Field Test Summary-FINAL (Attachment 2)

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

Attachment 2
Summary of Field Test Findings and
Field Test Report
August 14, 2013

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

PATH Study Summary of Field Test Findings
and
PATH Study Field Test Report

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

PATH Study Summary of Field Test Findings
Study
Component

Measure

Performance Standard

Field Test Result(s)

Comments

Methods, Procedures, and Systems
Field
Operations

Interview length—Household
screener (adults)

Projected time: 17 minutes

12.7 minutes for short version,
14.0 minutes for long version

Length is acceptable, use
strongest items from each
version.

Interview length—Individual
screener (adults)

Projected time: 6 minutes

6.4 minutes

Length is acceptable, no
changes needed.

Interview length—Extended
interview (adults)

Projected time: 69 minutes

42.0 minutes

Length is acceptable, no
changes needed.

Interview length—Extended
interview (youth)

Projected time: 35 minutes

32.2 minutes

Length is acceptable, no
changes needed.

Projected time: 19 minutes

7.8 minutes

Field
Operations

Interview length—Parent
interview (adults)
Staff time needed to finalize
activities at a dwelling unit

Projected time: 1 hour

1.1 hours

Length is acceptable, no
changes needed.
Length is acceptable, no
changes needed.

Biospecimen
Collection

Consent rate for blood, urine,
and buccal cell samples

Experience from other studies
varies, depending on factors
such as incentives and number
of visits. Range observed:
Urine – 60-95%
Blood – 55-83%

Consent rates:
Buccal cell – 74.0%
Urine – 59.6%
Blood – 46.9%

PATH Summary of Field Test Findings

i

Modifications have been
implemented to improve
consent rates, including:
modifying the structure of
incentives, providing additional
information in consent
materials on the importance of
biological specimens to the
study, providing interviewers
with specific information to
address common respondent
concerns, and enhancing
interviewer recruitment and
training.

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

Study
Component

Measure

Performance Standard

Biospecimen
Collection

Rate of successful collection of
samples among those who
consent

The goal was 100% among
those who consented.

Biospecimen
Collection

Average time between
interview and blood collection
visit
Distribution of times from
collection to processing

The goal was to collect blood
between 3 and 14 days after
the interview.
Biological specimens were to
be processed within 72 hours
of collection.

Processing at
the Repository

Success in obtaining the
expected number of aliquots
from each collected
biospecimen

Number of aliquots specified
by the processing protocol

Sampling

Unweighted housing unit
eligibility rate

Design assumption:
88.6%

Biospecimen
Collection

PATH Summary of Field Test Findings

Field Test Result(s)
Rate of collection among
those who consent:
Buccal cell – 98.3%
Urine – 81.8%
Blood – 83.1%
The decrease in rate of urine
and blood collection is due
mainly to the inability to
complete the second visit.
The average time between the
interview and blood collection
visit was 8.3 days.
The time from collection to
processing ranged from 22 to
65 hours for specimens
collected by interviewers and
from 24 to 74 hours for
specimens collected by
phlebotomists, depending on
the day of the week specimens
were collected.

All urine aliquots were
obtained when an adequate
volume of urine was collected.
Fewer than expected small
volume plasma and serum
aliquots were obtained.
89.8%

ii

Comments
The rate of urine collection is
expected to increase for
baseline, because collections
will be done during the first
home visit.

Range was expected, not
unusual.
Times from collection to
processing greater than 72
hours were due to FedEx and
repository availability
schedules during the
Christmas and New Year’s
holidays. This will be avoided
for baseline collections by
imposing “blackout” dates for
specimen collection on days
that will result in delays to
process.
The processing protocols for
plasma and serum have been
modified for baseline to
ensure an adequate number of
small volume aliquots will be
obtained.
Assumption is reasonable. No
change needed.

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

Study
Component
Sampling

Sampling

Measure
Unweighted distribution of
enumerated adults

Unweighted tobacco use
misclassification rates

PATH Summary of Field Test Findings

Performance Standard

Field Test Result(s)

Design assumptions:
Black, 18-24 years old,
tobacco user: 0.3%

Black, 18-24 years old,
tobacco user: 1.7%

Black, 18-24 years old,
tobacco nonuser: 1.7%

Black, 18-24 years old,
tobacco nonuser: 3.3%

Black, 25+ years old, tobacco
user: 2.1%

Black, 25+ years old, tobacco
user: 5.1%

Black, 25+ years old, tobacco
nonuser:
8.4%

Black, 25+ years old, tobacco
nonuser:
6.0%

Non-Black, 18-24 years old,
tobacco user: 2.2%

Non-Black, 18-24 years old,
tobacco user: 7.2%

Non-Black, 18-24 years old,
tobacco nonuser: 7.9%

Non-Black, 18-24 years old,
tobacco nonuser: 11.4%

Non-Black, 25+ years old,
tobacco user: 16.3%

Non-Black, 25+ years old,
tobacco user: 22.8%

Non-Black, 25+ years old,
tobacco nonuser: 61.0%
Design assumptions:
False positive rate:
18-24 years old: 5%
25+ years old: 5%

Non-Black, 25+ years old,
tobacco nonuser: 42.5%

False negative rate:
18-24 years old: 5%
25+ years old: 2%

False negative rate:
18-24 years old: 32.1%
25+ years old: 9.7%

iii

False positive rate:
18-24 years old: 4.1%
25+ years old: 14.1%

Comments
For the baseline wave,
increase the assumed tobacco
use rates among the
enumerated adults during
household screener. The
extent of increase will vary by
age and race group.

For the baseline wave,
increase the assumed
misclassification rates for
tobacco use status.

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

Study
Component

Measure

Performance Standard

Sampling

Frame coverage rate (count of
addresses in the address list
frame purchased from the
vendor versus count of housing
units in Census 2010)

Design assumption:
About 95% for urban areas
and 70% for rural areas.

Sampling

Unweighted response rates*

No design assumptions were
made for field test. Original
design assumptions for
baseline wave:
Household screener: 87%,
based on findings from the
2011 National Survey of Drug
Use and Health (NSDUH)
Extended adult interview
(across individual screener and
main interview): 90%

Field Test Result(s)
Above 93% for 14 of 15 field
test PSUs.

Household screener: 39.7%

Extended adult interview
(across individual screener and
main interview): 58.1%
Extended youth interview
(including only those directly
sampled as youth): 62.6%

Extended youth interview
(including those directly
sampled as youth and those
originally sampled as shadow
youth who later age into the
youth cohort): 90%
Flow of Scientific Information

PATH Summary of Field Test Findings

iv

Comments
Assumption is reasonable.
Geocoding errors may affect
the coverage rate in the way it
is currently calculated. Address
coverage enhancement
procedures will be
implemented in the baseline
wave to improve coverage.
Purpose of field test was to
assess the data collection
protocol and operations. This,
in addition to an abbreviated
follow-up period, provide
context for the field test
sample yields and response
rates. For the baseline wave,
lower the expected response
rates for both screener and
extended interviews.

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

Study
Component

Measure

Performance Standard

Field Test Result(s)

Comments
The protocol has been
simplified to make it easier for
interviewers to follow.
Questions have been revised
and restructured to make
them easier to read as
intended. Training procedures
have been improved.
Logic checks were
implemented to reduce the
amount of inconsistent data.
Because the adult and youth
interviews are selfadministered, eliminating all
inconsistent data is unfeasible.
No changes needed.

Questionnaire
Performance

Computer-assisted recorded
interviewing review

Interviewers followed protocol,
and questions performed as
expected.

Interviewers had difficulty with
some portions of the protocol,
and some questions were hard
to read as intended.

Questionnaire
Performance

Items that produce
inconsistencies

No inconsistent data.

Some inconsistent data were
found.

Questionnaire
Performance
Questionnaire
Performance

Interview breakoff rate

Interview breakoff rate < 1.0%

Interview breakoff rate < 0.5%

One-way frequency
distributions of both
continuous and categorical
variables
Low variation items

Frequencies in expected
ranges

Frequencies were in expected
ranges

No changes needed.

Variation as expected for each
item type

Variation was as expected for
each item type

No changes needed.

Questionnaire
Performance

PATH Summary of Field Test Findings

v

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

Study
Component
Field Test
Experiment:
Household
Screener
Incentive ($0
vs. $5 vs.
$10)

Measure

Performance Standard

Field Test Result(s)

Household screener response
rate

Household screener response
rate is associated with
incentive amount

Marginally significant
difference in screener
response rates:
$0: 36.7%
$5: 40.3%
$10: 43.0%

Household screener/youth
extended response rate

Household screener/youth
extended response rate is
associated with incentive
amount

No significant difference in
conditional Phase 2/Youth
extended response rates:
$0: 59.4%
$5: 59.7%
$10: 64.0%

Number of contact attempts
per screener

Number of contact attempts
per screener is lower for higher
incentive amount

No significant difference in
contact attempts needed:
$0: 5.2
$5: 5.0
$10: 4.7

PATH Summary of Field Test Findings

vi

Comments
Results suggested a positive
effect of incentive amounts on
the household screener
response rates

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

Study
Component
Field Test
Experiment:
Long vs. Short
Household
Screener

Measure

Performance Standard

Field Test Result(s)

Household screener response
rate

Household screener response
rates are higher for short
version

No significant differences in
screener response rates:
Long: 39.5%
Short: 40.3%

Individual screener/youth
extended response rate

Individual screener/youth
extended response rates are
higher for short version

No significant differences in
individual screener/youth
extended response rates
Long: 36.0%
Short: 37.1%

Proportion of adults classified
as tobacco users

Proportion of adults classified
as tobacco users is higher for
long version

Significantly higher proportion
of adults classified as users
with long screener:
Long: 39.9%
Short: 33.5%

Agreement between
household and individual
screener classifications

Agreement between
household and individual
screener classifications is
higher for long version

Marginally significant
difference in rate of
agreement between household
and individual screener
classifications:
Long: 92.3%
Short: 88.0%

Comments
The final version of the
screener for the baseline wave
combines the best features of
both versions; it is shorter than
the long version used in the
field test (but longer than the
short version).

*The sampling design assumptions are national, whereas the field test results are for a purposive selection of 15 PSUs.

PATH Summary of Field Test Findings

vii

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

Population Assessment of Tobacco and Health
(PATH) Study
Field Test Report

PATH Field Test Report

i

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

Table of Contents

Chapter

Page
1

Introduction ...........................................................................................
1.1
Organization of the Field Test Report ..............................
1.2
Limitations of the Field Test Data .....................................

1
1
1

2

Operational Components of the Field Test ......................................

2

2.1
2.2
2.3
2.4

Questionnaire Performance ................................................
Biospecimen Collection .......................................................
Interviewing and Field Operations ....................................
Sampling .................................................................................

2
6
8
17

Field Test Experiments ........................................................................

23

3.1
3.2
3.3

24
25

3

3.4

Field Test Screener Response Rates ..................................
Household Screener Tobacco User Rates.........................
Response Rates to the Phase 2 Screener and
Extended Interview ..............................................................
Agreement Between the Household and Phase 2
Screener Classifications ........................................................

25
26

Table
11

Comparisons of PATH Study Instrument Timings ......................................

Appendix A
Detailed Tables

PATH Field Test Report

ii

9

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

1. Introduction to the PATH Study Field Test Report

This report presents results from the Population Assessment of Tobacco and Health (PATH) Study
field test conducted between December 2012 and February 2013. The purpose of the field test was
to assess the performance of study components under field conditions. The field test results have
helped to inform the PATH Study’s baseline wave of data and biospecimen collection, as reported in
this Information Collection Request to the Office of Management and Budget.

1.1

Organization of the Field Test Report

This report focuses on the main components assessed in the field test for the PATH Study’s
baseline wave, specifically:


Questionnaires and instrumentation performance,



Biospecimen collection and processing,



Length of time for interviews,



Interviewer workload needed to collect data,



Operational yields of interviews and biospecimens collected,



Sample design, and



Two embedded experiments regarding the household screener interview.

The report includes a number of detailed tables in an appendix, except for Table 11, which is
embedded in the text of this report.

1.2

Limitations of the Field Test Data

A major purpose of the field test was to assess overall protocol operations, including what worked
well and what did not, and to guide decisions on needed improvements. It was not designed to yield
results that could be generalized to the population as a whole. As such, information in this report is
limited to the field test sample and to operational metrics specific to procedures used in the field

PATH Field Test Report

1

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

test. The field test was also based on a nonrandom sample of 15 primary sampling units (PSUs);
with one exception, explained later in the text, the data in this report are unweighted.
Important limitations of the field test include:


Operations. The field test implemented a limited protocol over a compressed data
collection period. Many of the cases were incomplete, because work on them ended
when the field test period ended. The refusal conversion protocol was briefly
implemented for test purposes only; by contrast, the main study will implement this
protocol on a consistent basis and for every appropriate case.



Sampling. PSUs for the field test were purposively selected because their
characteristics were of interest for the main study, such as population size, region,
urban/rural locations, racial representation, and tobacco use behavior. Sample sizes
were small, especially for rarer subgroups (e.g., 18-24 years old Black tobacco users);
unless otherwise noted, estimates herein for these subgroups are unreliable.



Response Rates. The field test focused on assessing the study protocol and overall
operations within a compressed period of time; these priorities came first.
Consequently, less emphasis was given to achieving high response rates or completing
open cases still in process. These limitations will not apply for the main study,
however, indicating that the low response rates in the field test should be viewed as
the “worst case.”

2. Operational Components of the Field Test

In this section of the report, field test findings are presented on questionnaire performance,
biospecimen collection, interviewing and field operations, and sampling.

2.1

Questionnaire Performance

Questionnaire performance was assessed in the field test by examining computer audio recordings
of select instruments, the interview breakoff rate, and data inconsistencies and distributions.

PATH Field Test Report

2

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

2.1.1

Computer Audio Recordings of Select Instruments

Computer Audio Recorded Interviewing (CARI) provides direct audio recording of the ambient
sound during certain portions of the interview, consisting mostly of the conversation of the
interviewer and respondent. Data in CARI recordings can be used for purposes of validating
interviews, reviewing protocols, and reviewing the performance of questions. CARI files are large
and can become operationally cumbersome; for this reason, use of CARI in the field test was limited
to specific aspects of the interview identified as most informative to record.
For the PATH Study field test, CARI was employed to review both the performance of questions in
the household screener and the biospecimen consent process.

Household Screener Interview CARI Review
Qualitative assessments of CARI recordings focused on problems with any questions, the questions
asked by respondents, the use of show cards, and uncertainties or contradictions in respondent
answers.
Table 1 provides an overview of the characteristics of respondents whose CARI recordings were
qualitatively assessed. The 547 recordings reviewed and analyzed were selected based on household
and respondent characteristics of interest, including age, gender, and household size.
The following issues were identified from the household screener recordings; the household
screener has since been revised for purposes of the upcoming baseline wave.


Household enumeration questions, including the number of people in the household
and their basic demographic characteristics, did not appear confusing or difficult to
either respondents or interviewers. An exception to this was the race question, where
the long list of response options (i.e., 14 options read out loud by the interviewer to
the respondent for each person who lives in the household) proved problematic.
Respondents appeared to be increasingly irritated by having to listen to this long list
for every person in the household, regardless of whether they were related to each
other. This led interviewers to take shortcuts by, for example, skipping to the category
the respondent had already identified for one household member to see if that same
category applied for another household member.



The tobacco use questions were difficult to administer, especially for the long version
of the household screener. Here too, the detailed nature of these questions seemed to

PATH Field Test Report

3

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

lead interviewers to take shortcuts. For example, interviewers skipped the information
they were supposed to read, such as information that appeared in parentheses and the
brand name examples. Respondents also seemed increasingly irritated by these detailed
questions.


In an effort to shorten the screener, the tobacco use questions were structured so that
a “yes” response to any tobacco use item skipped the respondent out of the rest of the
section. This skip seemed to confuse interviewers, however, because the questions that
prompted the skip were inconsistent in the section. This may have contributed to
respondent irritation as well.



Some questions appeared difficult for the interviewers to administer due to the way
they were formatted on the computer-assisted personal interviewing (CAPI)
questionnaire screen. For example, the question stem was visually separated from the
response options in some cases, which led interviewers to ask the question in an
awkward-sounding way or to mistakenly omit response options from the question.



The field test used a number of hard-copy show cards for the household screener,
such as show cards with pictures of tobacco products and cards with text response
options (such as the 14 race categories). Interviewers used the show cards
inconsistently.



The contact information questions were lengthy but generally straightforward to
administer and not problematic for respondents. This was not the case for the social
media questions about the respondent’s Facebook name or Twitter handle.
Respondents were often unfamiliar with what these meant and occasionally seemed
defensive or reluctant to provide the information.



Questions on the relationship of everyone in the household to the sampled individuals
were straightforward to administer and not problematic for respondents. Interviewers
had some difficulty ascertaining the direction of the relationship, for example, was
person X the son or father of person Y.

Biospecimen Consent CARI Review
CARI recordings of the biospecimen module were reviewed to understand respondents’ reasons for
refusing to provide biospecimens. The usefulness of the recordings was limited for this purpose,
however: by the time the recording began, the interviewer had already moved on to discuss which
specimens the respondent would consent to provide. Also, the length of the recording was shorter
than desired. Still, some useful observations were captured in the recordings. Refusal reasons for
not consenting to provide a blood specimen, for example, included not wanting another
appointment, fear of needles, and discomfort providing specimens; refusal reasons for urine
included concern about drug testing and the length of time for the collection.

PATH Field Test Report

4

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

2.1.2

Interview Breakoff Rate

A final breakoff refers to a questionnaire that began but was never completed. This discussion
focuses on only final breakoffs. A breakoff point is the last item or section in the questionnaire that
the respondent answered before the breakoff. Breakoff points are of interest if they occur in clusters
around certain questions or question types, or at certain points in the questionnaire.
In the PATH Study field test, the breakoff rate was very low, with only 4 final breakoffs in the adult
instrument and no final breakoffs in the youth instrument. The small number of breakoffs was
insufficient for detecting a meaningful or useful pattern.

2.1.3

Data Review

As noted, the PATH Study field test was not designed to yield results that could be generalized to
the population as a whole. Rather, its data were intended for use in improving operational and
procedural aspects of the PATH Study’s data and biospecimen collection protocol for the baseline
wave. Consequently, PATH Study field test data were examined for inconsistencies, problematic
frequency distributions, and notably low variance items.


Inconsistencies. Data from all instruments were reviewed using automated error
checking programs, cross tabulations, frequencies, and comment review to determine
if data had been captured and stored correctly, algorithms were working as expected,
and inconsistencies were properly identified. When an inconsistency was identified,
data management staff proposed a logic check that highlighted the inconsistency for
the respondent and prompted him/her to consider revising the response.



Frequency distributions. One-way frequency distributions of both continuous and
categorical variables were reviewed to identify questions that seemed to produce
responses outside the expected range and distribution for similar questions. No outof-range frequencies or unexpected distributions were identified.



Low variation items. Variation in responses was examined for both continuous and
categorical variables to identify questions that seemed to produce less variation in
responses than expected for similar questions. No low variation items were identified.

PATH Field Test Report

5

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

2.2

Biospecimen Collection

Biospecimen collection was assessed in the field test by examining biospecimen consent rates;
collection, packaging, and shipping; and processing at the repository.

2.2.1

Biospecimen Consent Rates

Table 2 presents the rates of biospecimen consent by biospecimen type and various respondent
characteristics. Some of the break-outs indicate that: (1) users of tobacco products consented at
higher rates than non-users, (2) users of controlled substances consented at a higher rate than nonusers (although the counts are not high), (3) consent rates trended downward from the 18-24 age
category to the 65 years and older category, and (4) persons with less education consented at a
higher rate than persons with more education.
2.2.2

Biospecimen Collection, Packaging, and Shipment

Blood was collected by a trained professional or phlebotomist during a second scheduled visit
following the interview. Table 3 shows the rate of successfully collecting each blood tube type over
all blood collections.
Table 4 shows the rate of successfully collecting urine from respondents who consented, 81.8
percent overall (234/286). The weight of collected urine is recorded on receipt at the repository.
79.5 percent of collections weighed >= 42.7 grams, the amount of urine needed to create all of the
intended aliquots. Persons 65 years old or greater had a higher rate of specimens weighing below the
42.7 gram threshold.
The buccal cell specimen was always collected a short time after the biospecimen consent was
administered. Table 5 shows that collection of the specimen was successfully collected for 98.3
percent of respondents who consented (349 out of 355).
When a biospecimen was not collected from a respondent who consented, the reason for not
collecting the specimen was recorded. This was an infrequent occurrence (n=29 blood and n=32
urine collections). The main reason for not collecting blood (n=14) and urine (n=14) was “no
show,” recorded when the respondent did not keep a scheduled appointment.

PATH Field Test Report

6

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

Blood was collected by phlebotomists at separate visits following the interviews; urine was collected
at most of these visits as well. Table 6 shows that for respondents who consented to blood
collection, a visit was conducted for 96.0 (213/222) percent of them, and blood was collected for
83.3 (185/222) percent. For the 4 percent of respondents for whom a visit was not conducted, the
reason was the inability to successfully schedule an appointment. A visit was conducted but a blood
specimen was not collected for 13 percent of respondents; this was due to 6 percent “no shows”
(respondent not at home) and no reason was recorded for the remaining 7 percent.
Blood specimens were to be collected within an optimal window from the 3rd to 14th day after the
adult interview if possible or within an acceptable window up to 21 days after the interview. Table 7
shows that the median number of days between the interview and day of blood collection was 7. In
6 of 15 PSUs, all collections were completed within the optimal window (Max <= 14 days); and, in
all but two PSUs, all collections were completed within the acceptable window (Max <=21 days). All
PSUs except one performed similarly to each other in terms of the median number of days. The one
exception was 10 days; this PSU also had a small number of collections (n=6).
The goal for shipping specimens was to minimize the time in transit using FedEx Priority Overnight
shipping. Ideally, specimens collected on weekdays would be delivered at the repository the next
morning. However, specimens collected late on Friday or on Saturday and Sunday would not ship
until Monday. Because transit times could be as long as 72 hours, packing and shipping procedures
had to be designed to keep specimens cold for 72 hours. Table 8 presents the times in transit
observed for the different packing day segments. Focusing on interviewer- and phlebotomistinitiated shipments, the median times in transit for specimens collected on weekdays before 6:00 pm
were less than 24 hours. For specimens collected on Fridays after 6:00 pm, the median time in
transit increased to 38 and 37 hours for interviewers and phlebotomists, respectively. For specimens
collected on Saturday the median time in transit was 65 and 74 hours for interviewers and
phlebotomists, respectively. For specimens collected on Sunday, the median time in transit was 42
and 47 hours for interviewers and phlebotomists, respectively. Some very long times in transit were
observed. These collections occurred during the Christmas and New Year holidays, when some
collected specimens had to be held for some days due to FedEx and repository schedules.
Table 9 presents the time from specimen receipt to the start of processing at the repository for the
collected specimens that were processed. The median time for all specimens was 3 to 4 hours, and
the maximum recorded time was 8 hours.

PATH Field Test Report

7

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

2.2.3

Processing at the Repository

Table 10 summarizes the number of aliquots created at the repository for each parent specimen
type. For urine, all expected aliquots were created when the minimum required amount (42.7 grams)
was collected. As expected, fewer aliquots were created when less urine was collected. For plasma
(EDTA) and serum, fewer 0.75mL vial aliquots were created than expected from the processing
protocol. The explanation is that these are the last aliquots created per the protocol and were not
made when the yield of plasma or serum was less than expected.

2.3

Interviewing and Field Operations

Interviewing and field operations were assessed in the field test by examining interview length of
time, interviewer workload, and interview yields.

2.3.1

Interview Length of Time

Testing the length of time (or “timings”) for the various interviews was a key objective of the field
test. Interview timing has an impact on respondent burden, operational planning, and cost.
Table 11 summarizes three sets of timings for the various instruments and components: (1) as
estimated for the PATH Study field test OMB package, (2) as ultimately desired for the full baseline
implementation, and (3) as actually experienced in the field test. The estimates presented in the field
test OMB package were based on in-house timing tests performed on the final field test versions.
These estimates were set slightly higher than the desired targets, because one purpose of the field
test was to include as much of the desired content as possible in the core instruments before
determining whether any items had to be cut for the baseline.
In terms of the instrument timings, the main goal for the field test was to measure the length of time
each instrument would require in a real-world setting that was as comparable as possible to the
baseline wave of data collection. The primary metric was the mean time required to administer each
instrument across all respondents; other indicators included standard measures such as the median,
minimum and maximum time required for each instrument.

PATH Field Test Report

8

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

Table 11.

Comparisons of PATH Study Instrument Timings

Interview
Household Screener

Projected mean
timing for field
test in OMB
package
(minutes)
17

Desired mean timing
target for full
implementation at
baseline
(minutes)
15

Actual mean timing
from the field test
(minutes)
Short: 12.7
Long: 14.0

Adults – Individual Screener

6

Adults – Extended Interview

69

Included in
Adult Extended
60

6.4

Adults – Biospecimen Collection Forms

9

Not projected

*

Adults – Tobacco Use Form (NEQ)

2

Not projected

3.5

Adults – Followup/Tracking Participant
Information Form (contact information)

6

Included in
Adult Extended

7.9

Youth – Extended Interview

35

30

32.2

Adult – Parent Interview

19

15

7.8

42.0

* Data not available; timings for the various biospecimen collection forms (other than the Tobacco Use Form) were not
tracked separately from the activities that involved these forms (e.g., urine collection, blood collection).

The timings presented exclude a few outliers that appear to be the result of human or machine error
(e.g., the computer continuing to record the time after the interview was completed). As a result, the
number of observations for the same instrument may vary between this section and other sections
of this report.

Phase 1 Household Screener
Table 12 presents the household screener timings, by screener version (long or short) and household
size. Differences between the two versions were as expected—the longer version took longer to
administer, on average. Overall, the mean time for the long version was 14.0 minutes, and for the
short version it was 12.7 minutes. The timings increased with household size in both versions; and,
for the same-sized household, the longer version always took more time on average than the shorter
version. For both versions, the timings ranged from under 3 minutes to approximately one-and-aquarter hours.

PATH Field Test Report

9

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

Phase 2 Individual Screener
As shown in Table 13, the mean time to complete the Phase 2 screener was 6.4 minutes across all
respondents. (For consistency with other tables and with the other panels in this table, the analysis
includes only those who were actually sampled in Phase 2 to continue in the survey and completed
the extended adult interview [n =461].) The range was from 2.0 minutes to 19.7 minutes.
Table 14 examines the Phase 2 screener timings by tobacco use status and the number of products
used. For this assessment of Phase 2 screener timings, tobacco use has been classified as ever user of
tobacco and never user of tobacco, because ever use of a given product triggers questions about that
product. The mean time for never users was 5.3 minutes and for ever-users it was 6.6 minutes. The
mean time generally increased as the number of products ever used increased. The maximum time
for completing the Phase 2 screener was 19.7 minutes, for a respondent who had ever used four or
more tobacco products.
Table 15 is similar to Table 14, but breaks out the Phase 2 screener timings for never users and ever
users by four age categories. The mean time generally increased with age for both never users and
ever users; the exception to this pattern was for the 65 years and older never user group, but this
finding is not reliable due to the small sample size. Across the two user groups, the ever users always
had higher timings than the never users for the same age group.

Adult Extended Interview
Table 13 examines the timings for completed adult interviews. It disaggregates the various
components and sections of the interview and also breaks the timings out by tobacco use status. As
compared with Tables 14 and 15, where tobacco use was characterized by ever use, this table
characterizes tobacco use by current use or no current use (i.e., never or former users).
Across all respondents included in the table (n = 461), the mean timing for the entire adult
instrument was 48.4 minutes, which includes the Phase 2 screener (6.4 minutes) and the balance of
the substantive sections of the adult extended interview (42.0 minutes). The mean timing for entire
adult instrument ranged from 16.5 minutes to over two-and-a-half hours (154.1 minutes).
The Contact section required 7.9 minutes on average. The Consent section and ACASI Tutorial
section (on answering self-administered questions) took a mean time of 12.0 minutes. The Total

PATH Field Test Report

10

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

Interview Process covering all the components of the adult interviewing process (Consent and
Tutorial, Phase 2 screener, Extended Interview, and Contact Information) took a mean time of 63.7
minutes. The Health Effects Outcome section averaged 14.1 minutes.
The Current Tobacco Users (n = 227) and Current Non-Users (n = 234) panels of Table 13 allow
comparisons between current tobacco users and current non-users in regard to the adult interview
timings. As observed in the Phase 2 screener, the timings for the users were consistently higher than
for non-users, for the overall extended interview and for the individual tobacco-related sections; but
the timings were more or less comparable for sections not directly dealing with tobacco use (i.e.,
most of the sections in the latter half of the instrument).
Table 16 breaks out the adult extended timings by age, education, and gender, and, as in Table 13,
further by current tobacco use status. The mean time increased with age and decreased with
education; men took slightly longer than women (all respondents). Each of these demographics is
known to interact with tobacco use status; and tobacco users, generally speaking, have more
questions to answer in the PATH Study interview. However, these demographic patterns do not
seem to be confounded by the tobacco-use interaction; that is, they manifest themselves consistently
for both tobacco users and non-users, albeit with lower mean timings for the non-users than the
users.
Table 17 further demonstrates the expected impact of current tobacco use on the adult extended
interview timings. Among current users, the mean time increased consistently with the number of
products currently used (although a small number of respondents were currently using 3, or 4 or
more products). This pattern was not evident for the nicotine dependence section, however, because
the dependence section questions were asked only once for tobacco in general, regardless of which
or how many products were used.

Youth Extended Interview
Table 18 shows that the mean time for the youth extended interview was 32.2 minutes for all
respondents, and longer for the small number of current tobacco users (49.7 minutes, n = 9)
compared to the large number of current non-users (30.9 minutes, n = 114). The timings ranged
from 11.0 minutes to 73.1 minutes.

PATH Field Test Report

11

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

Timings by instrument section indicate that the higher timings for current tobacco users are in the
sections on tobacco use. This can be seen by comparing means in the current tobacco users panel
to the means in the same respective rows in the current non-users panel. For example, the mean for
the cigarette use section is 5.7 minutes for tobacco users and 2.1 minutes for non-users. In contrast,
with a few exceptions, the means for the sections that do not deal directly with tobacco use (mainly
those at the bottom of the table) are similar between the two groups.
Table 19 shows virtually no difference in the timings for the younger youths (12 – 14 years old) and
the older ones (15 – 17 years old). The means and medians are nearly the same for the two age
groups, overall and for current tobacco users and non-users.
The number of products used seemed unrelated to the length of the youth interview for current
tobacco users, as compared to the adult interview, where the relationship was more apparent (Table
20 vs. Table 17). However, the small number of youth users (9) and the near absence of youths who
used more than one product suggest that these findings be considered highly preliminary.

Parent Interview
As seen in Table 21, the mean time to administer the Parent Interview was 7.8 minutes, with a range
of 3.0 minutes to 18.2 minutes.

2.3.2

Interviewer Workload: Interviewer Hours per Case

The overall hours per completed case has importance for estimating future data collection costs. To
compute hours per completed case for these analyses, a completed case was defined as a sampled
dwelling unit for which a household screener was completed, and the amount of time required for
that case covers all in-home activities related to screening, individual interviews, and any specimens
collection by the field interviewer. Hence, the results presented in this section are at the dwelling
unit level. Operationally, hours per case includes all time logged on interviewer tablet computers for
all PATH Study activities completed within those dwelling units. This measure excludes the sizable
number of interviewer hours allocated to contact attempts; it also excludes interviewer hours
allocated to administrative activities and hours expended by the phlebotomists.

PATH Field Test Report

12

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

As indicated in Table 22, the mean number of hours per completed case was 1.1 hours, and the
median was 0.6 hours. This table breaks out the number of hours for completed cases by the
number of sampled persons for dwelling units. The mean number of hours per completed case
increased steadily with the number of sampled persons, from 0.2 hours for no sampled person to 6.3
hours for four or more sampled persons.
The mean number of hours per completed case varied by the specific combinations of sampled
persons at dwelling units. The mean number of hours was smaller for one sampled youth (1.2 hours)
than for one sampled adult (1.5 hours). (For this table, sampled is defined as sampled at the Phase 1
screener stage.) Although one might expect to see fewer hours per sampled person in households
with multiple sampled persons, this was the case for youth but not for adults. The mean for one
sampled youth was 1.2 hours, but for two sampled youths, it was 2.0 hours (i.e., less than double the
hours for one youth); the mean for one sampled adult was 1.5 hours, but for two sampled adults it
was 3.4 hours (i.e., more than double the hours for one adult). For the same number of sampled
persons, one can see a consistent downward gradient as the mix goes from more adults/fewer
youths to more youths/fewer adults. This phenomenon reflects the countervailing effects of the
shorter youth interview and the need to contact and interview two persons per youth (youth and
parent) plus the additional time for the parent interview.
Table 23 breaks out the number of hours for completed cases by the number of extended interviews
completed per dwelling unit. The mean number of hours per completed case increased steadily with
the number of completed extended interviews, from 0.4 hours for no completed interviews to 7.0
for four completed interviews.
The number of hours per completed case varied by specific combinations of extended interviews
completed at dwelling units, and by the number of sampled adults who provided cheek cells and/or
urine samples. The mean number of hours was smaller for one sampled youth extended interview
(1.5 hours) than for one sampled adult extended interview (2.0 hours). For both youth and adults,
one can see efficiency in fewer hours per completed interview in households with multiple extended
interviews. The mean for one interviewed youth was 1.5 hours, but for two interviewed youths, it
was 2.3 hours (i.e., less than double the hours for one youth); the mean for one interviewed adult
was 2.0 hours, but for two sampled adults it was 3.7 hours (i.e., less than double the hours for one
adult). The same gradient observed in Table 22 for the numbers and combinations of sampled adults
and youths likewise occurs for the numbers and combinations of interviewed adults and youths.

PATH Field Test Report

13

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

In addition, this table indicates that the hours per case increased dramatically by the number of
adults who provided cheek cells and/or urine specimens. For example, the mean number of hours
was 0.6 for cases with no adults providing specimens and 4.2 hours for those with two adults
providing specimens. Potential reasons for the number of hours being high for the cases with
biospecimen collections include: (1) some collections required follow-up visits, which decreased
efficiency; and (2) perhaps more important, adults who provided biospecimens also completed
extended interviews, whereas the group with no specimens is a mixture of sampled adults who were
interviewed but provided no specimen and sampled adults who were not interviewed at all.

2.3.3

Response Rates

This section presents information on response rates based on completed interviews at various
sampling stages. Using “response rates” in this context is limited, however, because:


The sample is not a random sample.



The percentages are not weighted, as is best practice for reporting response rates.



The field test did not implement the full contact protocol over a full data collection
period as will apply to the main PATH Study survey. This means that many cases in
process were incomplete, because they stopped at the end of the field test period.

Table 24 presents the interview response rates for each sampling and interviewing stage, in the order
in which they occur in data collection. The field sample of addresses resulted in 2,944 addresses
confirmed (or presumed in the absence of evidence to the contrary) to be residential, which are
defined as the addresses eligible for PATH Study and at which a completed household screener was
sought. The field test completed the screener at 1,170 (39.7%) of these addresses.
The 1,170 completed household screeners resulted in 1,152 adults sampled for the Phase 2 screener.
The Phase 2 screening process was completed for 698 (60.6%) of these sampled adults. Notably, the
yields were higher for those sampled in the household screener as tobacco users (66.0%) than as
non-users (55.2%).
The 698 completed Phase 2 screeners produced 501 adults sampled for the field test, of which 480
(95.8%) yielded completed adult interviews. The target number of completed adult interviews for the
field test was 600. The obtained sample size of 501 adults was judged to be large enough to test the
instrument across a variety of respondent characteristics and all the sampling strata, as well as to test

PATH Field Test Report

14

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

the performance of the questions across various major skip pattern paths, while being small enough
to accomplish within the field test’s timeframe and budget.
The Phase 2 screener response rates by tobacco use status are associated with the adult interview
response rates by tobacco use status, because of two factors: (1) as discussed in Section 2.4, the
Phase 1 tobacco use status conformed well with the Phase 2 tobacco use status; and (2) most who
completed the Phase 2 screener and were sampled for the adult interview completed the adult
interview. With this consideration in mind, response rates for completed interviews were slightly
higher among those sampled at Phase 2 as tobacco users (96.9%) than among those sampled as nonusers (91.9%).
The household screener also produced 195 youths sampled for the field test. By design, the field test
subsampled youths in order not to greatly exceed the target of 100 completed youth interviews. This
target was set to produce a sufficient number of completed youth interviews to test the performance
of the questions across various major skip pattern paths, while balancing the data collection cost for
the youths with the demands of the various other objectives of the field test. Of these 195 sampled
youths, 122 completed the interview (62.6%). The response rates were nearly identical for younger
and older youths, a desirable result.
Among the 195 parents/guardians identified for the 195 sampled youths, the field test completed
interviews with 128 of them (65.6%). The response rates were nearly identical for parents of both
younger and older youths. Not shown in the table is the fact that every interviewed youth had a
matching interviewed parent; the exception was one emancipated youth for whom, by definition, a
parent/guardian was not applicable.
Table 25 examines nonresponse at each stage. For each set of non-responders, it breaks out the
percentages attributable to each of the major reasons for nonresponse. The reasons are as follows.


Refusal. Someone refused to participate in an interview. Typically, this is the targeted
person, but sometimes is another household member (a so-called gatekeeper) who
refuses as a proxy to allow access or to convey to the sampled person the request for
an interview.



Maximum Contacts. At least one contact was made with the targeted person (e.g., an
adult in the household to conduct the household interview, the sampled adult,
sampled youth, or youth’s parent), but after making repeated attempts to conduct the
interview, the effort was not successful before reaching the maximum of six in-person

PATH Field Test Report

15

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

attempts allowed by the data collection protocol (or after an increased number of
contacts was made beyond the minimum of six, as circumstances warranted).


No Contact. Similar to Maximum Contacts, except that, after at least six attempts, no
contact was actually achieved with the targeted person.



Language barrier. The targeted person did not speak English (the only language in
which the field test was conducted).



Physical Barrier. The interviewer was unable to gain physical access to the dwelling
unit, due to restricted access situations such as gated communities or controlled-access
buildings.



Other. For the field test, this category included cases that were still in process at the
end of the field period and did not qualify for the other categories in the table.

Typically, refusal is by far the most common reason for survey nonresponse, and that was the case
for all the discrete stages presented in Table 25. (The adult extended interview stage is excluded
from this statement because it is not discrete from the Phase 2 screener in the eyes of the
respondents, so refusals rarely apply to this stage.)
For the household screener, 59.7 percent of the nonresponse was due to refusals and 21.7 percent to
maximum contacts. This is as expected for an in-person household-based survey. Another 10.5
percent was due to language barrier; the addition of Spanish interviews to the main study is expected
to reduce this loss.
For the Phase 2 screener, 50 percent of nonresponse was due to refusal, and another 20 percent was
due to No Contact. The latter phenomenon means that the interviewer was never able to make
contact with the adult who was sampled from the household screener. The sources of nonresponse
were similar for Phase 1 tobacco users and non-users.
As noted, nonresponse was small for the adult interview that flows seamlessly from the Phase 2
screener. Most nonresponse fell into the All Other category, and may include breakoffs in the
middle of the instrument. Interpreting the percentages presented in this panel should consider the
small numbers in the nonresponse group. For example, the 85.7 percent of nonresponse in the All
Other category for all adult interviews is only 18 cases (out of the total of 21).
As with the adult interview, the preponderance of nonresponse to the youth and parent interviews is
due to refusal and non-contact, ignoring the All Other category which consists predominantly of inprocess cases on which work was stopped due to the end of the data collection period. The

PATH Field Test Report

16

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

distribution of the reasons for nonresponse was similar for parents of youths in both age groups and
for the youths in both age groups; again, the actual small numbers underlying apparently large
differences in percentages should be taken into account.
For the three individual-level interviews (adult, youth, and parent), the large percentages appearing
the All Other category are a function of small numbers and also of the artificial stopping of data
collection efforts at the end of the field test period.

2.4

Sampling

Sampling was assessed in the field test by examining unweighted data pertaining to design
assumptions; and weighted frame coverage rates, occupancy rates, response rates, and effective
coverage rates.

2.4.1

Unweighted Estimates for Examining Design Assumptions

The sample design was based on assumptions about housing unit eligibility rate, screener and main
interview response rates, distributions of 9-to-11 year-olds and 12-to-17 year-olds, distribution of
adults in the sampling domains cross-classified by age, race, and tobacco use, as well as rates of
misclassification of household members’ tobacco use by household informants. Tables 26 through
28 compare the unweighted estimates of housing unit eligibility rate, household response rate,
presence of youth, distribution of enumerated adults, and misclassification of tobacco use against
the expected quantities, to assess the accuracy of the design assumptions. Although the size of the
field test sample is too small to yield precise estimates, large differences between design assumptions
and the actual field test results can point to aspects of the sample design that need to be reviewed
and modified.
Table 26 shows that the design assumptions for housing unit eligibility rate and presence of youth
are reasonable. The unweighted screener response rate (raw sample yield) from the field test is 40.1
percent. The field test experience of the screener response rates needs to be qualified by issues
discussed earlier, namely the reduced protocol and shortened field period of the field test.
Nonetheless, the field test experience, in conjunction with screener response rates in other similar
household field studies such as the National Epidemiologic Survey on Alcohol and Related
Conditions (NESARC), argues for lowering the household screener response rate assumed for the
PATH Field Test Report

17

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

purpose of designing the main study, and the total number of sampled addresses would need to
reflect the revised expectation.
Table 27 gives the unweighted counts and proportions of the enumerated adults in the eight
sampling domains defined by age, race, and tobacco use. The field test result is based on the
information provided by the household informants in screened households. The design assumptions
were obtained by multiplying the population counts from Census 2010 (by age*race) by the
estimated tobacco use rates (by age*race) from the TUS-CPS conducted in 2006-2007 that included
both self-response and proxy-response. Alternative parameters of population distribution can be
obtained by using the estimated tobacco use rates from the National Survey of Drug Use and Health
(NSDUH) and the National Health Interview Survey (NHIS); and these parameters are also shown
in Table 27. The percentage columns show the distribution across the entire population (i.e., they
sum to 100 percent).
The distribution of the enumerated adults in screened households and the assumed population
distribution seem substantially different in some sampling domains. For example, the proportion of
non-Black 18 to 24 year-old tobacco users is 7.2 percent in the PATH Study field test and 2.2
percent in TUS-CPS 2006-2007; and the proportion of non-Black 25 years and older tobacco nonusers is 42.5 percent in the PATH Study field test and 61.0 percent in TUS-CPS 2006-2007. The
following factors may help explain the divergence of the field test result from the design
assumptions.


The field test result reflects the distribution of screener respondents in the 15 PSUs
chosen for the field test, while the design assumptions are based on national estimates.
The distribution of adults by age, race, and tobacco use in the 15 field test PSUs may
not align with that in the nation; indicators of higher tobacco use and minority
populations in some PSUs were one of the purposive sampling criteria for the field
test PSUs.



If the screener response rates differ by household characteristics (e.g., Black vs. nonBlack households, households with tobacco users vs. those without) or the household
informant tends to miss adults in particular sampling domains (e.g., young persons
who may be more mobile), then the tabulation based on the screener respondents’
information may diverge from the population distribution.



The small sample sizes in some domains make the estimates unstable (e.g., Black 18 to
24 year-old tobacco user).



The estimates of tobacco use rates differ considerably between the available external
data sources, as shown in the columns for TUS-CPS 2006-2007, NSDUH 2010, and
NHIS 2010. The difference between these external sources may be partly explained by

PATH Field Test Report

18

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

data collection mode, length of field period, and the form of the questions and the
tobacco products used in the definition of tobacco use. The PATH Study definition of
tobacco use is more inclusive than the definitions used in any of the other surveys.
Despite the various factors contributing to the differences between the field test results and the
design assumptions, Table 27 does clearly show that the observed tobacco use rates are noticeably
higher in all the four groups defined by age and race (i.e., Black 18-24 years old, non-Black 18-24
years old, Black 25 years and older, and non-Black 25 years and older). Although no formal statistical
tests were conducted, the consistency and magnitude of these differences indicate that consideration
should be given to increasing the Phase 1 adult tobacco use rates assumed for the purpose of
designing the sampling for main study, and the increase may vary across the four groups. As part of
the final sample design process, the PATH Study team will further investigate this issue by
examining the 15 field test PSUs individually, including the PSU characteristics (e.g., rural vs. urban),
population distribution, and response rate; this investigation will help decrease the uncertainty in
determining the sample design parameters for the main study.
Table 28 shows the tobacco use misclassification rates (i.e., false positive rates and false negative
rates of the Phase 1 reporting.) The cases in the supplemental sample1 were included in the analyses
to give a larger sample size for estimating the false negative rates.. The Phase 1 information was
provided by the household informant, and Phase 2 information was based on self-reporting. False
positive means that the sampled person was classified as a tobacco user based on the Phase 1
household screener but as a non-user based on the Phase 2 self-report; false negative means that the
sampled person was classified as a tobacco non-user based on the Phase 1 household screener but as
a user based on the Phase 2 self-report. For the purpose of sampling, an adult is considered as being
misclassified with respect to tobacco use status if the category assumed for sampling at Phase 1 differs
from the category we assumed for sampling at Phase 2. That is, if the household informant did not
provide enough information to determine tobacco use status at Phase 1, the household member in
question was assumed to be a user for sampling at Phase 1. So cases where the tobacco use status
was unknown at Phase 1 contribute to the denominator of the false positive rate. In addition, the
Phase 2 self-reported tobacco use information was simply categorized into “confirmed user” versus
“not a confirmed user”. Hence, cases with unknown tobacco use status at Phase 2 can potentially
increase the false positive rate (if sampled as a user) or decrease the false negative rate (if sampled as

1

A supplemental sample was selected to increase the sample size for evaluating the false negative rate of tobacco use status reported by the household
respondent.

PATH Field Test Report

19

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

a non-user). Such an approach for computing the misclassification rates is most relevant from the
perspective of sample design and monitoring.2
The assumed false positive rate going into the field test was 5 percent for both age groups. The
assumed false negative rate was 5 percent for the 18-24 years old group and 2 percent for the 25
years and older group. The observed false positive rate for the 25 years and older group is 9.1
percentage points higher than the assumed rate. The observed false negative rates are also much
higher in both age groups (approximately 32.1 percent for the 18-24 years old group and
approximately 9.7 percent for the 25 years and older group) than the assumed rates. The
misclassification rates are also reported separately for the short screener questionnaire and long
screener questionnaire. For the 25 years and older group, the false positive rate was 6.9 percent for
the short screener and 20.5 percent for the long screener; the false negative rate was 14.1 percent for
the short screener and 4.3 percent for the short screener. A similar pattern is observed for the 18-to24 year-old group in terms of the difference between the two screener versions. The short screener
resulted in lower false positive rates and higher false negative rates than the long screener. At the
time of designing the field test, limited information was available for estimating the rates of tobacco
use reported by a household informant on behalf of all adult household members. The information
gathered through the field test can now be incorporated into adjusting these assumptions.
The design assumptions were based on the limited data available on this topic, which derives from
random digit dialing telephone surveys done in the 1990’s that addressed only cigarette smoking.
Indeed, while 5 percent may have been used as the design assumption for the false negative rate for
18 to 24 year-olds, household respondents who were not themselves the sampled 18-to-24 year-olds
might often be completely unaware that young adults were using tobacco, a speculation consistent
with the high false negative rate for this age group. Alternately, the “wide net” approach to the
screening could be hypothesized to make the false positive rates higher than the false negative rates,
a relationship seen for the older adults.

2

The later discussion of false positives and false negatives in the context of the field test experiment using long and short versions of the household
screener focuses on instrument validity, rather than operational sampling. As such, it excludes all cases with undetermined tobacco use status from
the analysis.

PATH Field Test Report

20

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

2.4.2

Weighted Frame Coverage Rates, Occupancy Rates, Response Rates, and
Effective Coverage Rates

As mentioned in the Introduction, the 15 field test PSUs were purposively selected. Within each
PSU, multi-stage probability samples of adults, youth, and/or shadow youth were selected. The firststage sampling units were groups of Census blocks (referred to as segments); the second-stage
sampling units were housing units within sampled segments; and the last-stage sampling units were
adults, youth, and/or shadow youth from eligible responding households. For the adult sample,
tobacco users, 18-to-24 year-olds, and Black persons were disproportionately oversampled. To
understand how the target population responded to the survey, weighted household-level and
person-level response rates were calculated by PSU. At the same time, the coverage rates of the
address frame (for housing unit sampling) provided by the external vendor were also evaluated.
To facilitate the weighted analyses, within-PSU base weights were computed separately for each PSU
that took into account the selection probabilities for sampling within the PSU. (PSU-level weights
were not constructed because the PSUs for the field test were purposively selected.) The main
reason for creating base weights was to produce estimates of population proportions and totals that
would account for the oversampling of particular subgroups in the field test. The initial step of
weighting was to construct household base weights for the sampled households using the inverse of
the probability of selection for that household. The household base weights are used for calculating
weighted household response rates.
For adults, two sets of base weights were developed. The person-level base weight as the result of
Phase 1 screening and sampling (referred to as “Phase 1 adult sampling base weight”) was calculated
as the product of household-level base weight and the inverse of the person-level selection
probability for Phase 1 sampling. The Phase 1 person-level selection probability depended on both
household composition and predetermined rates of selection that were set according to a person’s
age, race and tobacco use status reported by the household informant during the Phase 1 screener.
The sampled persons, selected at Phase 1, were administered the Phase 2 screener, and “Phase 1
adult sampling base weight” was used to calculate the weighted Phase 2 screener response rate.
The second set of adult base weights was developed for calculating the weighted extended interview
response rate. The persons who responded to the Phase 2 screener were subsampled for the
extended interview based on the self-reported age, race, and tobacco use information, and the
corresponding weight (referred to as “Phase 2 adult sampling base weight”) is the product of the
“Phase 1 adult sampling base weight” and the inverse of the subsampling factor.
PATH Field Test Report

21

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

For both youth and shadow youth, the person-level weight was computed by multiplying the
household base weight by the inverse of the selection probability for the youth or shadow youth that
was sampled.
Ideally, sample yields and weighted response rates could be examined separately by version of the
household screener administered and incentive amount. However, the limited size from the field test
does not support such refined analysis by PSU.
Table 29 shows the address frame coverage rates, housing unit eligibility rates, and weighted
household response rates for the 15 field test PSUs. The address frame had very good coverage in
general, with over 93 percent coverage rate for 14 of the PSUs. The only exception is PSU 304
where the coverage rate is 88.3 percent. The coverage of address frame in a particular area depends
on various factors. The main study will evaluate the coverage of the address frame for the sampled
segments.
The housing unit eligibility rates generally align with what was assumed for the sample design (88.6
percent). A few PSUs seem to be outliers, such as PSUs 309 (75.3 percent) and 311 (82.6 percent).
This could be due to the small sample sizes and the clustering effect of multi-stage sampling, which
made the estimates unstable.
The weighted household response rates were above 30 percent for all the 15 PSUs except PSU 313
(24.3 percent). In general, household response rates seem higher in rural PSUs and PSUs with highly
educated populations.
Table 30 provides information about person-level response rates for adults. A sampled adult at the
end of Phase 1 screener was expected to first respond to the Phase 2 screener, and then (if sampled)
the extended interview. The table shows the unweighted counts of sampled persons, the unweighted
counts of respondents, and weighted response rates for both the Phase 2 screener and extended
interview. Phase 2 screener response rates were above 50 percent for all the PSUs except PSU 313
(30.4 percent). Extended response rates were above 88 percent for all the PSUs except PSU 310
(77.6 percent). Once the target respondents had completed the Phase 2 screener (which included an
extensive list of items), they were likely to cooperate further. Again, the sample sizes are very small
for some PSUs, so the adult-level response rates should be interpreted with caution.

PATH Field Test Report

22

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

Tables 31 and 32 show the weighted response rates for youth ages 12-17 and the weighted
proportions of youth ages 9-11 whose parents/guardians agreed to be contacted during follow-up
waves. These rates vary significantly across PSUs. due to the very small sample sizes and should not
be over-interpreted.

3. Field Test Experiments

One purpose of the field test was to conduct an experiment to compare two versions of the Phase 1
screener. The longer version of the screener included 11 questions about tobacco use administered
to the household informant about each adult member of the household. The shorter version
included four questions about each adult’s tobacco use. The main hypotheses were that the longer
screener would result in greater accuracy but lower response rates. Accuracy, in this case, would be
measured by agreement with the results of the Phase 2 screener questions. The Phase 2 screener was
the “gold standard,” because the data were obtained directly from the sampled person and the
questions were self-administered via audio computer-assisted self-interviewing (ACASI) rather than
administered by the interviewer. The short screener, which consolidated the different tobacco
products into fewer items, was hypothesized to result in higher response rates but reduced
agreement with the Phase 2 screener.
The field test also compared three levels of incentive for the screener. Any possible impact of the
longer screener on response rates might be reduced or eliminated if sample households received a
monetary incentive. Hence, the field test experiment crossed the incentive variable with the screener
version variable. Households were offered $0, $5, or $10 as a promised incentive for completing the
household screener. They were informed of the incentive in the advance letter, as well as at the time
of the interview. Each sample dwelling unit was randomly assigned to one of the two screener
versions and one of the three incentive amounts. The randomization took place centrally and was
done before the dwelling unit was fielded.
The analysis of the experiment focuses on four main outcomes: (1) household screener response
rates; (2) rates of tobacco use according to the household screener; (3) response rates to the Phase 2
screener and extended interview; and (4) agreement between the household and Phase 2 screener
tobacco use classifications. The hypothesis regarding the response rates to the Phase 2 screener and
adult extended interview was the larger incentive for the household screener might create good will

PATH Field Test Report

23

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

that carried over to subsequent data collection activities. All the analyses reported here are
unweighted.

3.1

Field Test Screener Response Rates

Table 33 shows the Phase 1 (household) screener response rates. The first two columns show the
residential occupancy rates for the sampled addresses; the next two columns show household-level
screener response rates; and the final two, the person-level screener response rates (percentage of
enumerated persons for whom screener data were obtained) within households in which at least one
person was enumerated and screened.
As the table shows, neither experimental variable seemed to affect the number of addresses that the
interviewers classified as occupied residences. Regardless of which screener was used or which
incentive was offered, interviewers found about 90 percent of the addresses to be occupied dwelling
units. An effect of incentive level on this outcome was not expected and none was found.
The incentive did seem to produce an increase in the household-level screener response rates: 36.7
percent of households completed the screener when no incentive was offered versus 40.3 percent
with the $5 incentive and 43.0 percent with the $10 incentive. The incentive effect on screener
completion was marginally significant when the clustering of the observations by PSU was taken
into account (Rao-Scott χ2=5.82, df=2, p<.06). Neither the version of the screener nor the
interaction between the screener version and incentive affected the household-level response rate
significantly.
Within households where full screener data was collected about at least one person, both the
incentive level and screener version affected the proportion of household members for whom
screening data were obtained. In general, these proportions were high (90 percent or higher), but
both the incentive and the screener version had significant effects on this variable. The person-level
screener response rate (the rate at which full screener data was collected about each individual
household member) was highest in the no incentive group (97.3 percent), intermediate in the $5
incentive group (95.3 percent), and lowest in the $10 group (94.2). The person-level screener
response rate was very high with the short screener (99.8 percent) and considerably lower with the
long screener (91.3 percent). When the clustering by PSU is taken into account, both effects are
statistically significant (Rao-Scott χ2=6.89, df=2, p<.05, for the incentive effect; Rao-Scott χ2=77.5,
df=1, p<.001, for the effect of the screener version). The difference in person-level response rates
PATH Field Test Report

24

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

between the two versions of the screener reflects the larger number of items in the long screener.
When a person was missing data for one or more relevant screener items, he or she could not be
classified as a tobacco user or not and was counted as a nonrespondent in this analysis. The finding
regarding the incentive effect is difficult to interpret.

3.2

Household Screener Tobacco User Rates

Table 34 shows the rates of reported tobacco use for households and adults for each incentive
group and for the two versions of the screener. The table shows rates for all users (users within the
last 30 days of cigarettes, cigars, pipes, or dissolvables and lifetime users for all other products), as
well as separate figures for persons who reported smoking cigarettes, cigars, or pipes. The short
screener did not explicitly mention cigarillos or filtered cigars. Households and individuals screened
with the longer version of the Phase 1 screener were more likely to be classified as tobacco users
than those screened with the shorter screener. The effect of the screener version was significant for
both the proportion of households classified as having at least one tobacco user and for the
proportion of adults classified as tobacco users (Rao-Scott χ2=9.21, df=1, p<.05 for the proportion
of households and Rao-Scott χ2=8.72, df=2, p<.01 for the proportion of adults); these significance
tests take into account the clustering of the sample by PSU. The screener version also affected the
proportion of adults reported to smoke cigarettes, pipes, or cigars (Rao-Scott χ2=4.60, df=1, p<.05),
again taking the clustering of the sample by PSU into account. The longer screener clearly classified
more persons as tobacco users than the shorter version.

3.3

Response Rates to the Phase 2 Screener and Extended Interview

The incentive had a statistically significant effect on household screener response rates (see Section
3.1). Table 35 shows the response rates for subsequent stages of data collection, including the Phase
2 screener and the extended interview. Because of the small sample sizes at these later stages, this
section of the report focuses on descriptive statistics. The combined adult Phase 2 screener
completion rate and youth extended completion rate was lower for sampled persons in households
that completed the long version of the household screener than the short version (59.4 percent
versus 62.9 percent); combining the adult Phase 2 and youth interview rates creates a metric
characterizing the response to the initial approach seeking an interview with all persons sampled at
Phase 1. The person-level completion rate was also lower for the households getting the longer
version of the screener (86.1 percent versus 87.5 percent); the person-level completion rate
PATH Field Test Report

25

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

combines the adult extended completion rate for all adults sampled at Phase 2 with the youth
interview completion rate; this combination creates a metric characterizing the response in terms of
the ultimate data collection goal of fully completed questionnaires.
The incentive variable also seemed to have a positive effect on response rates to the Phase 2
screener and extended interview.3 The $10 household incentive produced the highest response rate
at the next stage of data collection (the Phase 2 screener for adults, the extended interview for
youth) with a response rate at this stage of 64.0 percent (versus 59.7 percent for the $5 incentive
group and 59.1 percent for the no incentive group). Given the small sample sizes, these findings are
not statistically significant, but they suggest that the benefits of the screener incentive may have
carried over to the next phase of data collection.
The incentive had no apparent effect on the proportion of sampled persons providing
biospecimens. For example, 51.2 percent of those in the no incentive and $10 incentive groups
provided urine samples; the corresponding figure is 53.2 percent for the $5 incentive. These
differences across incentive groups are not statistically significant.

3.4

Agreement between the Household and Phase 2 Screener
Classifications

The analysis also examined how often tobacco use as reported by the household informant in the
screener agreed with tobacco use reported by the sample adult himself or herself in the Phase 2
screener. In many cases (57.2 percent of the time), these were the same person.
Table 36 shows the overall rates of agreement regarding whether the sample person was a tobacco
user, by the experimental variables and by whether the screener data were self-reported or provided
by a proxy. The longer screener produced a somewhat higher rate of agreement than the short
screener (92.3 percent classified the same way in the household and Phase 2 screeners for the long
screener versus 88.0 percent for the short screener). This difference was marginally significant (RaoScott χ2=3.70, df=1, p<.06). The rate of agreement was also higher when the same person provided
the data in both screeners (92.1 versus 87.3 percent agreement); this difference was significant (RaoScott χ2=7.70, df=1, p<.01).

3

Regardless of the household incentive amount, all sampled adults and youth received the standard incentive amount for completing the Phase 2
screener/adult interview ($35) and the youth interview ($25).

PATH Field Test Report

26

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

The kappa statistics, displayed in the left panel of the table, follow the same pattern as the agreement
rates, with higher kappas when the household screener data were self-reports and when the longer
screener was used. Kappa adjusts the agreement statistics for chance levels of agreement.

PATH Field Test Report

27

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

Appendix A
Detailed Tables

PATH Field Test Report

28

Population Assessment of Tobacco and Health (PATH) Study (NIDA)
Table 1 PATH Field Test Household Screener CARI Recordings
Reviewed - Counts by Household Informant and Household
Characteristics

Characteristic
Total

Number reviewed
547

Household Informant Characteristic
Age
18 through 24 years old

56

25 through 44 years old

210

45 through 64 years old

200

65 years and older

81

Sex
Female

293

Male

254

Household Characteristic
Number of Household Members
1

112

2

186

3
4+

98
151

Number of Adults in Household
1

141

2

293

3
4+

72
41

Number of Youth in Household
0

353

1

84

2

74

3

27
9

4+

Population Assessment of Tobacco and Health (PATH) Study (NIDA)
Table 2 PATH Field Test Biospecimen Consent Rates by Type of Specimen by Respondent Characteristics
Consent to Provide Specimens (%)

Consent to
Genetic
2
Testing

Completed
Interviews (n)

Buccal

Urine

Blood

All Three

Any
Specimen

480

74.0

59.6

46.9

44.8

74.8

95.5

Tobacco User

236

80.9

70.3

54.7

52.1

82.6

95.4

Non-user

244

67.2

49.2

39.3

37.7

67.2

95.7

Alcohol

345

75.9

62.3

47.0

44.6

76.8

95.8

Marijuana

138

83.3

68.1

47.8

47.1

84.1

97.4

18

88.9

83.3

66.7

66.7

88.9

100.0

Overall
Tobacco Status

Drug Use

Cocaine/Crack
Stimulants

9

88.9

88.9

66.7

66.7

88.9

100.0

14

92.9

92.9

78.6

78.6

92.9

100.0

118

66.9

52.5

47.5

46.6

67.8

93.8

18-24

160

80.6

62.5

50.6

48.1

80.6

95.3

25-44

173

72.3

59.0

44.5

42.8

72.8

95.2

45-64

116

70.7

60.3

48.3

45.7

73.3

95.3

65+
1
Education

31

61.3

45.2

35.5

35.5

61.3

100.0

Heroin and Other Drugs
No Drug Use
Age

< HS degree

65

84.6

72.3

63.1

60.0

86.2

96.4

HS degree/< 4 yr college degree

303

73.6

59.4

45.9

43.9

74.3

96.4

4 yr college degree +

110

69.1

52.7

40.0

38.2

70.0

92.2

Female

228

76.8

62.3

50.4

48.2

77.6

96.6

Male

252

71.4

57.1

43.7

41.7

72.2

94.5

Black

146

73.3

61.0

42.5

41.8

73.3

92.5

White

303

74.6

60.4

49.5

46.9

75.9

96.5

Other

31

71.0

45.2

41.9

38.7

71.0

100.0

61

77.0

55.7

45.9

42.6

77.0

97.9

419

73.5

60.1

47.0

45.1

74.5

95.2

Gender

Race

Ethnicity
Hispanic
Non-Hispanic

Note: Table covers respondents who completed the adult interview.
1
2

The sum of counts for this category does not equal the overall total due to missing values.

Denominator is the number of subjects who consented to any biospecimen collection

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

Table 3 PATH Field Test Blood Tube Collection Rates (as percent of blood collected) by Respondent Characteristics

93.6

%

171

n

91.4

%

171

n

91.4

%

171

n

91.4

%

LV01

Blood Tube

n

RD02

175

RD01
%

BT01

94.7

PX01
(PaxGene)

n

LV02
(Plasma-EDTA 2)

177

(Plasma-EDTA 1)

%

(Serum 2)

96.3

(Serum 1)

n

(Plasma-Citrate)

180

Blood
Collected
(n)
187

91.5
91.4

Blood
Consent
Given (n)
225

97
74

Overall

90.6
92.6

Tobacco Status

96
75

90.6
95.2
92.2
66.7

90.6
92.6

58
60
47
6

96
75

90.6
95.2
92.2
66.7

91.5
96.3

58
60
47
6

84.9
93.8
90.2

97
78

93.8
95.2
88.2
66.7

28
105
37

93.4

96.9

60
60
45
6

87.9
92.0
92.7

96.3

96.8

93.8
96.8
92.2
77.8

29
103
38

99

62

60
61
47
7

84.9
93.8
90.2

93.7
89.1

78

61

92.2

28
105
37

89
82

95.3

96.9

77.8

93.9
93.8
92.7

92.6
90.2

97.5

96.8

7

47

93.9

31
105
38

88
83

101
79

96.1

95.5

92.6
90.2

81

88.9

31

92.7

88
83

106

64

97.0

38

107

94.7
92.4

96

63

95.5

90
85

129

81

97.6
95.8
96.7

95.8
93.5

Tobacco user

77

62
61
49
8
32
107
40
91
89

91
86

Non-user
18-24

9

51

Age
25-44
56

33

11
41

45-64

< HS degree
41

112

65+
1
Education

44

95
92

139

115
110

HS degree/< 4 yr college degree
4 yr college degree +
Gender
Female
Male

Note: Table covers respondents who completed the adult interview and from whom blood was successfully collected.
The sum of counts for this category does not equal the overall total due to missing values.
1

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

Table 4 PATH Field Test Amount of Urine Collected by Respondent Characteristics (as percent of urine collections)

244

236

480

Completed
Interviews
(n)

100

120

166

286

Count

60.3

59.0

62.5

49.2

70.3

59.6

%

11

60

83

80

97

137

234

Count

80.6

80.9

78.6

85.7

81.4

80.0

80.8

82.5

81.8

%

9

31

8

6

13

11

18

19

29

48

Count

18.0

21.4

21.1

54.6

21.7

13.3

22.5

19.6

21.2

20.5

%

41

114

30

5

47

72

62

78

108

186

Count

82.0

78.6

79.0

45.5

78.3

86.8

77.5

80.4

78.8

79.5

%

Mass Urine Collected

Tobacco User

102

45.2

38

86.2

Age

142

57.1

62.3

31
17

27.0
14.3

84
102

73.0
85.7

>= 42.7 grams

Non-user
160
70

72.3

50

145

144

< 42.7 grams

18-24
173
14

59.4

Urine Collected

25-44
31

116

47

52.7

Urine Consent Given

45-64

65

58

180

Overall

65+
Education1
303

Tobacco Status

< HS degree
110
228

81.0
82.6

HS degree/< 4 yr college degree

252

115
119

4 yr college degree +
Female

Gender
Male

Note: Table covers respondents who completed the adult interview.
The sum of counts for this category does not equal the overall total due to missing values.
1

Population Assessment of Tobacco and Health (PATH) Study (NIDA)
Table 5 PATH Field Test Buccal Cell Collection Rates (as percent of consent given) by Respondent Characteristics
Buccal Consent Given

Buccal Specimen Collected

n

n

%

355

349

98.3

Tobacco User

191

187

97.9

Non-user

164

162

98.8

18-24

129

126

97.7

25-44

125

123

98.4

45-64

82

81

98.8

65+
Education1

19

19

100.0

Overall
Tobacco Status

Age

< HS degree
HS degree/< 4 yr college degree
4 yr college degree +

55

54

98.2

223

219

98.2

76

75

98.7

175
180

172
177

98.3
98.3

Gender
Female
Male

Note: Table covers respondents who completed the adult interview and consented to buccal cell collection.
1
The sum of counts for this category does not equal the overall total due to missing values.

Population Assessment of Tobacco and Health (PATH) Study (NIDA)
Table 6 PATH Field Test Hooper-Holmes Biospecimen Collection (as percent of consented) by Specimen Type and by Respondent
Characteristics
Consented for Blood Only
Visit
Blood
Conducted
Collected
n
%
n
%

n
Consented
Overall

3

3

100.0

2

Tobacco User

2

2

100.0

Non-user

1

1

100.0

18-24

1

1

100.0

25-44

1

1

45-64

1

1

n
Consented

Consented for Blood and Urine
Visit
Blood
Urine
Conducted
Collected
Collected
n
%
n
%
n
%

66.7

222

213

96.0

185 83.3 185

83.3

1

50.0

127

125

98.4

105 82.7 106

83.5

1

100.0

95

88

92.6

80

84.2

79

83.2

80

75

93.8

64

80.0

64

80.0

100.0

1

100.0

76

74

97.4

62

81.6

62

81.6

100.0

1

100.0

55

53

96.4

50

90.9

49

89.1

11

11

100.0

9

81.8

10

90.9

33

80.5

32

78.1

Tobacco Status

Age

65+
1

Education

< HS degree

0

HS degree/< 4 yr college degree

3

4 yr college degree +
Gender
Female
Male

3

100.0

2

66.7

0
2
1

40

97.6

129

94.9

44

43

97.7

41

93.2

41

93.2

110 80.9 111

81.6

2

100.0

1

50.0

113

107

94.7

94

83.2

92

81.4

1

100.0

1

100.0

109

106

97.3

91

83.5

93

85.3

Note: Table panel covers respondents
consenting to biospecimen collection
requiring Hooper Holmes visit - blood
consent only.
1

41
136

Note: Table panel covers respondents consenting to
biospecimen collection requiring Hooper Holmes visit blood and urine consent.

The sum of counts for this category does not equal the overall total due to missing values.

Population Assessment of Tobacco and Health (PATH) Study (NIDA)
Table 7 PATH Field Test Time (days) from Blood Consent to Time of Blood Collection,
by PSU
Blood
Consent
(n)

Collected (n)

Min

Max

Median

Mean

225

187

3.0

33.0

7.0

8.1

301
302

17
9

14
6

4.0

13.0

7.0

7.0

8.0

10.0

9.5

9.2

303

18

16

4.0

15.0

8.0

8.3

304

24

17

4.0

17.0

8.0

8.7

305

10

10

4.0

17.0

6.0

7.4

306

24

23

4.0

23.0

6.0

7.2

307

21

21

4.0

15.0

6.0

7.7

308

17

14

4.0

33.0

7.5

10.9

309

21

18

4.0

15.0

8.5

8.9

310

10

6

4.0

10.0

5.5

6.3

311

24

15

3.0

12.0

8.5

8.1

312

9

7

5.0

18.0

7.0

8.1

313

5

5

5.0

13.0

7.0

8.2

314

10

10

4.0

10.0

6.5

6.8

315

6

5

6.0

15.0

10.0

10.0

Overall

Blood

Days from Consent to Blood Collected

PSU

Note: Table covers consented and completed blood collection

Population Assessment of Tobacco and Health (PATH) Study (NIDA)
Table 8 PATH Field Test Specimen Time in Transit (hours) by Shipment Type by Packing Day
Segment
Shipment Type/Packing
Day Segment
Blood + Urine
Weekday <6pm
Weekday >= 6pm
Saturday
Sunday
Blood Only
Weekday <6pm
Weekday >= 6pm
Saturday
Sunday
Buccal + Urine
Weekday <6pm
Weekday >= 6pm
Saturday
Sunday
Buccal Only
Weekday <6pm
Weekday >= 6pm
Saturday
Sunday
Urine Only
Weekday <6pm
Weekday >= 6pm
Saturday
Sunday
Interviewer Shipments
Weekday <6pm
Weekday >= 6pm
Saturday
Sunday
Phlebotomist Shipments
Weekday <6pm
Weekday >= 6pm
Saturday
Sunday

n

Minimum

Maximum

Median

Mean

111
20
16
12

15.5
14.7
45.2
38.8

145.1
138.3
141.1
56.3

24.0
36.8
83.8
47.4

40.3
47.0
92.6
46.5

5
1
1

19.4
84.9
49.9

119.3
84.9
49.9

69.4
84.9
49.9

59.6
84.9
49.9

25
12
14
6

17.7
35.7
40.0
38.7

142.8
113.8
70.6
43.9

21.5
37.9
64.9
41.4

49.2
43.9
58.5
41.5

155
70
46
15

14.7
14.1
42.5
35.8

164.2
160.3
119.5
89.5

21.5
38.6
64.6
41.7

33.8
47.6
65.6
48.1

7
1

19.7
39.7

69.8
39.7

42.9
39.7

41.0
39.7

1

46.4

46.4

46.4

46.4

187
83
60
22

14.7
14.1
40.0
35.8

164.2
160.3
119.5
89.5

21.8
38.3
64.7
41.8

36.2
47.0
63.9
46.2

116
21
17
12

15.5
14.7
45.2
38.8

145.1
138.3
141.1
56.3

24.0
37.1
73.5
47.4

41.1
48.8
90.1
46.5

0

0

Note: Table covers specimen shipments received at repository and for which collection
date/time is available

Population Assessment of Tobacco and Health (PATH) Study (NIDA)
Table 9 PATH Field Test Time (hours) From Specimen Receipt to Start of Processing at the
Repository, by Specimen Type
Specimen Type

n
n
Received Processed

Time in Hours
Minimum

Maximum

Median

Mean

Blue Top 1

186

185

1.0

6.4

4.0

4.1

Lavender 1

180

176

0.8

5.9

3.5

3.3

Lavender 2

176
182

0.8

5.9

3.5

3.3

Red Top 1

179
183

1.0

6.2

4.1

4.0

Red Top 2

181

180

1.0

6.2

4.1

4.0

Urine

246

242

1.0

8.0

3.1

3.4

Note: Table covers specimens processed at the repository following receipt.

Population Assessment of Tobacco and Health (PATH) Study (NIDA)
Table 10 PATH Field Test Number of Aliquots Created per Parent, by Specimen Processing Protocol and Vial Size
Protocol

Collected
(n)

Aliquot Type
Buffy Coat

Plasma (EDTA)

179

Plasma-EDTA

Serum

Urine1 >= 42.7g

Urine2 <42.7g >=33.8g

Urine3 <33.8g

185
185

197

14

37

Number of aliquots per parent
Minimum

Maximum

Median

Mean

2

1.0

2.0

2.0

2.0

Expected
2

0.75

4.0

10.0

7.0

7.2

10

2

1.0

6.0

6.0

5.7

6

10

2.0

2.0

2.0

2.0

2

Plasma-Citrate

2

1.0

1.0

1.0

1.0

1

RBCs+WBCs

2

1.0

1.0

1.0

1.0

1

0.75

1.0

5.0

3.0

3.1

5

2

1.0

8.0

7.0

6.7

8

2

3.0

14.0

14.0

13.7

14
3

RBCs
Plasma (citrate)

Vial Size

Serum

Urine

Urine

Urine

6

2.0

3.0

3.0

3.0

10

1.0

2.0

2.0

2.0

2

2

8.0

10.0

8.0

8.6

14
3

6

2.0

3.0

3.0

2.9

10

2.0

2.0

2.0

2.0

2

2

1.0

12.0

8.0

7.1

14

6
10

1.0
1.0

3.0
2.0

3.0
2.0

2.2
1.8

3
2

Note: Table covers all specimens processed at repository.
Note: Expected column indicates the expected number of aliquots to be created as defined by the processing protocols

Population Assessment of Tobacco and Health (PATH) Study (NIDA)
Table 12 PATH Field Test Household Screener Timings (minutes), by Version and Household Size
n

Minimum

Maximum

Median

Mean

568

2.4

69.4

12.2

14.0

8.3

1

115

2.4

44.2

6.6

8.0

5.6

2

168

4.3

39.5

10.8

12.4

6.0

3

103

5.3

38.3

13.3

14.4

5.4

4

105

4.1

69.4

15.6

17.0

8.9

77

7.6

59.4

19.1

22.0

10.1

598

2.8

77.0

11.3

12.7

7.5

1

123

2.8

18.0

6.3

7.1

2.9

2

199

2.8

43.7

9.3

11.2

6.3

3

110

5.8

34.9

13.5

14.3

5.6

4

100

5.6

77.0

14.3

16.5

10.3

66

8.8

39.2

18.0

18.8

5.9

Long version

Std Dev

HH Size

5+
Short version
HH Size

5+

Note: Table covers all completed household screeners, excluding those with missing data and extreme
outliers resulting from in-field computer problems.

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

Table 13 PATH Field Test Adult Interview Timings (minutes), by Current Tobacco Use Status by Instrument Section

1.4
42.0

4.8

1.7
19.7

2.2

1.1

Std Dev
2.9

23.1

1.4

0.6

Minimum
2.3

150.3

15.9

7.6

48.1

4.8

1.5

52.6

5.2

1.8

19.6

2.4

1.1

Std Dev
3.0

14.2

0.8

0.4

Minimum
2.0

88.5

11.8

7.9

27.8

3.9

1.3

31.7

4.3

1.6

13.4

2.0

1.0

Hookah

Pipe

Cigar

E-cigarette

Cigarette

0.5

0.7

0.5

0.9

0.7

0.7

.

16.3

10.0

10.7

23.0

27.2

28.7

1.4

.

1.2

1.7

1.2

3.6

1.5

5.4

1.7

.

2.2

2.2

1.8

4.8

2.5

5.8

3.3

1.1

.

2.9

1.4

1.8

3.8

3.1

4.7

1.5

0.2

.

0.5

0.8

0.5

1.1

0.7

0.7

22.3

6.2

.

16.3

10.0

10.7

23.0

27.2

28.7

1.4

4.2

1.4

.

1.3

2.1

1.3

4.5

1.7

7.8

1.5

5.3

1.7

.

2.5

2.6

1.8

5.7

2.6

8.2

0.7

3.0

1.1

.

3.2

1.7

1.7

4.0

3.3

4.5

0.1

0.1

.

.

0.6

0.7

0.6

0.9

0.9

0.7

21.9

9.6

12.8

.

.

8.1

5.9

10.4

20.8

9.1

16.9

3.6

0.6

0.3

.

.

1.0

1.5

1.2

2.6

1.2

1.7

4.3

0.8

1.6

.

.

1.5

1.9

1.8

3.4

1.9

2.6

0.8

2.4

0.7

2.5

.

.

1.5

1.0

1.9

2.9

1.9

2.4

Std Dev
2.7

4.4

Current Non-Users (n = 234)
Maximum
Median
Mean
19.7
5.4
5.9

7.9
38.0

Current Tobacco Users (n = 227)
Maximum
Median
Mean
17.9
6.4
7.0

15.9

All Respondents (n = 461)
Maximum
Median
Mean
19.7
5.9
6.4

0.4
150.3

Minimum
2.0

0.8

Phase 2
Intro Demographics
14.2

Section
Tobacco Use
Extended (Excluding Contact)*

Smokeless Tobacco
.
6.2
4.6

4.3

1.6

Section

Dissolvable Tobacco
0.2
3.9

0.2

3.1

0.8

Polyuse
22.3

0.8

4.6

1.8

0.1

1.1

3.9

1.6

Nicotine Dependence
0.9

34.9

8.7

9.6

1.6

0.6

0.1

2.7

0.6

Packaging and Health Warnings

4.4

1.6

3.7

1.4

34.9

5.3

1.6

0.6

0.3

Product Regulation

0.7

1.7

1.7

1.5

1.5

5.0

8.7

0.5

0.6

1.1

Media Use

1.8

0.9

1.6

1.7

11.9

1.5

0.5

11.9

6.7

0.5

0.5

Second Hand Smoke Exposure

0.4

0.2

2.2

0.4

0.1

0.6

0.4

0.6

0.8

4.2

0.7

0.1

4.2

Peer and Family Influences

0.8

113.4

13.5

2.3

12.1

8.0

55.1

14.1

1.6

4.4

12.5

3.1

9.0

113.4

5.3

14.7

2.3

15.0

12.7

Health Effects Outcome

0.6

1.8

1.5

1.4

0.8

3.0

1.8

1.2

1.6

2.6

3.2

7.2

1.4

23.3

2.8

0.4

7.2

1.0

23.3

0.8

0.4

1.7

1.0

59.5

1.8

Industry Advertising/Promo

54.6

3.5

Additional Demographics

154.1

2.6

26.2

15.5

21.7

2.6
6.9

37.6

48.4

24.3

34.3

43.5

7.8
12.4

101.0

154.1

74.9

16.5

16.5

7.4
10.7

21.6

Entire Adult Instrument*

69.6

5.7

14.8
52.6

8.0
163.2

7.8

2.6
2.8

14.3
34.3

3.6

Contact

2.6
6.3

19.5

24.6

11.7

7.9
12.0

52.8

63.7

10.2

7.6
10.5

48.3

58.8

36.4

14.8
52.6

126.5

163.2

3.5

2.6
2.8

22.2

22.2

Consent and Tutorial
Total Interview Process*

Note: Table covers all completed adult interviews, excluding those with missing data and extreme outliers resulting from in-field computer problems.
*Because each respondent may not be routed through each section, the summary means will not necessarily equal the sum of the means of the various sections.

Population Assessment of Tobacco and Health (PATH) Study (NIDA)
Table 14 PATH Field Test Phase 2 Screener Timings (minutes), by Ever/Never Tobacco Use, and Number of
Products Ever Used
n

Minimum

Maximum

Median

Mean

Std Dev

43

2.2

10.6

4.4

5.3

2.4

418

2.0

19.7

6.0

6.6

2.9

1

91

2.0

14.1

5.5

5.9

2.6

2

82

2.3

12.6

5.2

5.8

2.6

3

63

2.1

14.8

5.5

6.3

3.0

182

3.3

19.7

6.8

7.3

3.0

All Never-Users
All Ever Users
# Products

4+

Note: Table covers all completed Phase 2 screeners, excluding those with missing data and extreme outliers
resulting from in-field computer problems.
Never user is any respondent who has never used any tobacco product.
Ever user is any respondent who has ever used any tobacco product.

Population Assessment of Tobacco and Health (PATH) Study (NIDA)
Table 15 PATH Field Test Phase 2 Screener Timings (minutes), by Ever/ Never Tobacco Use by Age
n

Minimum

Maximum

Median

Mean

Std Dev

43

2.2

10.6

4.4

5.3

2.4

18-24

20

2.2

10.4

3.8

4.4

2.1

25-44

11

3.3

10.6

4.4

5.0

2.3

45-64

9

4.2

10.2

7.2

7.4

2.0

65+

3

4.3

9.3

5.5

6.3

2.6

418

2.0

19.7

6.0

6.6

2.9

18-24

131

2.1

16.7

5.0

5.6

2.4

25-44

157

2.0

14.6

5.5

6.1

2.5

45-64

104

2.7

19.7

7.3

7.8

3.1

26

4.0

17.9

7.8

8.9

3.8

All Never-Users
Age

All Ever Users
Age

65+

Note: Table covers all completed Phase 2 screeners, excluding those with missing data and extreme outliers
resulting from in-field computer problems.
Never user is any respondent who has never used any tobacco product.
Ever user is any respondent who has ever used any tobacco product.

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

Median
48.1
46.4

Mean
52.0

20.6

19.1

Std Dev
21.8

58

91

65

n
227

54.9

34.1

26.4

29.6

Minimum
26.4

132.2

130.6

126.2

154.3

Maximum
154.3

80.2

65.2

56.7

53.8

Median
58.1

84.7

68.1

61.1

56.6

Mean
62.9

21.3

22.4

19.2

20.1

Std Dev
21.4

16

55

77

86

234

29.6

24.7

16.7

18.4

Minimum
16.7

93.7

107.2

101.6

84.7

Maximum
107.2

47.5

43.4

33.8

35.6

Median
37.9

51.7

49.2

36.6

38.7

Mean
41.4

16.9

18.8

12.9

14.1

Std Dev
16.1

Current Non-Users

Maximum
154.3
44.1
49.8

13

Current Tobacco Users

Minimum
16.7
154.3
46.2

22.7

All Respondents

Table 16 PATH Field Test Entire Instrument Timings (minutes), by Current Tobacco Use Status by Age, Education, and Gender
n
461
18.4
126.2

25.0

n

151
16.7
58.9

Overall
18-24
168
66.5

Age
25-44
54.8

21.9

64.8

14.5

130.6

51.4

132.2

44.6

24.7

107.2

13.3

29.6

18.4

40.1

29

37

38.5

113

18.9

37.4

45-64
65+
1
Education

67.4

34.5

64.5

93.7

101.6

127.5

60.1

16.7

29.6

127.5

21.8

67

18.4

81

116

21.3

104

23.1

61.7

< HS degree

17.5

13.8

61.4

17.8

57.9

51.5

38.7

55.2
43.2

44.0

53.6

47.2

35.2

93.1

39.1

40.1

154.3

93.7

154.3

88.1

26.4

16.7

107.2

30.5

21.8

18.0

26

248

16.7
117

117

132

107

125

22.2

HS degree/< 4 yr college degree

20.3

16.6

4 yr college degree +

22.8

Gender
49.2

19.8
54.5

22.6

45.8

61.3
51.0

64.3

130.6

58.3

154.3

58.0

18.0

130.6

16.7

154.3

219

26.4

242

30.5

Female

102

Male

The sum of counts for this category do not sum to the overall total due to missing values.

Note: Table covers all completed adult interviews, excluding those with missing data and extreme outliers resulting from in-field computer problems.
1

Population Assessment of Tobacco and Health (PATH) Study (NIDA)
Table 17 PATH Field Test Entire Instrument and Nicotine Dependence Section Timings (minutes) for Current
Tobacco Users by Number of Products Currently Used
n

Minimum

Maximum

Median

Mean

Std Dev

227

26.4

154.3

58.1

62.9

21.4

189
26
8
4

26.4

154.3

56.4

61.7

21.1

29.6

122.0

64.3

67.1

24.9

51.3
47.1

87.4
112.9

69.2
74.1

69.8
77.1

11.0
27.2

227

1.5

22.3

4.2

5.3

3.0

189
26
8
4

1.9

22.3

4.1

5.2

3.0

1.5

12.5

4.7

5.1

2.4

2.3
2.4

7.6
17.5

5.6
6.1

5.4
8.0

1.8
7.0

Entire Instrument
All Users
# Products
1
2
3
4+
Nicotine Dependence
Section
All Users
# Products
1
2
3
4+

Note: Table covers all completed adult interviews with tobacco users, excluding those with missing data and
extreme outliers resulting from in-field computer problems.

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

Table 18 PATH Field Test Youth Interview Timings (minutes), by Current Tobacco Use by Instrument Section

E-cigarette

Cigarette

Intro Demographics

0.2

0.4

0.1

0.9

0.4

5.8

3.8

6.8

18.5

10.1

14.4

5.1

0.4

0.4

1.5

1.3

1.5

3.3

2.0

2.0

0.9

Median
30.6

3.2

0.1

0.5

0.4

1.5

1.4

1.7

3.8

2.1

2.4

1.1

Mean
32.2

1.3

0.1

0.7

0.4

0.9

0.9

0.9

2.3

1.4

1.6

0.7

Std Dev
10.0

2.5

0.1

0.2

0.1

0.2

0.8

1.2

2.7

1.9

1.5

0.4

7.6

0.2

2.7

1.9

5.8

3.8

2.9

18.5

10.1

14.4

1.8

4.0

0.1

0.4

0.3

1.2

1.8

1.3

8.0

3.5

4.8

1.2

Median
47.5

1.6

4.1

0.1

0.9

0.5

1.7

2.0

1.6

7.8

4.4

5.7

1.1

Mean
49.7

0.2

0.5

0.5

1.6

0.1

0.9

0.6

1.7

1.1

0.6

4.8

2.8

3.7

0.5

Std Dev
13.2

0.1

0.3

0.2

.

0.8

1.1

.

0.1

0.1

0.1

0.2

0.2

0.4

0.1

0.9

0.4

8.1

2.8

2.3

3.7

1.3

.

1.6

8.3

.

6.7

3.5

4.6

3.7

6.8

9.5

6.9

6.4

5.1

1.8

0.2

2.8

0.8

0.6

1.2

0.4

.

1.4

2.9

.

0.4

0.4

1.6

1.2

1.6

3.2

1.9

2.0

0.9

Median
29.8

0.2

2.2

2.0

0.2

3.0

0.9

0.7

1.3

0.4

.

1.3

3.1

.

0.5

0.4

1.5

1.3

1.7

3.4

1.9

2.1

1.1

Mean
30.9

0.1

1.2

0.8

0.1

1.2

0.4

0.6

0.6

0.2

.

0.3

1.2

.

0.7

0.4

0.9

0.8

0.9

1.6

1.1

0.9

0.7

Std Dev
8.3

Current Non-Users ( n = 114 )

Cigar
0.2
3.5
0.1

1.7

0.8

0.5

0.4

0.7

1.8

Current Tobacco Users (n = 9 )

Pipe
0.1
6.7
2.9

2.4

0.7

0.5

1.0

0.6

5.4

0.2

All Respondents (n = 123 )

Hookah
0.1
0.2

0.8

1.7

0.5

1.3

0.5

0.0

6.7

Packaging/Health Warnings

Secondhand Smoke

Modified Risk Products

Cessation

Dependence

0.4

0.1

0.3

0.2

0.2

2.8

3.8

3.7

1.3

2.8

0.8

0.7

1.2

3.0

0.9

0.8

1.2

0.4

2.1

0.3

Marketing
8.1

0.1

0.6

0.1

Media Use

0.2

0.7

0.2

0.0

1.5

Accessibility

2.9

1.8

1.2

5.4

0.8

0.5

2.0

Psychosocial

Minimum Maximum
56.9
11.0

Smokeless Tobacco
0.1
8.3

0.4

0.2

0.7

1.3

2.4

0.7

0.5

1.3

2.3

3.9
0.2

0.1

1.9

3.8
0.2

0.2

7.0

7.0
1.3

0.2

0.5

1.6

0.1

0.1

Substance Use

1.1

1.3

Health Outcomes

0.5
0.7

0.1

Minimum Maximum
73.1
30.5

Dissolvable Tobacco
0.1
1.4

0.5

0.2

2.4

2.1

1.0

0.1

0.5

0.5
0.7

Minimum Maximum
73.1
11.0

Bidi/Kretek
1.1
1.4

0.8

0.2

0.6

3.8

0.8

3.0

0.7

0.1

0.9
2.0

Entire Extended

First Tobacco Product
2.4
0.7

0.4

0.6

0.7

2.2

2.9

0.2

1.5

0.8
1.9

Section

Reasons to Use
0.8
1.7
0.4

1.3

0.8

0.7

4.2

0.2

2.0

0.0

4.0
5.0

0.9
2.0

0.7

0.3

0.8
1.9

0.6
0.6

4.0
5.0

1.1
1.9
0.3

1.0
2.1
0.7

2.2
2.6

Ad Exposure

0.5

Demographics

Note: Table covers all completed youth interviews, excluding those with missing data and extreme outliers resulting from in-field computer problems.
Because each respondent may not be routed through each section, the entire extended mean will not necessarily equal the sum of the means of sections.

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

Current Tobacco Users

Current Non-Users

9.1

8.4

7.5

Std Dev

31.4

30.9

30.3

Mean

29.8

29.8

29.8

Median

53.7

56.9

56.9

Maximum

11.0

11.0

13.4

Minimum

58

n

56

114
.

13.2

14.1

Std Dev

48.6

49.7

49.8

Mean

48.6

47.5

47.4

Median

48.6

73.1

73.1

Maximum

48.6

30.5

30.5

Minimum

1

n

8

9
9.3

10.0

10.7

Std Dev

31.7

32.2

32.7

Mean

29.9

30.6

30.8

Median

53.7

73.1

73.1

Maximum

11.0

11.0

13.4

Minimum

59

n

64

123

All Respondents

Table 19 PATH Field Test Youth Interview Timings - Entire Instrument (minutes), by Current Tobacco Use and Age

Overall
12-14

Age
15-17

Note: Table covers all completed youth interviews, excluding those with missing data and extreme outliers resulting from in-field computer problems.

Population Assessment of Tobacco and Health (PATH) Study (NIDA)
Table 20 PATH Field Test Youth Interview Timings - Entire Instrument (minutes) for Current Tobacco Users by Number of
Products Currently Used
n

Minimum

Maximum

Median

Mean

Std Dev

9

30.5

73.1

47.5

49.7

13.2

1

7
1

30.5
47.5

73.1
47.5

47.3
47.5

50.1
47.5

15.2

2
3

1
0

48.6

48.6

48.6

48.6

All Users
# Products

4+

Note: Table covers all completed youth interviews, excluding those with missing data and extreme outliers resulting
from in-field computer problems.

Population Assessment of Tobacco and Health (PATH) Study (NIDA)
Table21 PATH Field Test Parent Interview Timings (minutes)

All interviews

n

Minimum

Maximum

Median

Mean

Std Dev

122

3.0

18.2

7.1

7.8

2.5

Note: Table covers all completed parent interviews, excluding those with missing data and extreme outliers
resulting from in-field computer problems.

Population Assessment of Tobacco and Health (PATH) Study (NIDA)
Table 22 PATH Field Test Number of Hours per Case by Case Characteristics
Characteristics
Overall

n

Minimum

Maximum

Median

Mean

1166

0.0

9.9

0.6

1.1

541
475
125
22
3

0.0
0.2
0.9
2.1
4.9

4.2
6.9
6.9
9.9
8.3

0.2
1.4
3.0
3.4
5.8

0.2
1.5
3.0
3.9
6.3

541

0.0

4.2

0.2

0.2

450
25

0.3
0.2

6.9
2.1

1.4
1.2

1.5
1.2

69
46
10

1.0
0.9
1.5

6.9
5.0
2.5

3.4
2.5
2.1

3.4
2.6
2.0

14
11
0

2.3
2.1

8.3
9.9

4.9
2.8

4.7
3.5

Number of Sampled Persons
0
1
2
3
>4
Combination of Sampled Persons
0 Sampled Persons
1 Sampled Person
1 adult, 0 youth
0 adults, 1 youth
2 Sampled Persons
2 adults, 0 youth
1 adult, 1 youth
0 adults, 2 youth
3+ Sampled Persons
2 adults, >1 youth
1 adult, >2 youth
0 adults, >3 youth

Note: Case is defined as a dwelling unit with a completed household screener. Hours cover those logged by the
field interviewer's computer whenever the interviewer was interacting with persons at the dwelling unit.
For combinations of sampled persons, sampled refers to the sampling from the Phase 1 screener.

Population Assessment of Tobacco and Health (PATH) Study (NIDA)
Table 23 PATH Field Test Number of Hours per Case by Numbers of Interviews per Case and Specimen
Collection Outcomes
n

Minimum

Maximum

Median

Mean

1166

0.0

9.9

0.6

1.1

718
333

0.0

6.9

0.3

0.4

1

0.2

4.6

1.9

1.9

2

100

1.6

6.9

3.4

3.3

3

13

2.3

9.9

4.3

4.8

4

2

5.8

8.3

7.0

7.0

5

0

6

0

>6

0

718

0.0

6.9

0.3

0.4

1 adult, 0 youth

295

0.7

4.6

1.9

2.0

0 adult, 1 youth

38

0.2

2.6

1.5

1.5

2 adults, 0 youth

54

1.6

6.9

3.8

3.7

1 adult, 1 youth

32

2.1

5.0

3.1

3.2

0 adult, 2 youth

14

1.7

2.9

2.3

2.3

2 adults, >1 youth

9

3.9

8.3

5.2

5.4

1 adult, >2 youth

6

2.3

9.9

4.1

4.7

0 adults, >3 youth

0

Overall
Number of Extended
Interviews Completed
0

Combination of
Extended Interviews
Completed
0 Completed
1 Completed

2 Completed

3+ Completed

Number of Adults
Providing Buccal Cells
and/or Urine Sample to
Interviewer
0

869

0.0

4.4

0.3

0.6

1

245

1.0

9.9

2.2

2.4

2

52

2.3

8.3

4.0

4.2

Note: Case is defined as a dwelling unit with a completed household screener. Hours cover those logged by
the field interviewer's computer whenever the interviewer was interacting with persons at the dwelling
unit.
Times include parent interview(s) when also completed.

Population Assessment of Tobacco and Health (PATH) Study (NIDA)
Table 24 PATH Field Test Unweighted Response Rate by Instrument and Various Characteristics
Eligible* (n)

Complete (n)

Unweighted
Response Rate
(%)

Household Screener
2944

1170

39.7

Long Version

1400

562

40.1

Short Version

1544

608

39.4

1152

698

60.6

Sampled in Household Screener as User

579

382

66.0

Sampled in Household Screener as Non-user

573

316

55.1

501

480

95.8

Sampled in Phase 2 Screener as User

390

378

96.9

Sampled in Phase 2 Screener as Non-User

111

102

91.9

195

128

65.6

12-14 Year Olds (age from Household Screener)

92

62

67.4

15-17 Year Olds (age from Household Screener)

103

66

64.1

195

122

62.6

12-14 Year Olds (age from Household Screener)

92

58

63.0

15-17 Year Olds (age from Household Screener)

103

64

62.1

All Household Screeners

Phase 2 Screener **
All Phase 2 Screeners
Tobacco Status

Adult Interview
All Adult Interviews
Tobacco Status

Parent Interview
All Parent Interviews
Youth Age Group

Youth Interview
All Youth Interviews
Youth Age Group

* El i gi bl e:
Hous ehol d s creener: a l l confi rmed/pres umed res i denti a l a ddres s es
Pha s e 2 Screener: a l l a dul ts s a mpl ed from hous ehol d s creener
Adul t i ntervi ew: a l l a dul ts s a mpl ed from Pha s e 2 s creener, excl udi ng a ge i nel i gi bl es from Pha s e 2
Pa rent i ntervi ew: pa rents of a l l youths s a mpl ed from hous ehol d s creener, excl udi ng youth a ge i nel i gi bl es from pa rent
i ntervi ew
Youth i ntervi ew: a l l youths s a mpl ed from hous ehol d s creener, excl udi ng youth a ge i nel i gi bl es from pa rent i ntervi ew
** Incl udes Fi el d Tes t s uppl ementa l s a mpl e of Pha s e 1 non-us ers for res pons e ra te ca l cul a ti on purpos es

Population Assessment of Tobacco and Health (PATH) Study (NIDA)
Table 25 PATH Field Test Interview Sources of Non-Response by Instrument and Various Characteristics
Reasons for Non-Response (percent)

Household Screener
All Household Screeners

Eligible*
(n)

Nonresponse
(n)

Refusal

Maximum
Language
No Contact
Contacts
Barrier

Physical
Barrier

All Other

2944

1774

59.7

21.7

6.2

10.5

1.1

0.8

Long Version

1400

838

21.0

6.0

10.0

0.7

0.6

Short Version
Phase 2 Screener **

1544

936

61.7
57.9

22.3

6.4

10.9

1.5

1.0

1152

454

50.2

6.2

20.3

8.1

5.3

9.9

Sampled in Household Screener as User

579

197

52.3

6.6

23.4

6.6

1.5

9.6

Sampled in Household Screener as Non-user

573

257

48.6

5.8

17.9

9.3

8.2

10.1

501

21

4.8

0.0

4.8

0.0

4.8

85.7

Sampled in Phase 2 Screener as User

390

12

8.3

0.0

0.0

0.0

8.3

83.3

Sampled in Phase 2 Screener as Non-User

111

9

0.0

0.0

11.1

0.0

0.0

88.9

195

67

13.4

1.5

4.5

3.0

0.0

77.6

12-14 Year Olds (age from Household Screener)

92

30

10.0

0.0

3.3

3.3

0.0

83.3

15-17 Year Olds (age from Household Screener)

103

37

16.2

2.7

5.4

2.7

0.0

73.0

195

73

20.5

1.4

6.8

2.7

0.0

68.5

12-14 Year Olds (age from Household Screener)

92

0.0

2.9

2.9

0.0

70.6

103

34
39

23.5

15-17 Year Olds (age from Household Screener)

17.9

2.6

10.3

2.6

0.0

66.7

All Phase 2 Screeners
Tobacco Status

Adult Interview
All Adult Interviews
Tobacco Status

Parent Interview
All Parent Interviews
Youth Age Group

Youth Interview
All Youth Interviews
Youth Age Group

* Eligible:
Household screener: all confirmed/presumed residential addresses
Phase 2 screener: all adults sampled from household screener
Adult interview: all adults sampled from Phase 2 screener, excluding age ineligibles from Phase 2
Parent interview: parents of all youths sampled from household screener, excluding youth age ineligibles from parent interview
Youth interview: all youths sampled from household screener, excluding youth age ineligibles from parent interview
** Includes Field Test supplemental sample of Phase 1 non-users for response rate calculation purposes
Refusal includes those who refused to participate in an interview.
Maximum Contacts: at least one contact was made but the effort was not successful before reaching the maximum of six in-person attempts.
No contact: similar to Maximum Contacts, except that, after six attempts, no contact was actually achieved with the targeted person.
Language barrier: the targeted person did not speak English (the only language in which the field test was conducted).
Physical barrier: the interviewer was unable to gain physical access to the dwelling unit, due to restricted access situations such as gated communities or
controlled-access buildings.
All other: miscellaneous reasons which include persons staying in a seasonal home, or final break offs; for the field test, this category also included cases
that were still in process at the end of the field period and did not qualify for the other categories.

Population Assessment of Tobacco and Health (PATH) Study (NIDA)
Table 26 PATH Field Test Unweighted Housing Unit Eligibility Rate, Weighted Household
Response Rate, and Proportion of Screener Responding Households with Eligible Youth

Housing Unit Eligibility Rate
Household Response Rate

Field Test Result (%)

Design Assumption (%)

89.8
40.1

88.6
90.0

10.9
16.2

9.3
16.0

Proportion of Screener Responding
Households with Youths
9 to 11 Years Old
12 to 17 Years Old

Note: field test not designed to be predictive of response rate

Population Assessment of Tobacco and Health (PATH) Study (NIDA)

Proportion (%)

Design Assumption
(TUS-CPS 20062007) (%)

NSDUH 2010 (%)

NHIS 2010 (%)

Field Test Result
37
74

Count

Table 27 PATH Field Test Unweighted Distribution of Enumerated Adults by Sampling Domain, Based on Information Provided by Household Informant

Sampling Domaina

113

Black - 18 to 24 Years Old, Tobacco User
Black - 18 to 24 Years Old, Tobacco Non-User

0.4
1.6
2.6
7.6
2.5

Black - 25+ Years Old, Tobacco User

0.6
1.3
4.3
5.9
3.1

160

0.3
1.7
2.1
8.4
2.2

251

1.7
3.3
5.1
6.0
7.2

Black - 25+ Years Old, Tobacco Non-User
Non-black - 18 to 24 Years Old, Tobacco User
133

Non-Black - 18 to 24 Years Old, Tobacco Non-User

8.0
17.6
59.7
503

7.4
20.9
56.4
939

7.9
16.3
61.0

Non-Black - 25+ Years Old, Tobacco User

11.4
22.8
42.5

Non-Black - 25+ Years Old, Non-tobacco User

Assignment to sampling domain is based on information from the Phase 1 household screener.

Note: The table covers all individuals as enumerated in Phase 1 household screener.
a

Population Assessment of Tobacco and Health (PATH) Study (NIDA)
Table28 PATH Field Test Rate of Misclassification - Household Informant Report of Household Member Tobacco Use
False Positive
a

Age Group

b

False Negative

Sample Size

Field Test
Rate (%)

Design
Assumption
(%)

c

Sample Size

Field Test
Rate (%)

Design
Assumption
(%)

18 to 24 Years Old
98

4.1%

5%

81

32.1%

5%

All Persons with Short Screener

43

0.0%

-

43

37.2%

-

All Persons with Long Screener
Non-Household Informants

55

7.3%

-

38

26.3%

48

6.3%

-

55

36.4%

-

All Persons Regardless of Screener Version

25 Years or Older
All Persons Regardless of Screener Version

276

14.1%

5%

155

9.7%

All Persons with Short Screener

130

6.9%

-

85

14.1%

2%
-

All Persons with Long Screener

146

20.5%

-

70

4.3%

-

94

14.9%

-

59

3.4%

-

Non-Household Informants

Note: table addresses Phase 1 tobacco use classification as implemented for sampling purposes. For sampling purposes only, members who
could not be classified with certainty were considered to be users.
a

Age group as reported in Phase 2 screener
False positive is defined where the Phase 1 screener classified person as tobacco user, while the Phase 2 screener self-report classified as non-user
c
False negative is defined where the Phase 1 screener classified person as tobacco non-user, while the Phase 2 screener self-report classified as user
b

Population Assessment of Tobacco and Health (PATH) Study (NIDA)
Table 29 PATH Field Test Address Frame Coverage Rates, Housing Unit Eligibility Rates, and Weighted Household
Response Rates by PSU
Housing Unit
Count From
Census 2010

301

Sampled
Housing
Units
210

Screener
Responding
Households
96

302

210

56

3,445,076

303

210

53

555,932

304

212

58

12,787

305

216

97

2,180,359

306

212

46

417,862

307

211

43

135,160

308

212

94

612,004

309

213

98

296,685

310

211

34

398,510

311

210

95

35,511

312

216

20

419,974

313

296

28

835,127

314
315

210

46

268,426

210

99

407,998

PSU

84,872

Address
Frame
Coverage
Rate (%)

Housing
Unit
Eligibility
Rate (%)

Household Screener
Response Rate (%)

100.2
97.6
98.4
88.3
95.9
104.9
100.1
101.7
94.4
94.6
95.5
96.8
93.6
98.4
98.1

88.9
96.2
91.9
81.1
93.7
83.3
88.5
96.7
75.3
91.2
82.6
91.6
93.4
92.8
97.2

42.7
40.1
31.2
58.5
40.1
45.4
50.4
33.7
41.4
36.3
58.3
30.1
24.3
32.5
49.0

Population Assessment of Tobacco and Health (PATH) Study (NIDA)
Table 30 PATH Field Test Weighted Phase 2 Screener Response Rate and Extended Response Rates Among Eligible Adults by PSU
Adults Sampled
for Phase 2
Screener

Respondents to
Phase 2
Screener

301

90

53

302

62

41

303

58

43

304

89

56

305

72

44

306

67

53

307

72

51

308

59

33

309

62

47

310

57

33

311

99

73

312

49

31

313

57

20

314

51

31

315

64

34

PSU

Weighted Phase
2 Screener
Response Rate
(%)

59.0
61.0
74.7
62.6
60.1
81.2
69.2
55.0
76.9
61.0
73.0
56.1
30.4
57.3
50.9

Adults Sampled
for Extended
Interview

Respondents to
Extended
Interview

Weighted Extended
Interview Response
Rate (%)

45

43

32

32

33

33

39

38

27

24

45

43

42

39

25

25

42

40

27

24

53

51

28

28

18

15

22

22

23

23

94.9
100.0
100.0
91.3
93.9
93.3
88.9
100.0
89.7
77.6
97.7
100.0
88.5
100.0
100.0

Population Assessment of Tobacco and Health (PATH) Study (NIDA)
Table 31 PATH Field Test Weighted Extended Interview Response
Rate for Youth Ages 12-17
PSU

Sampled Youth

Youth
Respondent

Weighted Youth
Response Rate
(%)

301

10

6

59.6

302

25

14

50.0

303

13

10

78.6

304

24

16

67.2

305

12

7

56.7

306

9

7

77.0

307

18

17

95.0

308

4

1

25.0

309

3

3

100

310

11

3

27.6

311

11

9

84.9

312

10

6

59.9

313

12

8

70.5

314

17

8

45.1

315

16

8

50.1

Population Assessment of Tobacco and Health (PATH) Study (NIDA)
Table 32 PATH Field Test Weighted Proportions of Youth Ages 9-11 (Shadow Youth)
Whose Parents/Guardians Agreed to be Contacted During Follow-up Waves

PSU

Sampled Shadow
Youth

Shadow Youth Whose Weighted Proportion
Parents/Guardians of Shadow Youth that
Agreed to be
Can Be Contacted for
Contacted for the
Follow-up Waves
Follow-up Waves
(%)

6

5

84.5

302

10

7

69.8

303

5

4

78.0

304

15

13

89.8

305

7

6

85.5

306

7

6

84.3

307

5

2

40.0

308

3

1

33.3

309

4

2

50.6

310

15

8

53.9

311

12

9

77.3

312

7

2

29.6

313

9

5

57.2

314

10

8

79.2

315

9

7

76.4

301

Population Assessment of Tobacco and Health (PATH) Study (NIDA)
Table 33 PATH Field Test Screener Response Rates, By Incentive Amount and Screener Version
Occupancy

Experimental
Condition
No Incentive
Long Screener
Short Screener
$5 Incentive
Long Screener
Short Screener
$10 Incentive
Long Screener
Short Screener
Overall
Long Screener
Short Screener

Household Screener

Residential
Dwelling Units

Household
Screener
Response
Rate

(%)

(n)

89.9
90.1
89.7
90.9
89.7
92.2
89.3
88.4
90.2
90.1
89.5
90.7

1,090
628
462
932
460
472
913
451
462
2,935
1,539
1,396

Sampled
Addresses

Residential
Occupancy Rate

(n)
1,212
697
515
1,025
513
512
1,022
510
512
3,259
1,720
1,539

Enumeration

Adults
Enumerated

Adults
Screened

(%)

(n)

(%)

36.7
35.7
38.1
40.3
40.2
40.5
43.0
44.1
42.2
39.8
39.5
40.3

815
384
431
750
365
385
803
411
392
2,368
1,160
1,206

97.3
94.3
100.0
95.3
91.0
99.5
94.2
88.8
99.7
95.5
91.3
99.8

Population Assessment of Tobacco and Health (PATH) Study (NIDA)
Table 34 PATH Field Test Household-Level and Adult-Level Rates of Tobacco Use, By Incentive Amount and Screener Version
Household-level Tobacco Use

Individual Adult Household Member Tobacco Use

(n)

(%)

(n)

(%)

Proportion of
Adults
Reported to
Smoke
Cigarettes,
Pipes or
Cigars
(%)

386

49.5

793

37.6

28.9

8.7

Long Screener

178

53.4

362

40.6

30.9

9.7

Short Screener

208

46.2

431

35.0

27.1

7.9

372

50.0

717

37.2

29.4

7.8

Long Screener

179

53.6

332

41.3

32.2

9.7

Short Screener

193

46.6

385

33.8

27.0

6.8

381

48.0

754

34.7

25.9

8.9

Long Screener

186

51.6

365

38.1

28.5

9.6

Short Screener

195

44.6

391

31.7

23.5

8.2

1,138

49.1

2,264

36.5

28.0

8.5

Long Screener

543

52.9

1,059

39.9

30.5

9.4

Short Screener

596

45.8

1,205

33.5

25.9

7.6

Experimental
Condition
No Incentive

$5 Incentive

$10 Incentive

Overall

Screened
Households

Proportion of
Households with
1+ Users

Note: Table covers all completed household screeners.

Proportion of
Adults for
Enumerated
Whom any
Adults in
Tobacco Use
Screened
was
Households
Reported

Proportion of
Adults
Reported to
Use Any
Other
Tobacco
Products
(%)

Population Assessment of Tobacco and Health (PATH) Study (NIDA)
Table 35 PATH Field Test Proportion of Households with Completed Extendeds, and Person-level Extended Completion Rate, by Household Screener
Version, by Incentive Amount
Household-Level Extended Completion Rates

Experimental
Condition

No Incentive
Long Screener
Short Screener
$5 Incentive
Long Screener
Short Screener
$10 Incentive
Long Screener
Short Screener
Overall
Long Screener
Short Screener

Households
with at Least
One Sampled
Person
(n)

Proportion of
Households with at
Least One
Completed
Extended
(%)

Proportion of
Households
with Extended
for Every
Sampled
Member
(%)

324
153
171
311
150
161
322
161
161
957
464
493

45.4
45.8
45.0
49.8
53.3
46.6
49.7
45.3
54.0
48.3
48.1
48.5

36.4
38.6
34.5
37.9
38.7
37.3
35.4
31.1
39.8
36.6
36.0
37.1

Note: Table covers all sampled persons.

Completion Rate of Sampled
Person(s) from Phase 1
Household Screener

Adults and
Youths
Sampled from
Household
Screener
(n)
453
229
224
437
221
216
455
234
221
1345
684
661

Adults
Completing
Phase 2
Screener/
Youths
Completing
Extended
(%)
59.4
62.0
56.7
59.7
60.2
59.3
64.0
59.0
69.2
61.0
60.4
61.7

Completion Rate of Final Sampled
Person(s)

Adults Sampled
from Phase 2
Person-level
Screener/Youths Completion Rate,
Sampled from
Excluding NonHousehold
sampled Adults at
Screener
Phase 2
(n)
(%)
220
112
108
237
122
115
238
115
123
695
349
346

89.1
92.9
85.2
82.3
82.0
82.6
89.1
85.2
92.7
86.8
86.5
87.0

Population Assessment of Tobacco and Health (PATH) Study (NIDA)
Table 36 PATH Field Test Agreement between the Household Screener and Phase 2 Screener on Tobacco Use, By Incentive Amount and
Screener Version
Proportion Classified the Same Way in
Household Screener and Phase 2
Screener (Any Tobacco Use)
All Phase 2
Respondents
All Cases
(n=664)

SelfNot SelfRespondent in Respondent
Household
in Household
Screener
Screener
(n=380)
(n=284)

Kappas (Any Tobacco Use)

All Cases
(n=664)

SelfNot SelfRespondent Respondent
in Household in Household
Screener
Screener
(n=380)
(n=284)

(n)

(%)

(%)

(%)

(%)

(%)

(%)

221

91.4

95.0

87.4

0.825

0.896

0.743

Long Screener

108

95.4

98.4

91.3

0.904

0.968

0.812

Short Screener

113

87.6

91.2

84.2

0.750

0.806

0.667

218

90.8

92.6

87.1

0.814

0.845

0.761

Long Screener

106

90.6

91.0

90.0

0.809

0.816

0.795

Short Screener

112

91.1

94.1

84.4

0.815

0.874

0.732

225

88.0

88.8

87.2

0.761

0.775

0.742

Long Screener

99

90.9

94.6

86.4

0.818

0.886

0.724

Short Screener

126

85.7

84.3

87.9

0.717

0.689

0.751

664

90.1

92.1

87.2

0.800

0.837

0.748

Long Screener

313

92.3

94.6

89.2

0.845

0.886

0.782

Short Screener

351

88.0

89.7

85.6

0.760

0.789

0.720

No Incentive

$5 Incentive

$10 Incentive

Overall

Note: Table covers all Phase 2 screener respondents whose tobacco use could be classified with certainty in Phase 1 screener.