considered answering by
the Internet ................................................................................................................................................. 27
Table 11. Stated mode preference to Census Bureau surveys among selected ABS respondents ............ 29
Table 12. Percent of respondents who said they would not complete the ACS in the other mode
if the mode they chose was not available by notification strategy ............................................................ 30
Table 13. Percent of ABS respondents who reported using the Internet at home, at work, or at
some other place by response group and notification strategy ................................................................. 31
Table 14. Internet usage (weighted mean of hours on the Internet per day and median
frequency of Internet activities performed) by ABS response group ......................................................... 32
Table 15. Demographic Characteristics of ABS Internet, Mail-IT, and Nonrespondents ............................ 34

iii

LIST OF FIGURES
Figure 1. April 2011 ACS Internet Test Materials in Initial and Replacement Mailings ................................ 4
Figure 2: ACS Internet Test (IT) and Attitudes and Behavior Study Timing.................................................. 5
Figure 3. ABS sample size and response rates.............................................................................................. 7

iv

EXECUTIVE SUMMARY
Test Objective
In April 2011, the U.S. Census Bureau conducted the 2011 American Community Survey (ACS)
Internet Test to evaluate the feasibility of providing an Internet reporting option to addresses
selected for the ACS (Tancreto, Zelenak, Davis, Ruiter, & Matthews, 2012). In addition to the
production sample which did not receive an Internet reporting option, four experimental
treatments or notification strategies were designed to test different ways to inform potential
respondents of the Internet option. Two “Push” strategies directed respondents to use the
Internet option (by providing only the Internet option in the initial mailing package) and the
other two “Choice” strategies offered the Internet and a paper form as alternate choices. The
main objective of that test was to determine the best way to present the Internet mode in the
ACS mailings to maximize self-response.
Within approximately three weeks of the April Internet Test, the Census Bureau conducted a
study called the Internet Follow-up (IFU) to collect qualitative feedback on the mailing materials
used in the test, to learn preferences with regard to reporting via the Internet, and to
understand the reasoning behind the selection of mode of response (or the decision not to
respond.) The results for this portion of the IFU, called the Attitudes and Behavior Study (ABS),
are reported here.

Methodology
The ABS was a computer-assisted telephone interview study. ACS responding and
nonresponding households with telephone numbers were sampled for the study. Three groups
from the Internet Test were sampled: Internet responders, mail responders and nonresponders.
In addition, a control group of mail respondents from the April 2011 ACS production sample
were sampled for the ABS.
The ABS interviews consisted largely of open questions, gauging respondents’ reactions to the
mailing materials used in the Internet Test, their preferences with regard to reporting via the
Internet, and their reasons for the selection of mode of response (or the decision not to
respond).
The ABS response rate was 84 percent for Internet responders; 87 percent for Internet Test mail
responders; 38 percent for nonresponders; and 82 percent for control mail responders. All
total, 1,700 interviews were available for analysis.

Research Questions and Results
What is the respondent’s motivation for responding to the ACS in the selected mode?
Internet respondents reported choosing that mode because it was easy, faster, and
convenient. Across all notification strategy treatments, reported ignorance of the paper
form was also a reason cited for completing the survey on the Internet.

v

The primary reason cited by mail respondents for using the paper form was ignorance of
the Internet option.
What are the reasons for not responding to the ACS?
Not receiving the ACS envelope and not opening the envelope were the top two reasons
given for not responding to the ACS across all notification strategies.
There was no support for a hypothesis mentioned in the literature (Millar & Dillman,
2011) that presenting respondents with a choice of mode, per se, “paralyzes” them,
leading to nonresponse.
Was the mode choice message effectively communicated in the different notification strategies?
Generally, the mode choice message was placed on materials that were salient to ABS
respondents.
The mode choice message was one of the more salient messages spontaneously recalled
from the letter, second only to the message informing respondents that survey
participation was mandatory.
When explicitly asked, not all Internet respondents were aware of the paper form, and
not all mail respondents in the Internet Test reported knowing there was an Internet
survey option. About 42 percent of the control mail respondents reported that there
was an Internet option, despite the fact that they did not have an Internet option nor
were they notified of one.
What are the characteristics of Internet respondents, mail respondents, and nonrespondents?
Internet respondents tended to use the Internet more hours on a daily basis and to
perform Internet activities more often than either mail respondents or nonrespondents.
How did Internet respondents, mail respondents, and nonrespondents interact with the online
ACS survey?
Most of the time, the online ACS form was completed at home, followed by at work, and
then at a place other than home or work.
What are the attitudes toward completing the ACS online?
Approximately 30 percent of the control mail sample said that they prefer the Internet.
The percent of mail respondents who reported that they would not complete an online
ACS form if the paper form was not available ranged from approximately 13 to 31
percent.

vi

1. BACKGROUND
1.1 Motivation for the April 2011 ACS Internet Test Attitudes and Behavior Study
The Attitudes and Behavior Study (ABS) was designed to measure why respondents chose the
Internet survey or paper form to respond to the April 2011 American Community Survey (ACS)
Internet Test and whether anything specific in the mailing materials tested affected mode
choice. The secondary objective was to measure why some people did not respond to the ACS
at all, and whether nonresponse had to do with the multi-mode offer in the test (Nichols, 2011).
A frequent survey finding is that when people are given a choice of responding by Internet or
mail, they are less likely to respond than when only a mail response is offered (Gentry & Good,
2008; Griffin, Fischer, & Morgan, 2001; Grigorian & Hoffer, 2008; Millar & Dillman, 2011; Smyth,
Dillman, Christian & O’Neill, 2010). We do not know whether they do not respond because they
actively cannot decide between the modes, whether they put off the decision so much so that
they forget about the task, whether the choice itself makes them reconsider reporting
altogether or makes the reporting task seem not as urgent, or whether some other mechanism
is at work. One explanation, what might be called, “mode paralysis,” is explained in Millar and
Dillman (2011). They cite psychological research which posits that “…every option has
opportunity costs associated with it, and when two options are compared to each other,
individuals must consider tradeoffs. This makes each option appear less appealing than it would
if offered alone, leading to no compelling reason to select either one… This suggests that by
offering a choice between Web or mail response, surveyors are certainly not encouraging
response and, in fact, they may even be discouraging it” (p. 252).
“Mode paralysis” was a concern of Census Bureau staff prior to the fielding of the April 2011 ACS
Internet Test. In fact, the ABS was designed as a precautionary measure, just in case the 2011
ACS Internet Test followed the pattern of the 2000 ACS Internet Test, where the panel with the
multiple-mode offer had a lower response rate than the control panel, which only offered the
traditional paper form (Griffin, Fischer & Morgan, 2001). In contrast to that study and the other
studies highlighted previously, there was no evidence of lower response rates in the
experimental groups who were offered a mode choice in the 2011 April ACS Internet Test
(Tancreto, Zelenak, Davis, Ruiter, & Matthews, 2012). Nonetheless, the ABS provides a rich
dataset to examine the knowledge, attitudes and behaviors associated with respondents and
nonrespondents of the ACS Internet Test.

1.2 Previous Testing
A portion of the 2011 ABS questions were similar to those questions used in the Applied
Behavior Analysis in the 2000 ACS Internet Test (Nichols & Marquis, 2001). In 2000, the Applied
Behavior Analysis questionnaire was administered to a sample of the ACS nonrespondents in the
experimental panel to find out why they did not respond to either mode. Prior to data
collection, mode paralysis was suggested as a reason for nonresponse in that test.
The results of the 2000 Applied Behavior Analysis were limited by the small sample size. Only 58
ACS “nonresponders” responded in the Applied Behavior Analysis study. Furthermore, the
study was conducted five to seven months after the mailout of the multi-mode ACS

1

experimental panel. Many of the “ACS nonresponders” who answered the Applied Behavior
Analysis questions did not even remember the ACS mailing package. Most of the people who
claimed to have opened the package said that they considered answering the ACS by paper.
Thus, the analysis did not find evidence to suggest that offering an Internet response option had
somehow led to the nonresponse. Additionally, the analysis did not find that the
nonrespondents were still trying to decide which mode to complete.
The 2011 ABS represents a more systematic attempt to uncover the reasons for nonresponse or
response by a particular mode. Approximately 1,700 households completed the ABS in 2011.
The follow-up interview took place much closer in time to when the ACS packages were mailed
out. In addition to surveying nonrespondents, the ABS also surveyed mail respondents and
Internet respondents to learn about their motivators for responding.

2. METHODOLOGY
2.1 ACS Internet Test Experimental Treatments
The 2011 April ACS Internet Test occurred concurrent with the April 2011 ACS panel. The
Internet Test ran from late March to late May. In addition to having a control group drawn from
the production sample for whom no Internet option for reporting was available, the April 2011
ACS had four experimental conditions or notification strategies (Zelenak, Ruiter, Davis, Horwitz,
& Tancreto, 2010). There were two “Choice” strategies and two “Push” strategies. The
strategies varied when the different reporting modes were offered and the emphasis placed on
the Internet reporting mode. These changes were implemented in the ACS mailing pieces sent
to respondents (pre-notice letter, initial questionnaire mailing, reminder postcard, and, for
nonrespondents only, replacement questionnaire mailing and an additional reminder postcard1).
In the two “Choice” strategy conditions, the paper form and the Internet reporting option were
offered concurrently. These treatments were called the Prominent Choice and the Not
Prominent Choice strategies. The strategies differed in the emphasis on the Internet reporting
option. The Prominent Choice strategy explicitly mentioned the Internet option in the text of
the letters and mailing materials and the Not Prominent Choice strategy only subtly mentioned
the Internet option on the paper form itself. As explained in Tancreto et al., (2012), “…this
strategy was to provide the Internet option to those who were looking for it while attempting to
alleviate a respondent’s tendency to do nothing when offered response mode choices... .”
The two “Push” strategies offered the Internet mode first. For these treatments, access
information for the Internet survey was offered at the time of the initial mailing and a paper
form was offered in a replacement mailing for households that had not responded. The two
Push strategies differed in how much time had elapsed between the mailing of the initial and
replacement mailings, with the Push Accelerated strategy sending the form approximately one
week earlier than the Push Regular strategy.
Copies of the questionnaire cover used in each strategy are presented in Appendix A. Copies of
the letters used in the initial mailings are shown in Appendix B. A copy of the letter used in the
1

The additional reminder postcard was mailed only to nonrespondents for whom the Census Bureau did

2

replacement mailing for the Push Regular treatment is presented in Appendix C. The Englishlanguage Internet Instruction Card is shown in Appendix D.
The letter sent in both the initial and replacement mailings was used to communicate the mode
choice. The letter in the initial mailing explained the response mode options (paper or Internet)
in the Prominent Choice and the two Push notification strategies. There was no mention of an
Internet reporting option in the letter for the initial mailing in the Not Prominent Choice
strategy. The letter in the replacement mailing followed the same pattern; that is, the two
reporting modes were mentioned in the Prominent Choice and two Push strategies only.
In the two Push strategies, the letter in the initial mailing instructed households to complete the
survey online, and, if they could not complete the survey online, they were to wait for a paper
questionnaire to be mailed. The same initial letter was used for both Push notification
strategies. The Push strategies’ letter emphasized that completing the survey online conserves
natural resources, saves taxpayers’ money, and allows for quicker data processing. The
replacement mailing letter was nearly identical between the two Push strategies, and offered
the choice of modes. The only difference was the reference to when the initial mailing package
was mailed (two weeks ago or three weeks ago).
In the Prominent Choice strategy letter, respondents were offered both response options (paper
and Internet) and no emphasis was given to either. The Internet option was mentioned first in
the letter in the initial mailing. The paper questionnaire was mentioned first in the replacement
mailing.
The Internet option was mentioned on the front cover of the paper questionnaire in all of the
notification strategies, with the offer being somewhat “buried” among other messages in the
Not Prominent Choice strategy.
In the Push strategies and Prominent Choice strategy, an Internet Instruction card was mailed in
both the initial and replacement mailings. This card contained yet another notification of the
mode choice for the Prominent Choice strategy and the log-in information for the Internet for
the two Push strategies.
In the mail control ACS panel, no mention of the Internet option was made in any of the mailing
materials.
Figure 1 displays the materials used in each of the mailings (initial or replacement) by
notification strategy and whether or not the materials contained reference to the Internet
reporting option, denoted by an (I).

3

Figure 1: April 2011 ACS Internet Test Materials in Initial and Replacement Mailings
Materials in either
the Initial or
replacement
Not Prominent
mailing
Prominent Choice
Choice
Push Regular
*
Form
Initial (I)
Initial (I)
Replacement (I)
Replacement (I)
Replacement (I)
Letter
Initial (I)
Initial
Initial (I)
Replacement (I)
Replacement
Replacement (I)
Internet
Initial (I)
Initial (I)
Instruction Card
Replacement (I)
Replacement (I)
Questionnaire
Initial
Initial
Replacement
Guide
Replacement
Replacement
FAQ
Initial
Initial
Initial
Replacement
Replacement
Replacement
Return Envelope
Initial
Initial
Replacement
Replacement
Replacement
* (I) means Internet option is mentioned on the material

Push Accelerated
Replacement (I)
Initial (I)
Replacement (I)
Initial (I)
Replacement (I)
Replacement
Initial
Replacement
Replacement

2.2 ACS Internet Follow-up Interview Design
This study was part of a larger follow-up survey conducted by the Census Bureau called the ACS
Internet Follow-up (IFU). A sample of the addresses selected for the ACS Internet Test was
recontacted for the IFU, which had two components: the ABS and the Content Reinterview
Assessment (CRA). The CRA is a separate project; its research goals are explained in Horwitz
(2011) and results are found in Horwitz (forthcoming).
Four groups were sampled for the IFU: ACS Internet Test mail respondents, ACS Internet Test
Internet respondents, Internet Test nonrespondents, and control mail respondents.
- There were 6,000 CRA sampled cases for the ACS Internet Test mail respondents and
560 were subsampled for the ABS. In the 560 cases with both an ABS and CRA
component, the ABS questions preceded the CRA questions.
- There were 6,000 CRA sampled cases for the ACS Internet Test Internet respondents and
560 were subsampled for the ABS. In the 560 cases with both an ABS and CRA
component, the ABS questions preceded the CRA questions.
- There were 2,900 ACS Internet Test addresses, which did not submit a response by
Internet or mail, sampled for the ABS. These are the nonrespondents2. These cases did
not receive the CRA questions.
- There were 140 mail control respondents from the production panel, who did not have
an opportunity to report online, sampled for the ABS. These cases did not receive the
CRA questions.
Internet respondents who reported in Spanish did not receive ABS questions, but could have
received the CRA questions.

2

Households that logged on to the Internet, but did not submit data over the Internet are included in the
nonrespondent universe in addition to those households which never logged on or returned a paper form.

4

The IFU Internet and mail universe for sample selection was households where the phone
number and respondent name were available. The IFU was conducted only with the original
ACS respondent. For mail respondents, if no respondent name was provided, Person 1 was used
as the default respondent.
Nonrespondent households were eligible for sample selection only if a telephone number could
be located. For these cases, the IFU (which consists only of the ABS questions) was conducted
with someone 15 years or older who typically handled the mail and was knowledgeable about
the household.
Mail and Internet households were selected for the IFU at two different times (April 12 and May
1). The early date allowed the Census Bureau to start interviewing very soon after an ACS
response was received. The goal of selecting the early date was to reduce memory error
associated with the ACS experience. Nonrespondent households were selected on May 1,
2011. This was the date that a telephone nonresponse follow-up operation would have started
if such an operation had been part of this test, just as was done with the production sample.
The IFU was conducted through a centralized Computer-Assisted Telephone Interview (CATI)
operation. Interviewers from the telephone centers in Jeffersonville and Tucson conducted the
interviews. Interviewers were trained for one day for this study. The ABS interviewing period
was from mid-April to mid-May of 2011.
Figure 2 shows the timing of the mailouts for the April ACS Internet Test, the sampling for the
ABS, and the ABS data collection.
Figure 2: ACS Internet Test (IT) and Attitudes and Behavior Study Timing
Not Prominent
Dates in 2011
Control
Prominent Choice
Choice
March 24
Mailed ACS
Mailed ACS IT
Mailed ACS IT
Pre-notice letter
Pre-notice letter
Pre-notice letter
March 28
Initial ACS mailout
Initial IT mailout
Initial IT mailout
March 31
Reminder Postcard Reminder Postcard Reminder Postcard
st
st
April 12
1 cut for ABS
1 cut for ABS
sample selection
sample selection
(mail and Internet)
(mail and Internet)
April 14
April 20
April 21
May 1

May 9
May 23
May 31

Replacement ACS
mailout
Cut for ABS sample
selection (mail)

Began ABS
interviewing
Ended ABS data
collection

Push Regular

Push Accelerated

Mailed ACS IT
Pre-notice letter
Initial IT mailout
Reminder Postcard
st
1 cut for ABS
sample selection
(Internet)

Mailed ACS IT
Pre-notice letter
Initial IT mailout
Reminder Postcard
st
1 cut for ABS
sample selection
(Internet)
Replacement ACS
IT mailout
Began ABS
interviewing

Began ABS
interviewing
Replacement ACS
IT mailout
nd
2 cut for ABS
sample selection
(NR,mail and
Internet)

Began ABS
interviewing
Replacement ACS
IT mailout
nd
2 cut for ABS
sample selection
(NR,mail and
Internet)

Began ABS
interviewing
Replacement ACS
IT mailout
nd
2 cut for ABS
sample selection
(NR,mail and
Internet)

Ended ABS data
collection
Ended IT

Ended ABS data
collection
Ended IT

Ended ABS data
collection
Ended IT

5

nd

2 cut for ABS
sample selection
(NR,mail and
Internet)

Ended ABS data
collection
Ended IT

2.3 Sample Selection and Response Rates
Within each of three groups in the ACS Internet Test (i.e., Internet respondents, mail
respondents, and nonrespondents), cases were selected from each of the four notification
strategies, with the goal of having 100 ABS respondents for each of the strategies within each
response group. All of those groups had the opportunity to report either by paper or by the
Internet to the ACS.
Sample was also selected to obtain 100 responses from mail respondents in the control panel part of the April 2011 ACS production sample. This group did not have the opportunity to
respond online. We selected the sample in this way so that we could compare findings among
the different notification strategies and compare findings among response group types.
The goal of 100 completed interviews per notification strategy per response group was met with
one exception. We did not have enough interviews for the Push Regular mail respondents. This
limitation is discussed in Section 3.
The ABS response rates were as expected, with relatively high rates for Internet and mail
respondents (between 71% and 100%) and lower rates for nonrespondents (between 33% and
42%). Having said that, the response rates for the nonrespondent group were actually better
than we had anticipated when the sampling plan was drawn up; thus, the nonresponse sample
available for analysis was substantially larger than expected.
The number of cases selected, weighted response rate (RR), and number of cases used for
analysis are shown in Figure 3. The response rate was calculated using the response rate
outlined in Marquette (undated) for production ACS CATI. Completes and partials were
included in the numerator and all cases in the denominator were considered eligible. The final
data file for analysis was created independently and some cases which were part of the
response rate calculation were not used or were missing in the analysis file. A detailed
explanation of the missing cases is available in the footnotes of Figure 3.

6

Figure 3: ABS sample size and response rates
Prominent
Response groups Control
Choice
Sample: 140
1
Internet
No sample
RR: 71.7%
Analysis: 105
2

Mail

Sample: 140
RR: 81.9%
Analysis: 113

Not Prominent
Choice
Sample: 140
RR: 83.4%
Analysis: 115

Sample: 140
RR: 90.8%
Analysis: 119

Sample: 140
RR: 85.0%
Analysis: 104

Push Regular
Sample: 140
RR: 85.8%
Analysis:119
3

Sample: 1
RR: 100%
Analysis:0

Push
Accelerated
Sample: 140
RR: 87.2%
Analysis:118
Sample: 140
RR: 84.5%
Analysis:112

Sample: 725
Sample: 725
Sample: 725
Sample: 725
RR: 33.8%
RR: 38.4%
RR: 41.4%
RR: 40.3%
Analysis:180
Analysis: 196
Analysis:191
Analysis:228
Source: U.S. Census Bureau, Attitudes and Behavior Study, 2011 ACS Internet Test Follow up
1
Two Internet cases were considered complete or partially complete cases for the response rate calculation, but were
not used during analysis because they contained very few responses. There were 15 missing Internet cases when files
were transferred for analysis. This issue was not resolved prior to writing this report.
2
There were 32 missing mail cases in the Internet Test when files were transferred for analysis. This issue was not
resolved prior to writing this report.
3
At the time of ABS sample selection, there was only one completed Push Regular mail respondent case that had
gone through Census Bureau processing. This case completed the ABS, but was not included in the analysis because
the results could not be generalized to that panel.
4
Approximately 26 percent of the cases in the nonrespondent group who answered the ABS were not used during
data analysis because we determined that they were actually late responders.
NR

4

No sample

2.4 ABS Questionnaire Design
A slightly different ABS CATI survey was administered to each of the three response groups (i.e.,
Internet, mail and nonrespondents). Many of the questions were identical across the surveys
and the same topics were covered with a few exceptions.
Staff from the Center for Survey Measurement (CSM) and ACS staff in the Decennial Statistical
Studies Division (DSSD) designed the ABS questionnaire to collect feedback on the salience of
the mailing materials and messages, knowledge of the mode choice, consideration of reporting
by the Internet, and mode preference. The questionnaires also included questions about
whether respondents tried to complete either the paper ACS form or the Internet survey
unsuccessfully and where break-offs occurred. Internet usage was collected for everyone and
demographics were collected for nonrespondents only.
Cognitive testing of the ABS was not conducted prior to the field period due to tight
questionnaire development timing. Internet usage and demographic questions were modified
slightly from the Census Barriers, Attitudes, and Motivators Survey (CBAMS) questions (Johnson,
2009). Some of the ACS demographic questions were lengthy and we did not want to use them
in these telephone contacts. Many questions in the main questionnaire were taken from the
Applied Behavior Analysis questionnaire of the 2000 ACS Internet Test (Nichols & Marquis,
2001). However, there were some new questions asked in the ABS which had not been fielded
previously and were not pretested.
The ABS questions for the three response groups, including the screener questions used to
identify the correct respondent, are listed in Appendices E-I.

7

The ABS contained many open questions to be field-coded by interviewers. Responding to open
questions relies on respondent recall rather than recognition and it is considered more difficult
than responding to closed questions (MacDougall, 1904; Cannell, Miller & Oksenberg, 1983;
Haist & Shimamura, 1992). The development of pre-coded response choices helped reduce the
amount of recoding needed. The goal of the open questions was to obtain salient themes
recalled by respondents and nonrespondents without the influence of response choices read by
an interviewer. Most of the questions and response choices were not pretested.
The ABS survey was approximately 10 minutes long and was only administered in English.

2.5 Analysis Methods
SAS (www.sas.com) was used to produce the statistics in this report. All results are weighted to
reflect their probability of selection into the ABS. They were not weighted to either the ACS
sample (to account for nonresponse) or to national levels. The author used the Taylor series
expansion method found in SAS proc surveyfreq and proc surveymeans to compute standard
errors that reflect the sample design. The standard errors for results discussed in the text are
found in the tables provided.
All statistical analyses employed the Wald Chi-square test of Independence for the comparisons.
Because so few comparisons were made, they were not adjusted for multiple comparisons.

3. LIMITATIONS
3.1 No Push Regular mail respondents were selected for the ABS.
The ABS sample selection occurred before a sufficient number of Push Regular mail respondents
had returned the completed mail form. That treatment had received the paper questionnaire
only 11 days prior to selecting the ABS sample (see Figure 2). At the time of the ABS sample
selection, only one ACS mail respondent in the Push Regular strategy had returned his/her form
and it had been checked in and keyed at the Census Bureau. This household was selected for
the ABS and actually completed the ABS. But because there was only one response in that
treatment for that response group, I removed it from the ABS analysis.

3.2 Nonresponse group for all four notification strategies is reduced for analysis in this
report.
Early in the analysis, the author observed that many of 1,082 nonrespondents reported mailing
in their ACS form. DSSD staff matched the entire ABS nonrespondent group to the ACS Internet
Test data and found that in fact 287 of them had actually completed the ACS, albeit later than
the cut off for the ABS nonresponse group selection. Some had responded before the ABS
interview took place and some responded after the ABS interview occurred. Most responded
using the paper form, but some used the Internet. All 287 late responders were removed from
analysis, which left only 795 true nonrespondents for analysis. This limitation did not affect the
analysis significantly as we exceeded our original sampling goal of 400 nonrespondents.

8

3.3 Low response rate for nonrespondent group
The ABS response rate for the nonrespondent group is low. It could be that the nonrespondents
who did not complete the ABS were different from those who did.

3.4 Coding of responses
Every open question in the ABS had an “other” category. The author coded all the responses in
the “other” categories with some help from a research assistant. There was no assessment of
coding reliability.

3.5 ABS Questionnaire Responses
There was some inconsistency between the field-coded responses and the records of the ACS
Internet Test mail-out. The author observed these errors primarily with regard to the mailing
materials recall questions and the query on use of the paper form. For example, some people in
the Not Prominent Choice strategy were field-coded as recalling having seen the Internet
Instruction Card. That treatment group did not receive the card. As another example, some
Internet responders in the Push strategies said they filled out the paper form, or used it as a
guide, but according to the Internet Test records, they never received a paper form. Coding or
recall errors such as these could have been caused by the questions, the codes provided to
interviewers, or interviewer miskeying. These issues are discussed below in the results section.
Respondents and nonrespondents especially might have felt pressure to give socially desirable
answers. For instance, a nonrespondent might not want to acknowledge that he/she did not
complete the ACS or a respondent might report that he/she read the ACS letter that came in the
materials.
The ABS respondents had to retrospectively describe past actions with the ACS materials that
they might not have encoded at the time, or to which they might have paid scant attention.

4. RESULTS
This section organizes the research findings into five topics corresponding to the research
questions outlined in Nichols (2011): 1) motivation for choosing one mode over the other and
reasons for nonresponse; 2) communication of the mode choice; 3) interaction with the online
ACS; 4) attitudes toward responding online; and 5) characteristics of Internet, mail and
nonrespondents.
Most of the analyses are presented by response group by notification strategy, since the results
may vary according to these dimensions. The attitudinal data and demographics are presented
by response group only.

9

4.1 What is the respondent’s motivation for responding to the ACS and what are the
reasons for not responding to the ACS?
The first topic of analysis examines the reasons for the ACS respondent or nonrespondent
behavior - response by paper, response by Internet, or nonresponse. The ABS attempted to
answer whether mode choice was a reason for nonresponse. The ABS also attempted to
determine if there are motivators that the Census Bureau can capitalize on to encourage a
response over the Internet. To answer this question, this report seeks to identify motivating
messages in the mailing materials. This section first reports on ACS respondents and then
nonrespondents.

4.1.1 What are the reasons for choosing the Internet survey to answer the ACS?
The ABS contained two questions to determine why a respondent used the Internet as his/her
reporting vehicle. First, the ABS interviewer asked the respondent if he/she knew there was a
paper form, “We received your completed survey over the Internet. As far as you can recall,
could you have chosen to respond using a paper questionnaire?” If the respondent knew that
he/she could have completed a paper questionnaire, the ABS asked an open-ended question,
“Why did you decide to use the Internet to complete the survey?”
Table 1 displays the percent of the 457 Internet respondents who did not know there was a
paper form, broken down by notification strategy. For these respondents, they used the
Internet because they did not know they had a choice.
Table 1. Percent of 457 ACS Internet respondents who did not know they could have chosen to report by paper by
notification strategy (standard error in parentheses)
Notification Strategies
Push
Prominent Choice
Not Prominent Choice
Push Regular
Accelerated
(n=105)
(n=115)
(n=119)
(n=118)
Did not know could
11.2%
9.7%
26.2%
22.1%
choose paper
(3.2)
(3.0)
(4.1)
(3.8)
Source: U.S. Census Bureau, Attitudes and Behavior Study, 2011 ACS Internet Test Follow up

There is a difference in the percent of respondents who knew about the paper form in the
different notification strategies. Combining both Push strategies, approximately 24.1 percent
(s.e. =2.8) of the Internet respondents in the Push strategies reported that they did not know
they could have chosen to report using a paper form, which is significantly higher than the 10.5
percent (s.e. = 2.1) of respondents in the Choice strategies who reported that they did not know
they could complete a paper form using a Wald Chi-Square Test (F=15.3, p<.0001). The finding
that proportionally more Push respondents did not know about the paper form compared to the
Choice respondents is not surprising, since the initial and replacement mailing for the two
Choice strategies contained a paper form, whereas only the replacement mailing of the two
Push strategies contained the paper form. Over half of the Internet respondents in the Push
strategies in the ABS never received the replacement mailing. Unless Push strategy respondents
read the letter, they would not know about the availability of a paper form.

10

For the Internet respondents who reported knowing about the availability of the paper form,
Table 2 provides the reasons these respondents decided to report via the Internet. The percent
of Internet respondents in each notification strategy who reported the reason is presented. The
percentages in the columns do not sum to 100 because only reasons provided by five percent or
more of the respondents in a notification strategy are presented in the table, and the ABS
respondents could have reported more than one reason. For example, a respondent could
report that she used the Internet because it was easy and faster.
Table 2. Percent of ACS Internet respondents who reported these reasons for deciding to complete the ACS on the
Internet by notification strategy (standard error in parentheses)
Notification Strategies
Reason for deciding to
Push
use the Internet to
Prominent Choice
Not Prominent Choice
Push Regular
Accelerated
complete the ACS
(n=93)
(n=104)
(n=88)
(n=92)
Easy
60.1%
69.1%
41.9%
46.8%
(5.4)
(4.8)
(5.3)
(5.2)
Faster
29.4
33.7
21.6
25.8
(5.0)
(5.0)
(4.4)
(4.6)
Convenient
30.8
18.3
23.8
20.5
(5.2)
(4.0)
(4.6)
(4.2)
Prefer to do everything
6.3
12.6
13.8
11.9
on the Internet
(2.6)
(3.4)
(3.7)
(3.4)
No paper form
0
0
4.6
9.7
(2.3)
(3.1)
Don’t know or Item NR*
2.8
2.6
7.9
7.7
(1.7)
(1.8)
(2.9)
(2.8)
Source: U.S. Census Bureau, Attitudes and Behavior Study, 2011 ACS Internet Test Follow up
* Respondent could not report don’t know and another reason.

Proportionally more Internet respondents in the Choice strategies (averaged 63.3% with a
standard error (s.e.) of 3.9) reported using the Internet because it was easy compared to the
44.4 percent (s.e., 3.7) of respondents in the Push strategies (Wald Chi-Square Test: F=12.4,
p=.0005). Other than this difference, the other main reasons reported for choosing the Internet
are similar across the notification strategies. The other main reasons are that the Internet is
faster and convenient.
About 6 to 14 percent of Internet respondents across notification strategies reported that they
prefer to do everything on the Internet. And about 5 to 10 percent of Push strategy
respondents reported that there was no paper form and that was a reason why they chose the
Internet.
Most Internet respondents could articulate a reason for choosing the Internet. The “do not
know” rate was approximately 8 percent or lower across the notification strategies.
There were other responses given, but not by more than 5 percent of the respondents in any of
the notification strategies. Some of those reasons included that they did not have to mail the
form, it was a new experience, they did not have a return envelope, the Internet was secure,
and that it saved money. In the two Push strategies, three respondents reported that they
thought they had to use the Internet, and two said the Internet option came first.

11

4.1.2 What are the reasons for choosing the paper form to answer the ACS?
Parallel to the ABS for Internet respondents, the ABS interviewer asked mail respondents two
questions about why they chose to report using the ACS paper form. First, the ABS interviewer
asked the respondent if he/she knew there was an Internet survey: “We received a completed
form in the mail from you. As far as you can recall, could you have chosen to respond over the
Internet?” If the respondent knew he/she could have completed the survey over the Internet,
the ABS interviewer asked an open-ended question, “Why did you decide to use the paper form
to complete the survey instead of using the Internet?”
Table 3 displays the percent of the 335 mail respondents who did not know there was an
Internet option, broken down by notification strategy.
Table 3. Percent of 335 ACS mail respondents who did not know they could have chosen to report via the Internet by
1
notification strategy (standard error in parentheses)
Notification Strategies
Push
Prominent Choice
Not Prominent Choice
Accelerated
(n=119)
(n=104)
(n=112)
Did not know could
37.0%
47.1%
38.4%
2
choose Internet
(4.4)
(4.9)
(4.6)
Source: U.S. Census Bureau, Attitudes and Behavior Study, 2011 ACS Internet Test Follow up
1
No data are present for the Push Regular notification strategy since only one mail respondent completed the ABS.
2
There were eight respondents whose response to the open-ended question indicated that they did not know about
the Internet, which differed from their answer to the earlier question about whether they knew about the Internet.
We included these eight respondents in the “Did not know could choose Internet” category.

About 37 to 47 percent mail reporters used the paper form because they did not know about
the Internet survey. The lack of knowledge of the mode choice was constant across notification
strategy. Approximately half of the Not Prominent Choice mail reporters said they did not know
about the Internet reporting option, which is not surprising since notification of that mode was
buried in text on the cover of the questionnaire. On the other hand, around 37 percent of
Prominent Choice and Push Accelerated mail reporters said they did not know about the
Internet option. It is surprising that so many mail respondents in the Push strategy reported not
knowing about the Internet option since no paper form was sent in the initial mailing package. It
is possible that those mail reporters never opened the initial mailing envelope.
For the mail respondents who reported knowing about the availability of the Internet reporting
option, Table 4 presents the reasons reported for choosing the paper form instead of the
Internet. Again, respondents could provide more than one reason for using the paper form and
only reasons mentioned by five percent or more of respondents are listed; thus, percentages in
the columns may not sum to 100 percent.

12

Table 4. Percent of ACS mail respondents who reported these reasons for deciding to complete the ACS using the
1
paper form instead of via Internet by notification strategy (standard error in parentheses)
Notification Strategies
Reasons for completing
Push
the ACS on the paper
Prominent Choice
Not Prominent Choice
Accelerated
form
(n=75)
(n=55)
(n=69)
Prefer paper/
49.3%
50.9%
26.1%
2
convenient
(5.8)
(6.8)
(5.3)
3
Computer issues
6.7
7.2
18.8
(2.9)
(3.5)
(4.7)
No access to the Internet 21.3
12.7
34.8
(4.8)
(4.5)
(5.8)
Inexperienced computer
6.7
9.1
7.2
user
(2.9)
(3.9)
(3.1)
Worried about Internet
4.0
1.8
7.2
security
(2.3)
(1.8)
(3.1)
Don’t know or Item NR
9.3
18.2
4.3
(3.4)
(5.2)
(2.5)
Source: U.S. Census Bureau, Attitudes and Behavior Study, 2011 ACS Internet Test Follow up
1
No data are present for the Push Regular notification strategy since only one mail respondent completed the ABS.
2
Includes “paper was faster,” “paper was easier,” “paper was available in the mailing package,” “can start and stop on
the paper form easily,” and ”the household has multiple respondents, so paper is convenient.” The Internet ABS
contained separate response fields for some of these reasons. Not making the precoded response fields parallel
between the mail ABS and the Internet ABS was an oversight in the design.
3
Includes “problems with their computer,” “problems with the ACS Website,” and responses that suggested the
respondent was tired of computer tasks (i.e., computer fatigue).

The reasons respondents chose to report by paper differs by notification strategy. In the Push
Accelerated strategy, about 35 percent of the respondents chose paper because they said they
did not have Internet access; about 26 percent reported choosing paper because it was
convenient and about 19 percent reported “computer issues.” These reports suggest that
approximately half3 of the respondents in the Push Accelerated strategy, who knew about the
Internet offer, used the paper form because they could not report via the Internet.
For the Choice strategies, the most frequently reported reason for choosing paper was that the
paper option was convenient, followed by reported lack of Internet access. In fact, combining
the two Choice strategies, 50 percent (s.e., 4.1) of the respondents chose the paper form
because they preferred it or it was convenient compared with the Internet option. That is
significantly higher than the 26 percent (s.e., 5.3) of respondents in the Push Accelerated panel
who mentioned paper preference as the reason for their choice (Wald Chi-Square Test: F=11.9,
p=.0007).
Less than 10 percent (s.e., 3.9) of the respondents in any of the notification strategies claimed
that “computer inexperience” was a reason for choosing the paper form. Less than 8 percent
(s.e., 3.1) of the respondents mentioned Internet security as a reason for choosing a paper form.
Approximately 18 percent of the respondents in the Not Prominent Choice treatment did not
give a reason for choosing paper over the Internet.

3

53.6%=18.8% (computer issues)+34.8% (did not have access)

13

4.1.3 What are the reasons for not responding to the ACS?
To elicit information related to the second goal of the project, the ABS asked a series of
questions to determine why some households never responded to the ACS. We never explicitly
asked, “Why didn’t you respond to the ACS?”4 Instead, the questions were designed to
determine where in the response process nonrespondents stopped. For example, did everyone
open the ACS envelope or did some not even attempt to open it? Did some consider answering
over the Internet and try to log on, but encounter difficulty? Were some still working on the
paper form at the time of the ABS interview? Finally, was there “mode paralysis?” For those
who were aware of both mode options, did some have so much difficulty choosing between the
two that they chose neither? Or did some put aside the ACS form and login information, only to
forget about it? See Appendix G for the question text.
Table 5 summarizes the reasons why households did not respond to the ACS. The table provides
the percent of the ACS nonrespondents in each notification strategy who reported the reason.
Standard errors are in parentheses. The columns in the table do not necessarily add to 100
percent because only reasons provided by five percent or more of nonrespondents are listed in
the table and multiple responses were allowed for some questions. Reasons denoted with an
asterisk (*) are those where multiple responses were not allowed. For example, if the
respondent reported not receiving the envelope, the interview concluded.

4

In future research, I recommend asking such a question at the end of the series to obtain the overall
reason for nonresponse.

14

Table 5. Percent of nonrespondents who reported these reasons for not responding to the April ACS by notification
strategy (standard error in parentheses)
Notification Strategies
Push
Reasons why did not
Prominent Choice
Not Prominent Choice
Push Regular
Accelerated
respond to the ACS
(n=180)
(n=196)
(n=191)
(n=228)
Did not get the
27.2%
31.6%
39.3%
36.8%
envelope*
(3.3)
(3.3)
(3.5)
(3.2)
Received envelope but
20.6
14.8
16.8
20.6
did not open it*
(3.0)
(2.5)
(2.7)
(2.7)
Opened envelope, but
7.8
10.2
5.8
4.8
does not remember if
(2.0)
(2.2)
(1.7)
(1.4)
he/she could report by
paper or by the Internet*
Too busy
11.1
8.7
10.5
10.5
(2.3)
(2.0)
(2.2)
(2.0)
Mailed paper form, but
8.9
9.7
4.7
6.6
no record was found at
(2.1)
(2.1)
(1.5)
(1.6)
the Census Bureau*
Need data from another
6.1
6.1
4.2
4.8
person
(1.8)
(1.7)
(1.5)
(1.4)
Questions were too
6.1
5.6
4.2
1.3
sensitive
(1.8)
(1.6)
(1.5)
(0.8)
1
Computer issues
4.4
2.6
3.7
5.7
(1.5)
(1.1)
(1.4)
(1.5)
2
Don’t know or Item NR * 3.9
2.0
4.2
3.9
(1.4)
(1.0)
(1.5)
(1.3)
Source: U.S. Census Bureau, Attitudes and Behavior Study, 2011 ACS Internet Test Follow up
1”
Computer issues” include “inexperienced computer user,” “computer is difficult,” “no interest in computers,” “no
computer access,” “lost login, User ID/login problems,” and “rather use paper.”
2
Includes “don’t know” to any of the questions in this series.
*Multiple answers were not allowed for these categories.

The two primary reasons given for nonresponse involved receipt of the envelope containing ACS
materials: many respondents claimed they that they either did not receive the envelope or that
they received the envelope but did not open it. This finding held regardless of notification
strategy. Over 46 percent (s.e., 3.6) of all ACS nonrespondents in this survey provided one of
those two reasons. Recall that respondents were those who reported that they handle the mail
in the household. 5
More of the Push strategy nonresponders said they did not receive the envelope compared to
those in the Choice strategies. Combining both Push strategies, 37.9 percent (s.e., 2.4) reported
that they did not receive the envelope, which is significantly higher than the 29.5 percent (s.e.,
2.4) in the Choice strategies who reported that they did not receive the envelope (Wald ChiSquare Test: F=6.3, p=.01). All of the ACS nonrespondent cases discussed in Table 5 should have
received all of the mailings, including both initial and the replacement mailing packages.
The thickness of the initial mailed packet was the only difference between the Push and Choice
envelopes. The Push strategies’ initial mailing envelope was thinner than the Choice strategies’
initial mailing, since it contained the ACS questionnaire in addition to a letter. All replacement
5

It could be that even though we screened for mail handlers, some of the ABS respondents were not the
household members who saw the mailings. This possibility should have affected all notification strategies
equally.

15

mailing envelopes were of the same thickness. It could be that the thinner envelope for the
Push strategies was not as noticeable as the thicker initial mailing for the Choice strategies, but
it is interesting that the replacement mailing package also was not memorable.
Although not shown in Table 5, the ABS asked for the reasons nonrespondents never opened
the envelope. Across the notification strategies, there were 145 nonrespondents who said they
received the envelope but did not open it. The reasons they provided for not opening the
envelope did not differ across strategies.
- “Too busy” was the answer recorded for 39.8 percent (s.e., 4.2) of those respondents.
- “Someone else in the household opened the envelope” was recorded for 10.5 percent
(s.e., 2.6) of those respondents.
- Other answers included, “haven’t gotten to it” (9.6 percent, s.e., 2.5) and “already
completed the Census” (9.1 percent, s.e., 2.4).
- Less than six percent of 145 nonrespondents said that they “forgot about it;” “lost
it;”“threw it out;” “were not familiar with the Census Bureau;” or that “the mailing did
not have their name on it (was not personalized);” “looked like junk mail;” or that it
“looked too big or looked like too much work.”
Table 5 shows that 11 percent or less of nonrespondents across the notification strategies said
that they opened the envelope but could not remember how they could respond to the survey
(i.e., whether by a paper form or by the Internet, or by another way).
Around 8 to 11 percent of ACS nonrespondents in the study said that they remembered they
could complete the survey by either the paper form or by the Internet, but they were too busy
to do so. Some respondents elaborated more on why they were busy. Approximately 20
percent of that group said they were busy due to the health issues of either themselves or
someone in their family (not shown in the table).
Approximately 4 to 10 percent said they actually did mail in a paper form. Upon review, there
was no record of a form being received by the Census Bureau for these cases. Some
respondents could have truly thought they mailed the ACS form back. Others might have said
they mailed the form back to provide a socially desirable answer. Still, others could have been
thinking of their census form which was mailed the year earlier. (As described earlier, some
respondents said they did not open the envelope because they already responded to the
decennial census.)
All told, about 75 percent of ACS nonrespondents in the study provided one of these five
reasons for not responding to the ACS: “did not receive the envelope,” “did not open the
envelope,” “did not know about the mode choice,” “was too busy,” or “did complete the ACS.”
With the exception of the mailing package thickness and the lack of prominence of the mode
choice in one treatment group, none of the reasons pertained to the mode choice offered within
the notification strategies.
Three other reasons were listed by more than 5 percent of the ACS nonrespondents in the
study. Two of them, needing data from another person and the sensitivity of the ACS questions,
relate to the design and content of the ACS. The ACS requires person-level information for
everyone in the household. It is possible, especially in households containing unrelated
individuals, that the original respondent did not know this information and would have had to

16

obtain it from another individual. Some respondents reported that some of the ACS questions,
for example income, were sensitive. Neither data sensitivity nor needing data from another
person were related to mode choice or the particulars of the experimental treatment mailings.
The last reason listed in Table 5 is “computer issues.” “Computer issues” include “inexperienced
computer user,” “computer is difficult,” “no interest in computers,” “no computer access,” “lost
login, User ID/login problems,” and “rather use paper.” It could be that, when presented with
the task of completing the questionnaire online or on paper, some potential respondents
focused on the online form, realized they either did not want to complete the form online or
could not do so, and then failed to complete the survey with the paper form. Table 5 shows that
of all ACS nonrespondents, approximately 6 percent or less fell into this category. In total, 33
nonrespondents cited computer issues as a reason for not completing the Internet form. In 13
of the 33 cases, the nonrespondent reported not knowing about the paper form. Eight of the 13
were cases in one of the Push strategies. Thus, there appeared to be a small number of people
for whom offering an Internet option and not emphasizing the paper form led to a nonresponse.
There were other reasons provided by the ACS nonrespondents, but never by more than 5
percent of them and so not reported in Table 5. Some of the other reasons for nonresponse
were:
- that the survey might have been a scam,
- the survey was too long,
- the survey was redundant or that they had already complied (perhaps thinking of
their Census form),
- they never got the form (three nonrespondents in the Push strategies only),
- they were still working on the form,
- they forgot about the survey,
- they said they mailed back a form, but actually they mailed back a blank form,
- they prefer the telephone,
- they said they submitted the Internet form but no record or only a blank survey was
found at the Census Bureau,
- there were language issues,
- they were confused,
- they were worried about ID theft or Internet security, or
- they lost their PIN.
From these data we do not find explicit indications that “mode paralysis” was a reason for
nonresponse. Difficulty making a choice was never explicitly mentioned by nonrespondents.
There was a response choice of “haven’t gotten to it” and “forgot” listed as precoded responses
in the ABS, but those choices were not frequently reported by nonrespondents and they are not
listed in Table 5. In fact, of the nine nonrespondents who reported forgetting about the survey,
only four reported knowing about the mode choice. Finally, in another series of questions (not
reported in the table) the interviewer asked nonrespondents whether they considered
completing the survey online or considered completing the paper form. Only 3.5 percent (s.e.,
0.7) of 795 nonrespondents over all the notification strategies reported that they considered
completing the survey by both modes. That would be an upper bound on the percent where
“mode paralysis” might have occurred. Based on these data, if “mode paralysis” did affect some
potential respondents, it appears to have affected very few of them.

17

4.2 Was the mode choice message effectively communicated in the different
notification strategies?
The second topic of analysis examines the mailing materials used in the four notification
strategies. The mode choice was communicated on different pieces of material and with
different emphasis in the four notification strategies.
The ABS attempted to determine whether the Census Bureau placed the mode choice messages
on the most memorable pieces of material and secondly, whether respondents (and
nonrespondents) saw the mode choice message.

4.2.1 Did respondents recall the mailing materials containing the mode choice message?
ABS respondents were asked an open question about the materials they remembered seeing in
the ACS envelope, “What materials do you remember seeing in the envelope?” The purpose of
asking this question was to see if the respondents recalled the mailing pieces which had the
mode choice information.
Obviously, the ABS respondents had to remember the envelope prior to telling the interviewer
what materials they remembered were contained in it. Some 87 percent (s.e., 1.9) of ACS mail
respondents and 80 percent (s.e., 2.0) of Internet respondents did remember the envelope.
However, as Table 5 previously showed, about 27 to 39 percent of ACS nonrespondents did not
remember the envelope. Only ABS respondents who remembered the envelope are included
the following analysis.
This analysis examines the free recall of each mailing piece, providing a conservative estimate of
how memorable each piece was. Interviewers field-coded respondents’ answers in these
categories: Form or Questionnaire, Letter, Questionnaire Guide, Frequently Asked Questions
(FAQ) Brochure, Return Envelope, Postcard6, Internet Instruction Card7 and Other (with a writein field)8. The interviewer was instructed to code all of the items mentioned.
Table 6 displays the percent of respondents who remembered the envelope and who
spontaneously recalled each mailing material piece. Results are summarized by response group
and notification strategy combination. Standard errors are in parentheses. Only materials that
were consistently reported by 10 percent or more of the respondents are listed in the table.
6

There was no postcard in the initial or replacement mailing envelopes; rather, the postcard was sent
between the initial and replacement mailings to encourage an ACS response and one was sent to
nonrespondents at the end of the ACS mailing cycle. The analysis presented below suggests that some
interviewers or respondents may have confused the postcard with the Internet instruction card during the
interviews.
7
None of the Not Prominent Choice mailings included an Internet Instruction Card.
8
Figure 1 in Section 2.1 shows the mailing materials sent in each package and whether the mode choice
message was present on the material. In the initial mailing for the two Push strategies, no form,
questionnaire guide, or return envelope were enclosed. Those items were only included in the
replacement mailing that was sent to initial nonresponders. Thus, early Internet reporters in the two
Push strategies should never have received a paper form.

18

Respondents could report multiple mailing pieces; thus, percentages within response groups
and notification strategies do not necessarily add to 100 percent.
Table 6. Percent of ABS respondents who opened the envelope and who recalled each piece of mailing materials by
response group (Internet, Mail and Nonrespondents (NR)) and notification strategy (standard error in parentheses)
Notification Strategies
Push
Prominent Choice
Not Prominent Choice
Push Regular
Accelerated
Internet n=88
Internet=97
Internet=91
Internet=90
Recalled
Response
Mail n=101
Mail n=91
Mail=0
Mail=98
Materials
groups
NR n=87
NR=101
NR=76
NR=88
Form
Internet
79.0%(4.6)
79.8 (4.5)
34.9 (5.0)
47.4 (5.3)

Letter

Internet
Instruction
card

Return
envelope

Questionnaire Guide

Don’t know

Mail

89.1 (3.1)

87.9 (3.4)

No sample

80.6 (4.0)

NR
Internet

67.8 (5.0)
33.3 (5.4)

85.1 (3.6)
34.9 (5.2)

65.8 (5.5)
41.0 (5.2)

70.5 (4.9)
35.7 (5.1)

Mail

19.8 (4.0)

17.6 (4.0)

No sample

18.4 (3.9)

NR
Internet

14.9 (3.8)
26.1 (4.9)

18.8 (3.9)
11.1 (3.4)

6.6 (2.9)
38.5 (5.1)

12.5 (3.5)
33.5 (5.0)

Mail

1.0 (1.0)

3.3 (1.9)

No sample

3.1 (1.7)

NR
Internet

8.0 (2.9)
17.4 (4.3)

3.0 (1.7)
13.2 (3.8)

5.3 (2.6)
9.9 (3.2)

9.1 (3.1)
4.4 (2.1)

Mail

21.8 (4.1)

17.6 (4.0)

No sample

14.3 (3.6)

NR
Internet

6.9 (2.7)
17.4 (4.3)

13.9 (3.5)
16.6 (4.0)

9.2 (3.3)
5.5 (2.4)

14.8 (3.8)
8.8 (3.0)

Mail

19.8 (4.0)

13.2 (3.6)

No sample

16.3 (3.8)

NR
Internet

8.0 (2.9)
6.5 (2.7)

15.8 (3.7)
0.7 (0.7)

6.6 (2.9)
8.7 (3.0)

12.5 (3.5)
4.4 (2.2)

Mail

1.0 (1.0)

0

No sample

6.1 (2.4)

NR

21.8 (4.5)

11.9 (3.2)

21.1 (4.7)

21.6 (4.4)

Source: U.S. Census Bureau, Attitudes and Behavior Study, 2011 ACS Internet Test Follow up

In the Choice and Push notification strategies, the questionnaire or form was the item recalled
by the most respondents, followed by the letter. The Internet Instruction card was recalled by
many Internet respondents. The questionnaire, the Internet Instruction card (available in three
panels), and the letter (for three panels) contained the mode reporting choice message.
For the Internet response groups, the form was spontaneously recalled by proportionally more
respondents in the Choice strategies than in the Push strategies (Wald Chi-Square Test: F=57.4,
p<.0001). The lower percent of Internet respondents who recalled the form in the Push
strategies may have resulted from the fact that anyone who responded early via the Internet
was never mailed the replacement package. For the Push strategy respondents who actually
received the replacement mailing containing the paper form, the percent who recalled it was

19

61.9 percent (s.e., 10.9) for Push Regular and 72.6 percent (s.e., 7.9) for Push Accelerated
strategies.
The letter was one of the primary means of communicating the mode choice for three of the
strategies. The letter seemed to be particularly salient for Internet respondents, across all
strategies. In fact, the percent of all Internet respondents (across the four treatments) who
spontaneously recalled the letter was 36.4 percent (s.e., 2.7). That is significantly higher than
the overall percent of all mail and nonrespondents, 15.7 percent of whom recalled the letter
(s.e., 1.4) (Wald Chi-Square Test: F= 45.1, p<.0001.)
After respondents were asked to spontaneously recall the materials, interviewers explicitly
asked respondents whether they had seen the letter, and, for those who saw it, if they read it.
When the question was explicitly asked, over 80 percent of both mail and Internet response
groups reported that they saw the letter and over 70 percent of both groups reported that they
read it. Approximately 63 percent of the ACS nonrespondents reported seeing the letter and 50
percent reported reading the letter when explicitly asked.
The questionnaire guide and the return envelope were also salient mailing pieces for many
respondents. In the Push strategies, the return envelope and the questionnaire guide were not
present in the initial mailing; this explains why fewer Internet respondents mentioned those
materials. These materials also did not contain the mode choice message in any of the
notification strategies.
The FAQ brochure and the postcard were not listed in Table 6 because they were not recalled
frequently by respondents. The FAQ brochure was recalled by less than 10 percent of
respondents regardless of notification strategy and response group, with one exception.
Nearly 12 percent (s.e., 3.7) of Internet respondents in the Prominent Choice strategy recalled
the FAQ brochure. Fewer than ten percent of respondents recalled the postcard, except in the
Push strategies for the Internet response group. For those two groups, the percent of
respondents who recalled the postcard exceeded ten percent. This might reflect interviewer
error. The respondents may have been describing the Internet Instruction card and the
interviewer selected the postcard by mistake. There are other instances of measurement error
in Table 6. For example, no one in the Not Prominent Choice strategy received an Internet
instruction card, yet around 3 percent of mail respondents and 11 percent Internet respondents
were coded as recalling one.

4.2.2 What messages in the letter were salient to respondents?
In three of the four strategies, the letter communicated the mode choice. The text for the
paragraphs containing the mode message was different among the four strategies. The ABS
asked an open question to determine what messages in the ACS letter were salient to the
respondent or nonrespondent who reported reading the letter: “What do you remember about
what the letter said?” The objective of this question was to determine whether the mode
choice message was memorable, and if so, if it was recalled at different rates in different
strategies.

20

Besides the mode choice message, the letter in the initial mailing for all strategies included
other topics, such as informing respondents that the survey was required by law, the
confidential nature of the survey, how the respondent was selected, and uses of the data. A
phone number was listed if the respondent needed help and the letter was signed by the U.S.
Census Bureau director. The text varied slightly in the replacement letter, but all the topics
were covered again and were identical across notification strategies. Additionally, the
replacement mailing letter emphasized the importance of the ACS and asked for the
respondents help in completing it. The replacement letter also said that the household may be
contacted in person or by phone if no response is received. See Appendix B and C for the
letters.
Table 7 contains a summary of the three messages recalled the most often for each response
group and notification strategy combination. The percent of respondents who said they read
the letter and who recalled the message for each response group and notification strategy is
listed in the row for each message. Standard errors are in parentheses. Percentages within
response groups and notification strategies do not necessarily add to 100 since respondents
could report multiple messages and only the top three messages reported by the most ABS
respondents are listed.
Table 7. Percent of ABS respondents who read the letter and who recalled these messages in the letter by response
group (Internet, Mail and Nonrespondents (NR)) and notification strategy (standard error in parentheses)
Notification Strategies
Prominent Choice
Not Prominent Choice
Push Regular
Push Accelerated
Internet=79
Internet=81
Internet=86
Internet=97
Response
Mail=90
Mail =72
Mail=0
Mail=69
Message
groups
NR=41
NR=52
NR=39
NR=43
Internet
33.6% (5.6)
39.0 (5.7)
30.2 (5.0)
46.0 (5.1)
Mandatory

Mode
Choice

Uses of
data

Don’t know

Mail

41.1 (5.2)

33.3 (5.6)

No sample

34.8 (5.8)

NR
Internet

51.2 (7.9)
27.1 (5.4)

42.3 (6.9)
25.7 (5.3)

28.2 (7.3)
16.2 (4.0)

18.6 (6.0)
19.6 (4.1)

Mail

3.3 (1.9)

2.8 (2.0)

No sample

21.7 (5.0)

NR
Internet

7.3 (4.1)
16.4 (4.2)

5.8 (3.3)
18.2 (4.4)

17.9 (6.2)
8.2 (3.0)

9.3 (4.5)
10.3 (3.1)

Mail

17.8 (4.1)

15.3 (4.3)

No sample

18.8 (4.7)

NR
Internet

24.4 (6.8)
18.9 (4.7)

7.7 (3.7)
12.0 (3.8)

10.3 (4.9)
15.0 (3.9)

14.0 (5.3)
13.5 (3.5)

Mail

31.1 (4.9)

26.4 (5.2)

No sample

24.6 (5.2)

NR
14.6 (5.6)
32.7 (6.6)
28.2 (7.3)
Source: U.S. Census Bureau, Attitudes and Behavior Study, 2011 ACS Internet Test Follow up

27.9 (6.9)

The message remembered by the most ABS respondents, regardless of notification strategy or
response group was the required-by-law/mandatory message. The mandatory message was in
the ACS letter, but it was also on the outside of the initial and replacement mailing envelopes,
and on the postcard, so the respondent could have seen the message several times.
Approximately 30 to 46 percent of Internet respondents recalled that message; about 33 to 41
21

percent of mail respondents and about 19 to 51 percent of ACS nonrespondents across the
notification strategies. Proportionally more nonrespondents in the Choice strategies (46.2%,
s.e., 5.2) recalled the mandatory message compared to the nonrespondents in the Push
strategies (23.4%, s.e., 4.7) (Wald Chi-Square Test: F=10.5, p=.002). There is no ready
explanation for these findings, since all the notification strategies received the same number of
mailings containing the mandatory message.
The mode choice was the next most salient message for many response group/ notification
strategy combinations. It was memorable for many who chose to report via the Internet
regardless of notification strategy. There was no difference between the percent of Choice
Internet respondents who spontaneously recalled that message (26.6%, s.e., 4.0) and the
percent of Push Internet respondents who recalled it (18.0%, s.e., 2.9), (Wald Chi-Square Test:
F=3.1, p=.08). However, the 21.7 percent of the Push Accelerated mail respondents who
recalled the mode choice was significantly higher than the 3.1 percent of Choice mail
respondents who recalled that message (Wald Chi-Square Test: F=12.9, p=.0004).
One of the larger differences in Table 7 is the proportion of Internet and mail respondents in the
Choice strategies who recalled the mode choice message. Proportionally fewer mail
respondents in the Choice strategies (3.1%, s.e., 1.4) recalled the mode choice message
compared to the percent of Internet respondents in those same strategies (26.6%, s.e., 4.0)
(Wald Chi-square Test: F=30.6, p<.0001). The salience of the mode message for mail
respondents in the Choice strategies was not related to whether mail respondents had Internet
access. That is, mail respondents in the Choice strategies who had Internet access were no
more likely to spontaneously recall the mode choice message than the mail respondent who did
not have Internet access (Wald Chi-square Test: F=0.05, p=0.82).
In Table 7, the “data uses” message appears to be a very salient for many ACS respondents and
nonrespondents, even more so than other messages we think might motivate a response, like
data confidentiality. In the second paragraph, the letters list the types of things the ACS data
are used for, like building schools and hospitals. Perhaps one reason this message was
mentioned frequently across all response groups was because an entire paragraph was devoted
to it.
Many other messages were mentioned by respondents, but not by more than 7 percent of
them. These messages are not included in the table. Approximately 7 percent (s.e., 0.9) of
respondents (over all response groups and notification strategies) mentioned that their
household was randomly selected. And 5 percent (s.e., 0.8) mentioned that the letter said the
ACS was important to complete, or to fill it out. Many messages were not mentioned by more
than approximately 4 percent of ABS respondents. Those messages included the phone number
for help, that someone may call or visit, the confidentiality of the data, and an explanation of
the ACS. Only 1.3 percent (s.e., 0.4) mentioned that the letter was signed by the director.
The percent of respondents who reported “don’t know” to this question was high, ranging from
12 to 19 percent for Internet reporters, 25 to 31 percent for mail respondents and 15 to 33
percent for nonrespondents9. There was no discernible pattern for a “don’t know” response by
notification strategy.
9

See Table 7 for the standard errors on the percentages.

22

4.2.3 Did respondents mention the “green” messages in the Push strategies?
The Push strategy’s letters contained some “green” messages, informing respondents that
completing the survey online conserves natural resources, saves taxpayers’ money and allows
for quicker data processing. These messages were not mentioned by many respondents
spontaneously when asked what the letter said.
In the ABS, the interviewer asked Internet respondents, “Was there anything in the mailing
materials that made you decide to use the Internet to complete the survey?” If there was,
respondents were asked to describe what in the mailing materials motivated them. Only 34.3
percent (s.e., 1.0) of Internet responders in the Push strategies said that something in the
materials motivated them to respond by the Internet. The motivator reported the most often
by Push strategy respondents (at around 17%) was the lack of a paper form in the envelope,
followed by “easier” at around 15 percent, and “prefer the Internet” at around 11 percent.
Some respondents in the Push strategies did mention the messages of conserving natural
resources and processing data, but of all the reasons expressed by respondents, those were
mentioned about 11 percent of the time or less.
For the Choice strategies, 36.3 percent (s.e., 1.1) said that something in the materials made
them decide to use the Internet to complete the survey. The two motivators most frequently
reported by Choice Internet respondents were that the form looked too long (23%) and that an
Internet option was offered (20%). The Choice strategies materials did not explicitly contain
any green messages. However, conserving natural resources, saving money and processing data
efficiently were mentioned around 6 percent of the time or less.

4.2.4 Did respondents recall the mode choice message?
The discussion in Section 4.2.2 suggests that the mode choice was a salient message given that
many respondents spontaneously recalled it. In addition to determining if respondents would
spontaneously recall the mode choice message, the ABS explicitly asked respondents whether
they knew they had a choice.
The mode choice knowledge data were collected slightly differently by response group. For mail
and Internet responders the information was collected by two questions: “What do you
remember about what the letter said?” and “We received a completed (survey over the
Internet/form in the mail from you). As far as you can recall, could you have chosen to respond
(using a paper questionnaire/over the Internet)?” The second question was asked only if the
respondent did not mention that he/she knew about the mode choice in the first question. In
the analysis, the answers to those two questions are combined to determine whether the
respondent knew about the mode choice.
For ACS nonrespondents, everyone who said they opened the envelope received the question,
“There were several different ways to respond to the survey. As far as you recall, how could you
have chosen to respond to the survey: by paper? by the Internet? or by another way?”

23

Table 8 shows the percent of ABS respondents who reported that they knew about both the
paper and Internet response options by response group and by notification strategy. In this
table I also present responses received from 113 respondents in the mail control group, who
responded to an abbreviated ABS. As a reminder, the mail control group was not notified on
any of the mailing materials of the Internet reporting option and they could not report via the
Internet to the ACS.
Table 8. Percent of respondents who reported knowing about both Internet and paper reporting modes by response
group and notification strategy (standard errors)
Notification Strategies
Not Prominent
Push
Prominent Choice
Choice
Push Regular
Accelerated
Internet=105
Internet=115
Internet=119
Internet=118
Response Control
Mail=119
Mail =104
Mail=0
Mail=112
groups
Mail=113
NR=180
NR=196
NR=191
NR=228
Internet
No sample
87.6% (3.3)
89.1% (3.1)
68.8% (4.3)
71.9% (4.2)
Mail

42.5% (4.7)

58.0 (4.5)

51.0 (4.9)

No sample

NR
No sample
10.6 (2.3)
5.6 (1.6)
15.2 (2.6)
Source: U.S. Census Bureau, Attitudes and Behavior Study, 2011 ACS Internet Test Follow up

60.7 (4.6)
11.4 (2.1)

Approximately half of mail respondents in all the notification strategies reported that they could
respond using the Internet. However, about 42 percent of the mail respondents in the control
panel also reported that they could respond over the Internet. There was no mention of an
Internet option in any of the control mailing materials. Respondents answering this question
may have assumed that the Census Bureau offered an Internet response option because the ABS
asked about it and because online options are readily available for common tasks such as
shopping, banking and bill payment. Other government tasks such as tax filing can also be done
online. There could have been mail respondents and nonrespondents in the notification
strategies who also reported knowing about an Internet option for these same reasons.
Likewise there could have been Internet respondents or nonrespondents who reported that
there was a paper form, even though they really did not know about it.
With that said, these data should be viewed in conjunction with the spontaneous recall of mode
choice found in Table 7. Some of the findings with these data match the results found with the
spontaneous recall of mode choice. For example, across all notification strategies, it appears
that a larger proportion of Internet respondents reported knowing about the mode choice
compared with the proportion of mail and nonrespondents.
Some of the findings which were not significant with the spontaneous recall of mode choice are
significant with these data. For example, collapsing across the two Choice strategies and the
two Push strategies, these data show that Internet respondents in the Choice strategies were
more likely to report knowing about both modes (i.e., the paper form) than Internet
respondents in the Push strategies (Wald Chi-Square Test: F=11.3, p<.0001). ACS sample in the
Choice treatments received the form in the initial mailing whereas the ACS sample in the Push
treatments did not receive the form in the initial mailing. It makes sense that when the form is
not included, the percent who know about the paper mode is lower.

24

Some of the findings with these data do not match the data in Table 7. For example, in the mail
response group, there were no differences in mode choice knowledge between respondents in
the Choice versus Push Accelerated strategy and there were differences with the spontaneous
recall data. Additionally, in Table 8, nonrespondents were the least likely to know about the
mode choice option, across all notification strategies.

4.3 How did Internet respondents, mail respondents and nonrespondents interact with
the online form?
The next set of analyses look at results from the behavioral data collected in the ABS. The ABS
asked questions about where Internet respondents completed the online form, whether they
used the paper questionnaire as a guide as they completed the online form, and whether they
had difficulty with the online ACS. The ABS also asked mail respondents and nonrespondents if
they considered using the Internet form, if they tried to access the online survey and whether
they had difficulty. This section first examines where Internet reporters completed the survey
and whether they used the paper form as a guide, and then it examines any problems found
with the Internet survey. Finally, the section turns to whether the mail respondents and
nonrespondents in the Push strategies consider answering over the Internet more so than those
in the Choice strategies.

4.3.1 Did Internet respondents complete the ACS at home, work, or some other place?
The ABS asked Internet respondents where they completed the ACS - at home, at work, or at
another place. Table 9 contains the percent and standard error of Internet respondents who
completed the ACS at each of those places.
Table 9. Where Internet responders completed the online ACS (n=453)
Place
Percent who completed
ACS there
(standard error)
Home
87.1 (1.7)
Work
8.6 (1.4)
Library
0.8 (0.4)
Other
3.3 (0.9)
Refuse
0.2 (0.2)
Source: U.S. Census Bureau, Attitudes and Behavior Study, 2011 ACS Internet Test Follow up

As Table 9 shows, the majority of ACS respondents completed the Internet survey from home.
Only 13 percent completed it away from home, either at work, the library or someplace else.
The other places mentioned include other relatives’ homes, places with WiFi access, at school,
and at the hospital.

25

4.3.2 Did Internet respondents use the paper form as a guide and did they fill out the
paper form?
In the ABS, all Internet respondents who reported that they knew there was a paper form were
asked, “Did you use the paper form as a guide while you filled out the survey on the Internet?”
And they were asked if they completed the paper form since it has been suggested that this may
be the case.
Of the 281 Internet respondents who received a paper form (i.e., those were respondents in the
Not Prominent Choice and Prominent Choice strategies and those who received the
replacement mailing package in the two Push strategies), approximately 14.7 percent (s.e., 2.4)
reported using the form as a guide while they were completing the survey online. And 12 of
those 281 respondents (about 3 percent, s.e., 0.9) completed the paper form, with 11 of the 12
completing the paper form before they went online to complete the ACS. There may be some
error associated with these results because nine of the 162 Internet reporters in the two Push
strategies, who did not receive the replacement mailing and hence did not receive a paper form,
reported that they knew about the form and used it as a guide. There is the ability to download
an example of the paper form from the ACS website. Perhaps these respondents did that.

4.3.3 Were there usability problems with the online ACS?
Horwitz, Tancreto, & Zelenak (forthcoming) fully documents user interface problems with the
online ACS. The ABS collected only self-reports of usability problems with the online survey.
In the ABS, only two Internet respondents (out of 455) reported difficulty logging onto the
online survey. One Internet respondent said that the Website would not come up right away
and he/she had to try several times. Another respondent said he/she received an error message
that too many people were trying to log on.
Twenty-two Internet respondents (5.2 percent, s.e. 1.1) said they had difficulty with the online
survey itself. Six of the 22 had issues with the questions and survey items collected. For
example, one respondent reported having difficulty with the electric and gas bill question.
Another respondent said that the questions were not logical for the unemployed. These are not
usability problems with the online application. Another six respondents had website problems,
such as problems finding the correct website, and four said they had a slow Internet connection.
Three respondents said their session timed out. Two people said it took too long and one
person said there was no progress indicator.
These 22 respondents reported using either Internet Explorer or Google Chrome browsers with
DSL, a Cable Modem or a broadband plan at their home to complete the ACS online survey. No
one reported using dial-up, fiber-optic or satellite service.

26

4.3.4 Did mail respondents or nonrespondents in the Push strategies consider answering
the ACS using the Internet survey more so than the mail and nonrespondents in the
Choice strategies?
If the mail respondents and nonrespondents knew about the Internet option, the ABS
interviewer asked, “Did you consider answering over the Internet?” The objective was to
determine which notification strategies encouraged mail respondents and nonrespondents to
consider the Internet mode.
Table 10 displays the percent and standard errors of mail respondents and nonrespondents who
considered using the Internet to report. Results are collapsed across the two Push strategies
(for nonrespondents) and the two Choice strategies (for mail and nonrespondents). ABS
respondents and nonrespondents who reported knowing about the Internet mode choice are
included in this analysis.
Table 10. Percent of ABS mail respondents and nonrespondents who considered answering by the Internet (standard
errors)
Response Group
Choice strategies
Push strategies
Mail

1

37.0%(4.9)
60.9% (7.3)
(n=100)
(n=46)
Nonrespondents
53.3 (7.6)
39.3 (5.8)
(n=45)
(n=72)
Source: U.S. Census Bureau, Attitudes and Behavior Study, 2011 ACS Internet Test Follow up
1
Contains only Push Accelerated responses.

Table 10 shows that a larger proportion of mail respondents in the Push Accelerated strategy
considered answering the ACS over the Internet compared to the proportion of mail
respondents in the Choice strategies (Wald Chi-square Test: F=7.4, p<0.007).
There is a much smaller difference between nonrespondents in the Choice and Push strategies,
and the results go in the opposite direction. More Choice nonrespondents than Push
nonrespondents considered answering the survey over the Internet; however this difference is
not significant (Wald Chi-square Test: F=2.2, p=0.14).

4.3.5 Did mail respondents or nonrespondents try to access the Internet form or start
the Internet form? If they did so, were there usability problems with the form or login
task?
Horwitz et al., (forthcoming) provides a full evaluation of these questions.
Although some mail respondents and nonrespondents considered responding by Internet, very
few of them tried to access the Internet form or start it.
Of 65 mail respondents who reported that they considered answering the ACS over the Internet,
only 14 (~20%) reported logging into the online survey and 12 of those were from the Push
Accelerated strategy. Six of the 14 had difficulty with the login and only one of the 14 actually
started the survey.
27

The six mail respondents who had difficulty with the login task had either DSL or Cable service at
home. Two of the six did not know what website to get on. One reported that he/she could not
bring up the questionnaire on the Web and two said that the system did not recognize the user
ID. Still another mail respondent said the website would not let him/her go further.
Of the 113 nonrespondents who considered answering the ACS over the Internet, only 13
(~12%) reported that they logged into the online form. Census Bureau records showed that
only five of the 13 actually got to the online survey. Of the eight who do not have a record of a
successful login, five said that they had difficulty with the login procedure. Six of the 13 claimed
they started the survey and three of them said they finished, but there was no record of a data
submission. For two of the three who said they finished, there was no record of them ever
accessing the online survey.
Of the 13 nonrespondents who reported logging in, four were from the Prominent Choice and
eight were from the two Push strategies. Only one nonrespondent from the Not Prominent
Choice strategy reported that he/she logged in.

4.4 What are the attitudes toward completing a form online?
This set of analyses examines results from the attitudinal data collected in the ABS. The ABS
asked nonrespondents, respondents in the mail control panel, and mail respondents in the four
notification strategies who did not know there was an online ACS reporting option, which mode
they preferred for answering the ACS. Additionally, the ABS asked mail and Internet
respondents who knew about both reporting modes whether they would answer the survey
with the other mode if their current mode choice was not available. The section first shows the
preference data and then provides the attitudes towards filling out a survey in another mode.

4.4.1 What is the mode preference for nonrespondents, mail control, and mail Internet
Test respondents who did not realize there was an online survey option?
Three slightly different questions were asked to obtain the mode preference for
nonrespondents, mail control respondents, and mail respondents in the notification strategies
who did not know about the Internet reporting option.
The ABS asked nonrespondents who reported receiving the ACS envelope, “Some people were
offered two options for responding to the survey- filling out the paper form or going to the
Internet and filling out the survey there. Which of those two ways to answer the survey do you
prefer?” The ABS asked nonrespondents who did not remember the ACS envelope but who
reported that they would have completed a questionnaire had they received a request from the
Census Bureau, a slightly different question, “If you had a choice of answering on the Internet or
answering on paper, which would you prefer?”
The ABS asked mail control respondents and Internet Test mail respondents who did not know
about the Internet mode, “The Census Bureau is considering offering the survey over the
Internet, in addition to having the paper form. Which of those two ways to answer the survey do
you prefer?”
28

Table 11 contains the percent of ABS respondents who prefer each of the mode choice
combinations for each of the response groups shown. Results are collapsed across the four
notification strategies for the nonrespondents and the mail respondents in the Internet Test
(Mail IT).
Table 11. Stated mode preference to Census Bureau surveys among selected ABS respondents (standard errors)
Mode preference
Nonrespondents who Nonrespondents who Mail Control
Mail IT responders
remembered the ACS did not remember
responders (n=113)
who did not know
envelope (n=350)
the ACS envelope
there was an online
(n=431)
ACS survey (n=144)
Prefer paper
64.6 %(2.6)
57.3% (2.4)
56.6% (4.7)
Prefer the
20.9 (2.2)
30.2 (2.2)
33.6 (4.5)
Internet
Prefer neither
8.5 (1.5)
6.0 (1.2)
1.8 (2.4)
mode
Prefer both
4.2 (1.1)
4.2 (1.0)
6.2 (2.3)
modes about the
same
It depends
0.9 (0.5)
0.7 (0.4)
1.8 (1.2)
Don’t
0.9 (0.5)
1.7 (0.6)
0
know/Refused
Source: U.S. Census Bureau, Attitudes and Behavior Study, 2011 ACS Internet Test Follow up

75.7%(3.5)
13.9 (2.8)
2.1 (1.2)
4.2 (1.7)

2.8 (1.4)
1.4 (1.0)

Paper was the preferred mode for completing the survey, with approximately half or more of
the respondents in each of the response groups indicating that as the preferred mode as shown
in Table 11. The percent who prefer paper appears to vary across the response groups shown,
with approximately 76 percent of the mail respondents in the Internet Test who were unaware
of the Internet option preferring paper compared to about 57 percent of mail control
respondents and about 57 percent of nonrespondents who did not remember the ACS envelope.
Likewise, the percent who prefer the Internet appears to vary across the response groups
shown, with about 14 percent of mail respondents in the Internet Test groups (who said they did
not know there was an Internet option) preferring the Internet to about 34 percent of the mail
control group preferring the Internet.

4.4.2 Would respondents have completed the ACS if their first mode choice was not
available?
For mail respondents and Internet respondents who said they knew about the other reporting
mode, the ABS asked, “If the (paper form/Internet) was not an option, would you have
completed (the survey on the Internet/a paper form)?” The objective of this question was to
obtain an indicator of how response rates would be affected if there was only an online survey.
Table 12 displays the percent of respondents who would not complete the survey in a different
mode, broken down by notification strategy by response group.

29

Table 12. Percent of respondents who said they would not complete the ACS in the other mode if the mode they
chose was not available by notification strategy (standard errors)
Not Prominent
Push
Push
Prominent Choice
Choice
Regular
Accelerated
Mail respondents
21.7 %(5.0)
13.2% (4.7)
No sample 31.3 %(5.7)
who would not
(n=69)
(n=53)
(n=67)
complete an Internet
ACS
Internet respondents
9.5 (3.1)
6.9 (2.5)
5.0 (2.0)
1.7 (1.2)
who would not
(n=103)
(n=115)
(n=119)
(n=118)
complete the ACS on
a paper form
Source: U.S. Census Bureau, Attitudes and Behavior Study, 2011 ACS Internet Test Follow up

As shown in Table 12, about 13 to 31 percent of mail respondents in the different notification
strategies reported that they would not complete an online ACS form if the paper form was not
available. Ten percent or less of Internet respondents report that they would not complete an
ACS paper form if the online form was not available.

4.5 What are the characteristics of Internet respondents, mail respondents and
nonrespondents?
The final set of analyses examines Internet penetration and usage and then turns to the
demographics. The survey asked questions about Internet penetration and usage for all
respondents as well as the demographics for the ACS nonrespondents. Demographics for mail
and Internet responders were from their original ACS submissions.

4.5.1 What percent of mail respondents, Internet respondents and nonrespondents use
the Internet, and therefore have Internet access?
In the ABS, interviewers asked if the ABS respondent used the Internet at home, at work, or at
some other place. Table 13 shows the percent of ABS respondents who reported using the
Internet at one or more of those places. The data are displayed by response group by
notification strategy. Approximately 74 percent of the control mail respondents reported using
the Internet. Pew Research Center measured Internet penetration at 78 percent of adults in
May 2011 (http://www.pewinternet.org/Static-Pages/Trend-Data/Whos-Online.aspx). The
control population of mail reporters is in line with their statistic.

30

Table 13. Percent of ABS respondents who reported using the Internet at home, at work, or at some other place by
response group and notification strategy (standard errors)
Notification Strategies
Control
Prominent Choice
Not Prominent
Push Regular
Push
Mail=113
Internet=105
Choice
Internet=119
Accelerated
Response
Mail=119
Internet=115
NR=191
Internet=118
groups
NR=180
Mail =104
Mail=112
NR=196
NR=228
1
Internet
No sample
96.9% (1.6)
100 .0% (0)
97.4% (1.5)
97.5% (1.5)
2

Mail

74.3% (4.1)

68.9 (4.2)

79.8 (3.9)

No sample

58.0 (4.7)

3

NR
No sample
63.9 (3.6)
69.9 (3.3)
64.9 (3.5)
63.6 (3.2)
Source: U.S. Census Bureau, Attitudes and Behavior Study, 2011 ACS Internet Test Follow up
1
By definition, everyone should have had access to the Internet. However, a small number refused a question or were
ABS breakoffs which resulted in less than 100 percent for some notification strategies.
2
Less than one percent were ABS breakoffs, don’t know or refusals. Those are included in the denominator.
3
Approximately three percent were ABS breakoffs, don’t know or refusals. Those are included in the denominator.

By definition, everyone who reported by the Internet should have Internet access. Footnote 1
in Table 13 explains why most but not all respondents in the group reported using the Internet.
Lower percentages of the mail and nonrespondent groups reported using the Internet. On
average, Internet access for nonrespondents was less than 70 percent (s.e., 3.3) across all
notification strategies.
The lowest penetration of Internet access for all response group/notification strategy
combinations was 58 percent in the mail, Push Accelerated strategy. This lower percentage can
be explained by the fact that more ACS respondents in the Push strategies responded by
Internet than in the other notification strategies (Tancreto et al., 2012). The ABS data show that
those in the Push Accelerated strategy who waited for and returned the paper form soon
enough to be selected into the ABS were less likely to have Internet access than others in the
Choice strategies who returned a paper form (Wald Chi-Square Test: F=8.3, p=.004). This
finding suggests that the Push Accelerated strategy encouraged those who had Internet access
to report with that mode. These data also support results found in Table 4 where approximately
11 percent of the mail respondents in the Push Accelerated strategy said they chose mail
because they did not have Internet access.
Not shown in Table 13 is the place where the person had access to the internet. Home was
reported as the primary place where people used the Internet, followed by work, and then some
other place. Across all notification strategies, 96 percent of Internet reporters used the Internet
at home, 64 percent used the Internet at work, and 39.3 percent used the Internet at some
other place. For mail reporters 64 percent used the Internet at home and 33 percent used it at
work, and 20 percent used it at another place. Nonrespondents appear similar to the mail
reporters with 63 percent using the Internet at home, 33 percent at work and 18 percent at
another place.

31

4.5.2 How much, and for what purposes, do the mail respondents, Internet
respondents, and nonrespondents use the Internet?
The ABS sought to determine if completing or not completing the online survey was related to
time spent on the Internet or other tasks performed on the Internet. To do this, the ABS asked
a series of questions about Internet usage. If people used the Internet, the ABS asked how
many hours per day they spent on the Internet and how often they performed particular tasks
associated with the Internet. For these activity questions the response scale was very often,
often, sometimes, rarely, or never.
Table 14 contains the answers to these Internet usage questions. Results are collapsed across
the four notification strategies for each response group: Internet, mail respondents in the
Internet Test (indicated by Mail IT), and nonrespondents. Results from the mail control
response group are shown separately. Only respondents who reported that they used the
Internet are included in this analysis.
Table 14. Internet usage (weighted mean of hours on the Internet per day and median frequency of Internet activities
1
performed) by ABS response group
Response Groups
Internet usage
Internet
Mail-IT
Nonrespondents
Mail- Control
How much
Average # hours
3.6 (0.3)
2.4 (0.2)
2.3 (0.1)
2.7 (0.3)
on Internet/day
(n=438)
(n=226)
(n=505)
(n=82)
(weighted mean)
Activities
accomplished
(median)
Visit news, maps, Often
Sometimes
Often
Often
information on
(n=438)
(n=221)
(n=506)
(n=82)
the Web
Visit social
Sometimes
Never
Rarely
Rarely
networking sites
(n=439)
(n=221)
(n=506)
(n=83)
Use for bill
Often
Sometimes
Sometimes
Often
paying or
(n=439)
(n=221)
(n=505)
(n=83)
financial
transactions
Use for emailing
Often
Often
Often
Often
(n=438)
(n=221)
(n=506)
(n=83)
Other
Rarely
Never
Never
Rarely
communication
(n=438)
(n=221)
(n=506)
(n=83)
beside email
Entertainment,
Sometimes
Rarely
Rarely
Rarely
like music or
(n=436)
(n=221)
(n=506)
(n=83)
videos
Filling out
Rarely
Rarely
Never
Rarely
surveys
(n=436)
(n=221)
(n=508)
(n=83)
Source: U.S. Census Bureau, Attitudes and Behavior Study, 2011 ACS Internet Test Follow up
1
The sample size for each of the response groups differs by question because answers of don’t know or refusals are
removed as well as any item nonresponse. Unweighted median values are shown for the task completion response
data

The results of Internet daily usage by response group are consistent with expectations. That is,
Internet responders reported on average the highest number of hours per day spent on the

32

Internet. The nonrespondents on average spent the same number of hours on the Internet as
people in the control panel, and in the mail Internet Test panel. This finding suggests that
infrequent Internet usage was not a reason for nonresponse among those who had Internet
access.
Although no statistical testing was conducted with these data, the median answer for each
activity showed that Internet respondents generally reported performing the Internet activities
more often than the mail (IT) responders, except for email, which all response groups used
often, and completing surveys, which no one did often. Using the Internet for entertainment,
like music or videos and visiting or using social network sites, were activities where the median
response for Internet respondents was sometimes, while all the other groups reported a median
of rarely or never for those activities.
Based on these data we see that filling out surveys is something people rarely do on the
Internet, with nonresponders answering that they never fill out a survey on the Internet. The
median answer when asked about filling out surveys on the Internet was “rarely” for both
Internet responders and mail responders in the Internet Test. This suggests that ACS Internet
Test mail responders might well respond to some other surveys online, and but that filling out
surveys online in general is a rare event.

4.5.3 What are the demographic characteristics of ABS Internet responders, mail
responders and nonresponders?
The ABS collected demographic information for the nonresponders. Tancreto et al. (2012)
examined whether there are demographic differences for Internet and mail responders.
Because of its small sample, the ABS was not designed to examine those demographic
differences or to compare these data to the larger Internet Test dataset. Instead Table 15
provides a snapshot of the demographics for the three response groups in the ABS. The
demographics for the Internet and mail reporters come from the ACS Internet Test data
collection itself. The demographic questions asked of nonrespondents in the ABS were slightly
different from the ACS questions included in the Internet Test. The ABS demographic questions
used CBAMS as a model for the demographic questions of tenure, race, Hispanic origin, and
education instead of the sometimes longer questions used in the ACS.

33

Table 15. Demographic Characteristics of ABS Internet, Mail-IT, and Nonrespondents – (standard errors)
Response Groups
Characteristic
Internet
Mail -IT
Non-respondents
Age (mean years)
47.7 (0.7)
58.3 (1.0)
56.5 (0.6)
(n=456)
(n=332)
(744)
Female (%)
45.4%(2.4)
58.0% (2.7)
61.0% (1.8)
(n=457)
(n=333)
(n=771)
Race
(n=456)
(n=331)
(n=769)
White (%)
85.5 (1.7)
90.9 (1.6)
78.5 (1.5)
Black
5.3 (1.1)
4.8 (1.2)
10.5 (1.1)
AIAN
0.2 (0.2)
0.9 (0.5)
0.5 (0.3)
Asian
5.7 (1.2)
1.5 (0.7)
1.5 (0.4)
Hawaiian/OPI
0
1.5 (0.7)
0.3 (0.2)
Other
1.5 (0.6)
0.6 (0.4)
4.3 (0.7)
Multirace
1.7 (0.6)
1.2 (0.6)
2.1 (0.5)
Hispanic (%)
5.1 (1.1)
1.2 (1.1)
11.0 (1.1)
(n=455)
(n=313)
(n=770)
Education
(n=457)
(n=320)
(n=770)
Less than HS (%)
2.6 (0.8)
6.9 (1.4)
11.6 (1.2)
HS graduate
12.5 (1.6)
28.4 (2.5)
29.9 (1.7)
More than HS
84.9 (1.7)
64.7 (2.7)
56.2 (1.8)
Hhld income
25K<50K
50K-75K
2
Not available
1
(median)
(n=279)
(n=648)
Renter (%)
18.8 (1.9)
23.8 (2.4)
15.1 (1.3)
(n=441)
(n=324)
(n=768)
Source: U.S. Census Bureau, Attitudes and Behavior Study, 2011 ACS Internet Test Follow up
1
The median for the weighted data was used. Income for nonrespondents was collected by a range question.
2
The household income data for the Internet response group was inconsistent with the Mail response group income
data format. This difference was not resolved prior to the due date of this report.

From these data, the general finding is that nonrespondents were more frequently black and
Hispanic, and had a higher proportion with less than a high school education than either the
mail respondents in the Internet Test or the Internet respondents. Internet respondents in the
ABS were male and with more than a high school education than either mail respondents or
nonrespondents. Mail respondents in the ABS were white more than the other response
groups.

5. SUMMARY
The factors that affected a respondent’s mode choice decision varied across notification
strategies and by the mode ultimately chosen.
Many of the Choice strategy Internet respondents used it because of the perceived advantages
of the Internet: it was easy, convenient, and fast. Many Push strategy respondents used the
Internet for those same reasons. However, about 24 percent of the Push strategy respondents
used the Internet because they did not know of any other way to report, most likely because
there was no paper form in the initial mailing package. Other Push strategy respondents
actually reported that not having the paper form motivated them to use the Internet to report.
The “green” messages in the Push strategies’ letter concerning saving resources and faster

34

processing were not mentioned frequently by Internet respondents as motivating them to use
the Internet.
Not knowing about the Internet reporting option was the reason why about 37 to 47 percent of
mail respondents across the notification strategies reported using the paper form. Those
respondents also overwhelmingly (at about 76 percent) reported preferring the paper form over
the Internet, if it was offered. A notable finding is that many Push strategy mail respondents,
who knew about the Internet option, considered the Internet option and literally could not
report online because they did not have Internet access or because they had computer
problems. This was not the case with the Choice mail respondents. The difference in the
Internet access rates between Push and Choice mail respondents suggests that the Push
Accelerated strategy did push people to use the Internet who might have very well completed
the paper form if it was in the initial mail package. When asked, roughly 31 percent of the Push
Accelerated mail respondents reported that they would not complete the ACS using the
Internet, if that was the only mode offered. This suggests that there is a certain percent of
compliant responders who cannot or would not report to an online ACS.
When asked to recall the mailing materials, respondents’ answers indicate that the current
placement of the mode choice message is reasonable, especially ones placed on the form and
the letter. Those pieces were recalled most frequently by respondents and nonrespondents. It
seems that the letter was important to Internet reporters as they recalled it often. For the Push
strategies there was no form provided initially so the letter was the only thing to read. But,
even in the Not Prominent Choice strategy, many Internet respondents recalled the letter
spontaneously. In that strategy there was a form in the initial mailing and the letter did not
mention the Internet option. This finding suggests that one characteristic of ACS Internet
reporters was that they were letter readers. Another characteristic found in the ABS data is that
Internet respondents use social media sites and online entertainment more often than mail or
nonrespondents.
The ABS data showed that the mode choice was not communicated well enough so that all the
sampled households in the Internet Test knew about it, even when asked explicitly about it.
Internet respondents knew about the mode choice the most frequently, followed by mail
respondents and then nonrespondents. This finding was consistent across notification strategy.
Knowing about both modes becomes critical when respondents cannot complete the survey
online. The ABS found that 13 of the 795 nonrespondents could not complete the Internet
survey and did not know about the availability of the paper form. Had these 13 nonrespondents
known about the paper form, they might have responded.
ABS results do suggest that offering the Internet mode was not a primary reason for
nonresponse. Less than 6 percent of nonrespondents cite computer issues (such as computer
and Internet access) as a reason for not responding to an Internet option. Even fewer
mentioned Internet security, Internet experience and even losing the pin as reasons for
nonresponse. Very few nonrespondents attempted to access the online form and had
difficulty. Instead the salience of the mail package and being busy were the main reasons
reported for nonresponse.

35

Although a paper form was preferred to an Internet survey by respondents who did not know
about the Internet survey or did not have the online survey available, an Internet reporting
mode option is perhaps assumed by many, as 42.5 percent of mail control respondents, who
had no notification of an Internet option, reported knowing that there was one.
The motivator for conducting an ABS was to determine if there was any indication of “mode
paralysis.” The April 2011 Internet Test did not find dampened response rates with the panels
that offered multiple modes (Tancreto et al., 2012). However, it could be that offering an
Internet option encouraged some to respond who might not have otherwise, replacing any
nonresponse due to “mode paralysis.”
The ABS data did not contain explicit evidence for “mode paralysis” in the notification strategies.
None of the respondents commented that they could not decide between the two modes. Of
the very few respondents who said they forgot about the survey, not even half of them reported
knowing about the mode choice. Nonrespondents were the least likely of all ABS response
groups to know about the mode choice, across all notification strategies. Insofar as the detailed
questions employed here can address “mode paralysis,” the ABS data suggest an upper bound
for it as no more than about 4 percent of the nonrespondents who reported considering both
reporting options.

6. FUTURE RESEARCH
These ABS data suggest a few research avenues for both communicating an ACS mode choice
and converting nonrespondents into respondents.
The ABS data support placing the mode choice information on the form, letter, and Internet
instruction card, especially the letter and form. If the form were included in a multi-mode data
collection, research could concentrate on improving the Internet option notification on the
form, since it is the most salient piece of mailing material.
The ABS results lend support to the design that does not place the form in the initial mailing.
The ABS data show that the “no form” design, with a fallback paper option, appears to convert
mail respondents into Internet respondents, without any obvious negative consequences on
those who would have not responded in the first place.
Improving overall ACS response rates should center on testing different mailing envelopes and
postcards, as almost half the nonrespondents reported not knowing about the mailings or not
opening them. Creating a notification strategy that uses large pieces of mailing material that do
not require opening could be an avenue for research. Placing notifications in addition to the
mandatory message on the envelope might be another avenue for investigation.
Finally, more research into the messages on the materials might be useful to help improve
response rates. It seems that a portion of nonrespondents did not understand the distinction
between the decennial census and the ACS. They apparently assumed that they completed the
ACS when they completed the census. In addition to the mandatory message, the “data uses”
message also appears to be salient and perhaps a motivator for response.

36

Acknowledgements
I would like to thank the following people for their valuable contributions and assistance to the
development and analysis of these projects: Mary Davis, Jennifer Tancreto, Brenna Matthews,
Mary Frances Zelenak, Rachel Horowitz, Karen King, Tony Tersine (DSSD); Debbie Griffin, Cynthia
Ramsey (ACSO); Victor Quach, Kathy Ashenfelter, Peter Miller, Terry DeMaio, Jenny Childs
(CSM); Jason Arata, Ed Dyer (TMO); Nancy Bates (DIR) and my critical reviewers.

References
Cannell, C. F., Miller, P. V., & Oksenberg, L. (1981). Research on interviewing techniques.
Sociological Methodology, 12, 389-437.
Gentry, R. & Good, C. (2008). "Offering Respondents a Choice of Survey Mode: Use Patterns of
an Internet Response Option in a Mail Survey.” Paper Presented at the Annual Conference of
the American Association for Public Opinion Research, New Orleans, LA.
Griffin, D., Fischer, D., & Morgan, M. (2001). “Testing an Internet Response Option for the
American Community Survey,” Paper Presented at the Annual Conference of the American
Association for Public Opinion Research, May 17-20.
Grigorian, K. & Hoffer, T. (2008). “2006 Survey of Earned Doctorates Mode Assignment Analysis
Report.” Prepared for the National Science Foundation by the National Opinion Research
Center. Chicago, IL: University of Chicago.
Haist, F., Shimamura, A. P., & Squire, L. R. (1992). On the relationship between recall and
recognition memory. Journal of Experimental Psychology: Learning, Memory, and Cognition, 18,
691-702.
Horwitz, R. (forthcoming). "The April 2011 American Community Survey Internet Test: Content
Reinterview Analysis” U.S. Census Bureau: Decennial Statistical Studies Division.
Horwitz, R., Tancreto, J., & Zelenak, M.F. (forthcoming). "Use of Paradata to Assess the Quality
and Functionality of the ACS Internet Instrument” U.S. Census Bureau: Decennial Statistical
Studies Division.
Horwitz, R. 2011. “The 2011 American Community Survey (ACS) Internet Test Instrument Design
Evaluation.” U.S. Census Bureau: Decennial Statistical Studies Division. January 10, 2011.
Johnson, K. (2009). “Census Barriers, Attitudes, and Motivators Survey Methodology Report,”
C2PO 2010 Census Integrated Communications Research Memoranda Series No.8, January 6,
2009.
MacDougall, R. (1904). Recognition and recall. Journal of Philosophy, Psychology, and Scientific
Methods, I. 229-233.

37

Marquette, E. (undated). “Specifications for Calculation of American Community Survey
and Puerto Rico Community Survey Housing Unit Mode Level Response Rates.” U.S. Census
Bureau: 2009 American Community Survey Sampling Memorandum Series #ACS09-S-25.
Millar, M. & Dillman, D. (2011). Improving response to Web and mixed-mode surveys. Public
Opinion Quarterly, 75 (2), 249-269.
Nichols, E. (2011). "2011 ACS Internet Test: Attitudes and Behavior Study Reinterview Plan.”
U.S. Census Bureau: Center for Survey Measurement. November 23, 2011.
Nichols, B. & Marquis, K. (2001). “Results from the Applied Behavior Analysis of American
Community Survey Nonrespondents,” U.S. Census Bureau: Statistical Research Division,
Human-Computer Interaction Memorandum Series #43. September 19, 2001.
Pew Research Center (2012). Demographics of internet users. Internet and American Life Spring
Project’s Spring Tracking Survey [Data file]. Retrieved from http://www.pewinternet.org/StaticPages/Trend-Data/Whos-Online.aspx
Smyth, J.D., Dillman, D.A., Christian, L.M., & O’Neill, A. (2010). Using the Internet to survey small
towns and communities: Limitations and possibilities in the early 21st century. American
Behavioral Scientist. 53: 1423-1448
Tancreto, J., Zelenak, M. F., Davis, M., Ruiter, M., & Matthews, B. (2012). “2011 American
Community Survey Internet Tests: Results from First Test in April 2011” U.S. Census Bureau:
Decennial Statistical Studies Division. February 21, 2012.
Zelenak, M.F., Ruiter, M., Davis, M., Horwitz, R., & Tancreto J. (2010). “Project Plan for The 2011
American Community Survey (ACS) Internet Test” U.S. Census Bureau: Decennial Statistical
Studies Division. October 21, 2010

38

Appendix A April 2011 ACS Internet Test Questionnaire Cover
Not Prominent Choice: Questionnaire Cover

A-1

Prominent Choice and Push strategies: Questionnaire Cover

A-2

Appendix B April 2011 ACS Internet Test Letter for the Initial Mailing
Package
Not Prominent Choice: Letter in Initial Mailing Package

B-1

Prominent Choice: Letter in Initial Mailing Package

B-2

Push strategies: Letter in Initial Mailing Package

B-3

Appendix C April 2011 ACS Internet Test Replacement Letter
Push Regular: Letter in the Replacement Mailing Package

C-1

Appendix D April 2011 ACS Internet Test Internet Instruction Card
Prominent Choice: Initial Mailing Package Instruction Card (Front Side –
English)

Push strategies: Initial Mailing Package Instruction Card (Front Side –
English)

D-1

Appendix E ABS Questions for Internet Respondents
ABS Questions
Assumptions: Previous questions will provide the introduction of the interviewer,
verification of the telephone number and address. Additionally, we will ask for the
respondent by name and set a callback if not available. Most response options are not read
to respondents. Don’t know and refuse are available response options for each question.
Screener
May I speak to ?
This call is about a Census Bureau survey you recently completed. First we’d like to thank
you for completing the survey online. Today, we’d like to collect some additional
information.
1.

2.

2a.

If you recall, you recently received a large envelope from the U.S. Census Bureau.
The Census Bureau wants to improve its mailing materials. What materials do you
remember seeing in the envelope? (Mark all that apply)
❑ Form or Questionnaire
❑ Letter – (if marked, then go to Q2.a)
❑ Questionnaire Guide
❑ FAQ Brochure
❑ Return Envelope
❑ Postcard
❑ Internet Instruction Card
❑ Other – Record main ideas
❑ Don’t remember the envelope
If "Letter" NOT marked in Q1
Did you see a letter about the survey?
❑ Yes
❑ No - Skip to 3
Did you read the letter [Fill if "letter” marked in Q1: that was in the large
envelope]?
❑ Yes
❑ No - Skip to Q3

E-1

2b.

3.

4a.

4b.

What do you remember about what the letter said?
❑ Could have chosen between the Internet and the paper form
- If marked, skip to Q4a.
❑ Mentioned Internet only
❑ Mentioned paper form only - If marked, skip to Q4a.
❑ Listed uses of the data
❑ Randomly selected
❑ Required by law - mandatory
❑ Answers are kept confidential
❑ It provided phone number for help
❑ Signed by director
❑ Someone may call or visit
❑ Confusing
❑ Other- Record main ideas
We received your completed survey over the Internet. As far as you can recall,
could you have chosen to respond using a paper questionnaire? (Mark all that
apply.)
❑ Yes
❑ No - Skip to Q4b
Why did you decide to use the Internet to complete the survey? (Mark all that
apply)
❑ There was no paper form/ no other way to respond
❑ Easy
❑ I prefer the Internet/Do everything on the Internet (pay bills, shop, etc.)
❑ Secure
❑ Convenient
❑ Faster processing
❑ Saves money
❑ Other - Record main ideas
If the Internet was not an option, would you have completed a paper form?
❑ Yes
❑ No
❑ Maybe

E-2

5.

Was there anything in the mailing materials that made you decide to use the
Internet to complete the survey?
❑ Yes, - Please describe what in the materials motivated you to use the
Internet.(Mark all that apply.)
❑ Message about conserving natural resources
❑ Message about saving money
❑ Message about processing data efficiently
❑ Uses of the data
❑ Randomly selected
❑ Required by law
❑ Answers are kept confidential
❑ Someone may call or visit
❑ Paper form looked too long
❑ Other -Record main ideas
❑ No
6.
Did you have any trouble with your User ID or PIN?
❑ Yes -- Can you tell me more about that? - Record main ideas
❑ No.
7.
Did you have any difficulty using the Internet to complete the survey?
❑ Yes -- Can you tell me more about why answering the survey using the
Internet was difficult? (Mark all that apply.)
❑ Slow Internet connection
❑ Session timed-out
❑ Couldn’t get back into the survey
❑ Power problems
❑ Other -Record main ideas
❑ No
If acknowledged there was a paper form in Q1, Q2b, or Q3 then ask Q8, Q9 and Q10;
otherwise skip to end.
8.
Did you use the paper form as a guide while you filled out the survey on the
Internet?
❑ Yes
❑ No
9.
Did you complete the paper form?
❑ Yes
❑ No – Skip to end
10.
Was that before you started the survey on the Internet, while you were on the
Internet, or after you completed the survey on the Internet?
❑ Before respondent started the survey on the Internet
❑ While the respondent was on the Internet
❑ After the respondent completed the survey on the Internet
END - GO TO INTERNET USE

E-3

Appendix F ABS Questions for Mail Respondents
ABS Questions
Assumptions: Previous questions will provide the introduction of the interviewer, verification
of the telephone number and address. Additionally, we will ask for the respondent by name
and set a callback if not available. Most response options are not read to respondents. Don’t
know and refuse are available response options for each question.
Screener
May I speak to ?
This call is about a Census Bureau survey you recently completed. First we’d like to thank you
for completing the questionnaire. Today, we’d like to collect some additional information.
1.

2.

2a.

2b.

If you recall, you recently received a large envelope from the U.S. Census Bureau.
The Census Bureau wants to improve its mailing materials. What materials do you
remember seeing in the envelope? (Mark all that apply)
❑ Form or Questionnaire
❑ Letter - If marked, skip to Q2a.
❑ Questionnaire Guide
❑ FAQ Brochure
❑ Return Envelope
❑ Postcard
❑ Internet Instruction Card
❑ Other – Record Main ideas
❑ Don’t remember the envelope
If "Letter" NOT marked in Q1
Did you see a letter about the survey?
❑ Yes
❑ No - Skip to Q3
Did you read the letter [Fill if "letter” marked in Q1: that was in the large envelope]?
❑ Yes
❑ No - Skip to Q3
What do you remember about what the letter said?
❑ Could have chosen between the Internet and the paper form
- If marked, skip to Q4a.
❑ Mentioned Internet only- If marked, skip to Q4a.
❑ Mentioned paper form only
❑ Listed uses of the data
❑ Randomly selected
❑ Required by law - mandatory
❑ Answers are kept confidential
❑ It provided phone number for help
❑ Signed by director
❑ Someone may call or visit
❑ Confusing
❑ Other- Record main ideas

F-1

3.

We received a completed form in the mail from you. As far as you can recall, could
you have chosen to respond over the Internet?
❑ Yes - If control panel skip to Q13; otherwise skip to Q4a.
❑ No - Skip to Q13
4a.
Why did you decide to use the paper form to complete the survey instead of using
the Internet?
❑ No access to Internet
❑ Lost/misplaced internet log-in instructions
❑ Inexperienced computer user
❑ Worried about security
❑ Convenience of the paper / Prefer paper forms
❑ Other – Record main ideas
4b.
If the paper form was not an option, would you have completed the survey on the
Internet?
❑ Yes
❑ No.
❑ Maybe
5.
Was there anything in the mailing materials that made you decide to use the paper
form to complete the survey?
❑ Yes - Please describe what in the mailing materials motivated you to use the
paper. Record main ideas
❑ No
If in Q4a the respondent reports they do not have Internet access or they lost the log-in
instructions, then skip to Internet Use; otherwise continue to Q6a.
6a.
Did you consider answering over the Internet?
❑ Yes
❑ No - Skip to Q6d
6b.
Did you try to log on?
❑ Yes - Skip to Q7
❑ No
6c.
Why didn’t you log onto the Internet? (Mark all that apply)
❑ I am too busy
❑ Forgot to
❑ Haven’t gotten to it (Respondent implies they will)
❑ The instruction card was too confusing
❑ Lost/misplaced the logon information
❑ User id/PIN login problems
❑ Other – -Record main ideas
Skip to Internet Use

F-2

6d.

7.

8.

Why didn't you want to answer over the Internet? (Mark all that apply)
❑ No access to Internet
❑ Inexperienced computer user
❑ Worried about security
❑ Convenience of the paper
❑ The instruction card was too confusing
❑ The survey was too long to do on the Internet
❑ No interest in the computer
❑ Other – Record main ideas
Skip to Internet Use
Did you have any trouble with your User ID or PIN?
❑ Yes - Can you tell me more about that? Record main ideas
❑ No
Did you start to answer the questions on the Internet?
❑ Yes
❑ No - Why didn’t you start to answer the questions on the Internet? Record main
ideas and Skip to Internet Use

9.

Did you have any difficulty using the Internet to answer the survey?
❑ Yes - Can you tell me more about why answering the survey using the Internet
was difficult?
(Mark all that apply.)
❑ Slow Internet connection
❑ Session timed-out
❑ Couldn’t get back in to the survey
❑ Power problems
❑ Other -Record main ideas
❑ No

10.

Did you finish answering the survey on the Internet?
❑ Yes
❑ No - Why didn’t you finish answering the survey on the Internet? -Record main
ideas and Skip to Q12

11.

Did you submit the completed survey on the Internet?
❑ Yes – About when did you submit the survey?
❑ Last week
❑ A few weeks ago
❑ Last month
❑ Sometime in March
❑ Sometime in April
❑ Sometime in May
❑ Other- Record main ideas
❑ No -- Why didn’t you submit the survey on the Internet?

F-3

Record main ideas

12.

13.

Did you complete the paper form before or after you went to the Internet?
❑ Completed paper BEFORE Internet
❑ Completed paper AFTER Internet
❑ Respondent indicated they completed the paper and Internet at the same time.
Skip to Internet Use
The Census Bureau is considering offering the survey over the Internet, in addition to
having the paper form. Which of those two ways to answer the survey do you
prefer? (Mark only one)
❑ Paper– Why do you prefer paper? Record main ideas
❑ Internet – Why do you prefer Internet? Record main ideas
❑ Neither option
❑ Prefer both options about the same
❑ It depends-- Please give some examples of when you would prefer to respond
over the Internet Record main ideas
Go to Internet Use

F-4

Appendix G ABS Questions for Nonrespondents
ABS Questions
Assumptions: Previous questions will provide the introduction of the interviewer and
verification of the telephone number and address. Most response options are not read to
respondents. Don’t know and refuse are available response options for each question.
Screener
May I speak to someone at least 15 years old who usually handles the mail and is
knowledgeable about the household at {read address} and the people in that household?
What is your name?
I’m calling from the U.S. Census Bureau about the American Community Survey that was
recently mailed to your address.
I am required by law to tell you that this survey is authorized by Title 13, Sections 141, 193,
and 221 of the United States Code. The average interview takes about 15 minutes. This
survey is mandatory and your cooperation is very important. All the information you provide
will remain completely confidential.
1.
Do you remember receiving a large envelope from the Census Bureau in March or
April?
❑ Yes
❑ No (probe for someone who opens the mail and may have seen the envelope and
restart the interview)
(if nobody remembers it, skip to hypothetical questions, H1).
2.
Did you open it?
❑ Yes - Skip to Q3
❑ No -- Go to Q2a
❑ DK -- Skip to hypothetical questions, H2
2a.
Why didn't you open it? (Mark all that apply)
❑ Already completed the Census
❑ I am too busy
❑ Forgot to
❑ Haven’t gotten to it (Respondent implies they will)
❑ Looked too big/too much work
❑ Threw it away
❑ Didn’t look official/Looked like junk mail
❑ Not familiar with Census Bureau
❑ It wasn’t addressed to me (said “Resident”)
❑ Other - Record main ideas
Skip to hypothetical questions, H2

G-1

3.

The Census Bureau wants to improve its mailing materials. What materials do you
remember seeing in the envelope? (Mark all that apply)
❑ Form or Questionnaire
❑ Letter – (if marked, then go to Q4.a)
❑ Questionnaire Guide
❑ FAQ Brochure
❑ Return Envelope
❑ Postcard
❑ Internet Instruction Card
❑ Other – Record main ideas

4.

If "Letter" NOT marked in Q3
Did you see a letter about the survey?
❑ Yes
❑ No - Skip to Q5a if "Form or Questionnaire" marked in Q 3, Otherwise go to Q5.

4a.

Did you read the letter [Fill if "letter” marked in Q3: that was in the large envelope]??
❑ Yes
❑ No - Skip to Q5a if "Form or Questionnaire" marked in Q3, Otherwise go to Q5.

4b.

What do you remember about what the letter said? (Mark all that apply)
❑ Could have chosen between the Internet and the paper form
❑ Mentioned Internet only
❑ Mentioned paper form only
❑ Listed uses of the data
❑ Randomly selected
❑ Required by law - mandatory
❑ Answers are kept confidential
❑ It provided phone number for help
❑ Signed by director
❑ Someone may call or visit
❑ Confusing
❑ Other- Record main ideas

5.

If "Form or Questionnaire" not marked in Q3
Did you see a questionnaire or form to fill out?
❑ Yes
❑ No - Skip to 6

5a.

Also ask if "Form or Questionnaire" marked in Q3
What do you remember about the questionnaire or form? - Record main ideas

G-2

6.

There were several different ways to respond to the survey. As far as you recall, how
could you have chosen to respond to the survey: by paper? over the Internet? or by
another way?
❑ Paper – Mail/Send it in
❑ Internet (if marked then go to Q7a)
❑ Other- Record main ideas
❑ Don't Remember

Internet Check
If "Internet” is marked in Q6, then ask this series beginning with 7a; otherwise Skip to Paper
Check
7a. Did you consider answering over the Internet?
❑ Yes
❑ No - Skip to 7d
7b. Did you try to log on?
❑ Yes - Skip to 7e
❑ No
7c. Why not? (Mark all that apply) If necessary: Any other reasons?
❑ I am too busy
❑ Forgot about it
❑ Haven’t gotten to it (Respondent implies they will)
❑ The instruction card was too confusing
❑ Lost/misplaced the logon information
❑ User ID/PIN login problems
❑ Other – Record main ideas
Skip to Paper Check
7d. Why didn't you want to answer over the Internet? (Mark all that apply)
❑ No access to Internet
❑ Inexperienced computer user
❑ Worried about security
❑ Convenience of the paper
❑ The instruction card was too confusing
❑ The survey was too long to do on the Internet
❑ No interest in the computer
❑ Other – Record main ideas

7e.

Skip to Paper Check
Did you have any trouble with your User ID or PIN?
❑ Yes -- Can you tell me more about that? Record main ideas
❑ No

G-3

7f.

Did you start to answer the questions on the Internet?
❑ Yes
❑ No -- Why didn’t you start to answer the questions on the Internet? Record main
ideas and Skip to Paper Check

7g.

Did you finish answering the survey on the Internet?
❑ Yes
❑ No -- Why didn’t you finish answering the questions on the Internet? Record
main ideas and Skip to Paper Check

7h.

Did you submit the completed survey on the Internet?
❑ Yes – About when did you submit the survey?
❑ Last week
❑ A few weeks ago
❑ Last month
❑ Sometime in March
❑ Sometime in April
❑ Sometime in May
❑ Other- Record response
❑ No -- Why didn’t you submit the survey on the Internet? Record main ideas

Paper Check: If "Paper” is marked in Q6, then ask this series beginning with 8a; otherwise Skip to
Check Other
8a.

Did you consider answering the survey on the paper form?
❑ Yes
❑ No -- Why didn’t you consider answering the survey on the paper form? Record
main ideas and Skip to Check Other

8b.

Did you start to answer the questions on the paper form?
❑ Yes
❑ No -- Why didn’t you start to answer the questions on the paper form? Record
main ideas and Skip to Check Other

8c.

Did you finish answering the survey on the paper form?
❑ Yes
❑ No -- Why didn’t you finish answering the questions on the paper form? Record
main ideas and Skip to Check Other

G-4

8d.

Did you mail in the completed survey?
❑ Yes – About when did you mail the survey?
❑ Last week
❑ A few weeks ago
❑ Last month
❑ Sometime in March
❑ Sometime in April
❑ Sometime in May
❑ Other- Record response
❑ No -- Why didn’t you mail in the completed survey? Record main ideas

Check Other: If "Paper” and “Internet" are marked in Q6, then Skip to Internet Use Questions
and then go to the Demographic Questions. Otherwise, ask Q9.
9.

Some people were offered two options for responding to the survey - filling out the
paper form or going to the Internet and filling out the survey there. Which of those
two ways to answer the survey do you prefer? (Mark only one)
❑ Paper– Why do you prefer paper? Record main ideas
❑ Internet– Why do you prefer the Internet? Record main ideas
❑ Neither option
❑ Prefer both options about the same
❑ It depends - Please give some examples of when you would prefer to respond over
the Internet -Record main ideas

Go to Internet Use then go to Demographic Questions
HYPOTHETICAL QUESTIONS
H1.
If you had received a large envelope from the Census Bureau, do you think you would
have opened it?
❑ Yes
❑ No -- Why wouldn’t you have opened the envelope?
H2.

If you had opened it and found a questionnaire to fill out, do you think you would
have completed it?
❑ Yes
❑ No -- Why wouldn’t you have completed the questionnaire if you found one in
the envelope?

H3.

If you had a choice of answering on the Internet or answering on paper, which would
you prefer?
❑ Paper– Why do you prefer paper? (Record main ideas)
❑ Internet– Why do you prefer the Internet? Record main ideas
❑ Neither option
❑ Prefer both options about the same
❑ It depends-- Please give some examples of when you would prefer to respond over
the Internet -Record main ideas
Go to Internet Use then go to Demographic Questions

G-5

Appendix H ABS Questions for Internet Use – For everyone in the ABS
ABS Questions
The next questions are about your Internet use.
1.
Do you use the Internet at home?
[] Yes
[] No
2a.

Do you use the Internet at work?
[] Yes
[] No

2b.

Besides at home or at work, do you use the Internet at some other place?
[] Yes
[] No – (If Q1=No and Q2a=No and not an Internet responder then go to closing)

3. If Internet responder, based on input flag: Where did you complete the survey?
❑ Home
❑ Work
❑ Library
❑ Other -Record main idea
If 'No" to Q1, Q2a and Q2b then go to end.
Else, if No to Q1 and “Yes” to Q2a or Q2b, skip to question 6.
4. What kind of Internet access do you have at home? Do you have dial-up service; DSL
service; cable modem service; fiber-optic service; mobile broadband plan for a computer
or cell phone; satellite service; or some other service?
❑ Dial-up service
❑ DSL service
❑ Cable modem service
❑ Fiber-optic service
❑ Mobile broadband plan for a computer or cell phone
❑ Satellite service
❑ Some Other service, Record main idea
5. What Internet browser do you typically use at home?
❑
❑
❑
❑
❑
❑

Windows Internet Explorer (IE), ____________Version
Mozilla Firefox, ____________Version
Chrome, ____________Version
Safari, ____________Version
Opera, ____________Version
Other -Record main idea

H-1

6. On an average day, about how many hours do you use the Internet, either at home, work
or somewhere else? (If necessary, Please include time you use the Internet on your phone
or on a regular computer.)
❑ Less than one hour Would you please describe how often you use the Internet?
For example, once a week, once a month?
❑ 1 hour
❑ 2 hours
❑ 3 hours
❑ 4 hours
❑ 5 hours
❑ 6 hours
❑ 7 hours
❑ 8 hours
❑ 9 hours
❑ More than 10 hours
7.
For the following activities or types of websites, please indicate how often you use or
visit them. Please answer very often, often, sometimes, rarely, or never?
a. How often do you use or visit news, map directions, job hunting, reviews or other
information or research gathering Internet sites?
❑Very often
❑Often
❑Sometimes
❑Rarely
❑Never
b. How often do you use or visit the Internet for… Facebook, MySpace, Twitter, blogging,
or other social networking? (Is it very often, often, sometimes, rarely, or never?)
❑Very often
❑Often
❑Sometimes
❑Rarely
❑Never
c. How often do you use or visit the Internet for… Paying bills, shopping or other financial
transactions? (Is it very often, often, sometimes, rarely, or never?)
❑Very often
❑Often
❑Sometimes
❑Rarely
❑Never
d. How often do you use or visit the Internet for… Emailing? (Is it very often, often,
sometimes, rarely, or never?)
❑Very often
❑Often
❑Sometimes
❑ Rarely
❑Never

H-2

e. How often do you use or visit the Internet for… Instant messaging, telephone calls,
Skype, or other means of communication besides email? (Is it very often, often,
sometimes, rarely, or never?)
❑Very often
❑Often
❑Sometimes
❑Rarely
❑Never
f. How often do you use or visit the Internet for… Viewing videos, downloading music or
podcasts, or other forms of entertainment? (Is it very often, often, sometimes, rarely,
or never?)
❑Very often
❑Often
❑Sometimes
❑Rarely
❑Never
g. How often do you use or visit the Internet for… Filling out surveys? (Is it very often,
often, sometimes, rarely, or never?)
❑Very often
❑Often
❑Sometimes
❑Rarely
❑Never
CLOSING -- If nonrespondent, go to demographic questions
Thank you for your time. / Thank you very much for your participation in this important
survey.
Your cooperation has made it possible for us to collect better data in the American
Community Survey. You’ve been very helpful.

H-3

Appendix I Demographic Questions for Nonrespondents
Now I am going to ask you some questions regarding your demographic information.
1. Including yourself, how many people live in your household?
2. ♦ Ask or verify. Are you male or female?
❑ Male
❑ Female
3. What is your age?
___ years old - If only 1 person in the household (from Q1) SKIP to question 5.
Otherwise go to question 4.
98 DON'T KNOW - Go to 3a
99 REFUSED - Go to 3a
3a. In which of these age categories do you belong?
READ LIST.
❑
❑
❑
❑
❑
❑
❑

Less than 18 years old
18-24
25-34
35-44
45-54
55-64
65 or older

If only 1 person in the household (from Q1) SKIP to question 5.
4. Are there any children in school who are under 18 living at home with you?
❑ Yes
❑ No
5. What is the highest grade or year of regular school you completed?
READ LIST IF NECESSARY.
❑ Less than high school graduation
❑ High school graduate
❑ Some college
❑ College graduate
❑ Postgraduate

6. Are you of Hispanic, Latino or Spanish origin?
❑ Yes
❑ No

I-1

7. I’m going to read a list of race categories. You may choose one or more races.
(Fill if Q6 = ‘yes’: For this survey, Hispanic origin is not a race.)
Are you White; Black, African American, or Negro; American Indian
or Alaska Native; Asian; Native Hawaiian or Other Pacific Islander; or Some other race?
♦ Enter all that apply, separate with commas.
❑ White
❑ Black, African American, or Negro
❑ American Indian or Alaska Native
❑ Asian
❑ Native Hawaiian or Other Pacific Islander
❑ Some other race
Now I’m going to ask about the place where you live…
8. Do you rent or own your house, apartment, or mobile home?
❑ Rent
❑ Own
❑ OTHER
8b. Which best describes this building? Is it a mobile home, single family house, building
with two or more apartments, boat, RV, or van?
❑ Mobile home
❑ Single family house (attached or detached)
❑ Building with 2 or more units or apartments.
❑ Boat, RV, van, etc
9. For last year - 2010 - what was the total income of your household BEFORE TAXES?
♦ Read all answer categories if necessary.
❑ Less than $25,000
❑ $25,000 to less than $50,000
❑$50,000 to less than $75,000
❑$75,000 or more

I-2