Impact of T-ACASI on Estimates of Youth Smoking Prevalence: Results of UMASS Tobacco Study

1
Impact of T-ACASI on Estimates of Youth
Smoking PrevalenceResults of UMASS Tobacco Study

• Lois Biener,1 Charles F. Turner, 2 Amy L.
  Nyman1

1Center for Survey Research University of
Massachusetts Boston
2 Health Behavior Measurement Program Research
Triangle Institute
Presented at American Public Health Association
Meetings, Phila., Nov. 12, 2002
Funded by The National Cancer Institutes State
Community Tobacco Control Research Initiative
2
Introduction
A key concern for tobacco researchers is
obtaining accurate information on youth smoking
behaviors. School-based self-administered surveys
under-represent school drop-outs, truants and
absentees, and securing cooperation from school
systems is difficult. In addition, obtaining
parental consent can reduce both the response
rate and the representativeness of the sample.
Random-digit-dial (RDD) telephone surveys can
recruit representative samples, but they yield
lower estimates of smoking prevalence than
in-school, self-administered surveys. Although
some of this discrepancy may arise from
over-reporting in school settings, childrens
worries about being overheard by parents and
reluctance to disclose smoking to adult
interviewers is likely to produce under-reporting
in telephone surveys. We present results of an
experiment in which these factors were reduced.
3
Methods
A representative sample of 3800 youth, aged 12 to
17 were recruited for the UMass Tobacco Study.
Two-thirds were randomly assigned to be surveyed
about their smoking experiences using standard
interviewer-administered questioning (T-IAQ)
one-third were assigned to a private
computer-administered self-interview (T-ACASI).
In T-ACASI respondents listen to pre-recorded
questions and enter answers on the keypad of a
touchtone telephone. Before speaking with the
youth, permission was obtained from a parent or
guardian who had, in most cases served as the
informant for a household screener which
enumerated the household and collected
information on the smoking status of all resident
adults.
4
Data Collection Protocol
Interviewers introduced the study to all youth
respondents, and conducted the first 2 sections
of the interview (18 questions) which dealt with
details of school attendance, after-school
activities, and television viewing
patterns. Respondents assigned to T-ACASI were
then connected to the computerized portion of the
interview and were told that they would be
connected back to the interviewer when the
computerized portion was completed. After
completing the 2 sections on smoking behavior
(mean time 7.5 minutes), respondents were
automatically returned to the interviewer and
completed the remaining 6 sections of the
interview.
5
Research Questions

• Is T-ACASI technology acceptable to youth 12 17
  years of age?
• Are youth more likely to report smoking behavior
  with T-ACASI than with a live interviewer?
• Are youth more likely to report susceptibility to
  smoking with T-CASI?

6
Results

• Response rate was somewhat lower for T-ACASI
  (60.8) than T-IAQ (65.8).
• Some proportion of the T-ACASI decrement was due
  to technological problems and interviewer error.

7
Survey Items For Prevalence Indices
Item Respondents N Mode Difference
Have you smoked at least 100 cigarettes in your life? All youth 12-17 3844 Not significant
Have you ever experimented with cigarette smoking, even a few puffs? All youth except established smokers (those whove smoked 100 cigarettes) 3541 Not significant
Have you ever smoked a whole cigarette? Everyone who has puffed 890 P.001
Think about the past 30 days. Did you smoke a cigarette, even a puff, on any of those days? Everyone who has experimented 1191 Not significant
When did you last smoke or puff on a cigarette? Smoked or puffed but not in the past 30 days 779 Not significant
8
Estimates of Smoking Prevalence
9
Ever Smoked Whole Cigarette
p lt .05
N127 N340 N419
Age
10
Lifetime SmokingHas had at least a puff
p lt .05
N1213 N1330 N1277
Age
11
Past Year SmokingSmoked at least 1 cigarette in
past year
p lt .05
N1213 N1330 N1277
Age
12
Past Month SmokingSmoked at least 1 cigarette
past month
p lt .05
N1213 N1330 N1277
Age
13
Estimates of Susceptibility to Smoking
14
Do you think you will try a cigarette soon?
Never smokers and puffers
p lt .05
N1148 N1109 N811


15
Do you think you will smoke in the next year?


p lt .05
N1211 N1328 N1277
16
Would you smoke if best friend offered a
cigarette?
N1211 N1328 N1273

p lt .05
17
Do your parents know that you smoke?
p lt .05
N373
18
Would/Do parents disapprove a lot of your
smoking?
p lt .05

N377 N3463


19
Results Smoking Prevalence Estimates

• Estimates of lifetime smoking were 36 higher for
  the youngest respondents with T-ACASI compared to
  T-IAQ.
• Estimates of past year and past month smoking
  were 60 to 70 higher for youth 14-15 years old
  with T-ACASI.

20
Results Susceptibility to smoking

• Youth 12 to 15 who were least experienced with
  cigarettes were twice as likely to admit
  intentions to try a cigarette soon with T-ACASI
  than T-IAQ.
• All youth were slightly less likely to report a
  strong commitment not to smoke in the future with
  T-ACASI.

21
Results Reports of parental attitudes

• Smokers were 50 more likely to report that their
  parents knew that they smoked and less likely to
  report parental disapproval when speaking with a
  live interviewer than when responding to the
  automated survey.

22
Conclusions

• The traditional telephone interview of
  Massachusetts youth results in under-reporting of
  current smoking among 14-15 year olds by about 5
  percentage points.
• Traditional telephone interviewing appears to
  under-estimate youth susceptibility to future
  smoking, especially among those with the lowest
  levels of experience with cigarettes.

23
Conclusions (continued)

• The use of T-ACASI methodology does not remove
  the discrepancy in smoking prevalence estimates
  between self-administered, school-based surveys
  and home-based telephone surveys of adolescents.

24
Past Month Smoking Massachusetts Youth by Mode
25

• Additional research can help to explore the
  proportion of the remaining discrepancy which is
  due to over-reporting in self-administered
  surveys and that due to under-reporting in
  telephone interviews.