Adolescent sexual behavior, drug use, and violence: Increased reporting with computer survey technology

Science

Volumn 280, Issue 5365, 1998, Pages 867-873

  Turner, C F ^a   Ku, L ^b   Rogers, S M ^a   Lindberg, L D ^b   Pleck, J H ^c   Sonenstein, F L ^b  

^a RTI INTERNATIONAL   (United States)

^b URBAN INSTITUTE   (United States)

^c UNIVERSITY OF ILLINOIS AT URBANA CHAMPAIGN   (United States)

Author keywords

[No Author keywords available]

Indexed keywords

ADOLESCENT; ADULT; ARTICLE; CHILD BEHAVIOR; COMPUTER; HEALTH SURVEY; HOMOSEXUALITY; HUMAN; HUMAN EXPERIMENT; INTERVIEW; INTRAVENOUS DRUG ABUSE; MALE; NORMAL HUMAN; PRIORITY JOURNAL; QUESTIONNAIRE; SEXUAL BEHAVIOR; VIOLENCE;

ADOLESCENT; ADOLESCENT BEHAVIOR; COMPUTERS; DATA COLLECTION; ETHNIC GROUPS; HUMANS; INTERVIEWS; MALE; PRIVACY; QUESTIONNAIRES; RISK-TAKING; SEXUAL BEHAVIOR; SOCIOECONOMIC FACTORS; SUBSTANCE-RELATED DISORDERS; UNITED STATES; VIOLENCE;

EID: 0032496321     PISSN: 00368075     EISSN: None     Source Type: Journal    
DOI: 10.1126/science.280.5365.867     Document Type: Article

References (60)

1. F. L. Sonenstein, J. H. Pleck, L. Ku, Fam. Plann. Perspect. 21, 132 (1989); J. A. Catania et al., Science 258, 1101 (1992); E. Laumann, J. Gagnon, R. Michael, S. Michaels, Social Organization of Sexuality (Univ. of Chicago Press, Chicago, IL, 1994); A. Johnson, J. Wadsworth, K. Wellings, J. Field, S. Bradshaw, Sexual Attitudes and Lifestyles (Blackwell, London, 1994); P. B. Davis, R. L. Yee, J. Chetwynd, N. McMillan, AIDS 7, 1509 (1994); K. Tanfer, Fam. Plann. Perspect. 25, 83 (1993).
2. F. L. Sonenstein, J. H. Pleck, L. Ku, Fam. Plann. Perspect. 21, 132 (1989); J. A. Catania et al., Science 258, 1101 (1992); E. Laumann, J. Gagnon, R. Michael, S. Michaels, Social Organization of Sexuality (Univ. of Chicago Press, Chicago, IL, 1994); A. Johnson, J. Wadsworth, K. Wellings, J. Field, S. Bradshaw, Sexual Attitudes and Lifestyles (Blackwell, London, 1994); P. B. Davis, R. L. Yee, J. Chetwynd, N. McMillan, AIDS 7, 1509 (1994); K. Tanfer, Fam. Plann. Perspect. 25, 83 (1993).
3. F. L. Sonenstein, J. H. Pleck, L. Ku, Fam. Plann. Perspect. 21, 132 (1989); J. A. Catania et al., Science 258, 1101 (1992); E. Laumann, J. Gagnon, R. Michael, S. Michaels, Social Organization of Sexuality (Univ. of Chicago Press, Chicago, IL, 1994); A. Johnson, J. Wadsworth, K. Wellings, J. Field, S. Bradshaw, Sexual Attitudes and Lifestyles (Blackwell, London, 1994); P. B. Davis, R. L. Yee, J. Chetwynd, N. McMillan, AIDS 7, 1509 (1994); K. Tanfer, Fam. Plann. Perspect. 25, 83 (1993).
4. F. L. Sonenstein, J. H. Pleck, L. Ku, Fam. Plann. Perspect. 21, 132 (1989); J. A. Catania et al., Science 258, 1101 (1992); E. Laumann, J. Gagnon, R. Michael, S. Michaels, Social Organization of Sexuality (Univ. of Chicago Press, Chicago, IL, 1994); A. Johnson, J. Wadsworth, K. Wellings, J. Field, S. Bradshaw, Sexual Attitudes and Lifestyles (Blackwell, London, 1994); P. B. Davis, R. L. Yee, J. Chetwynd, N. McMillan, AIDS 7, 1509 (1994); K. Tanfer, Fam. Plann. Perspect. 25, 83 (1993).
5. F. L. Sonenstein, J. H. Pleck, L. Ku, Fam. Plann. Perspect. 21, 132 (1989); J. A. Catania et al., Science 258, 1101 (1992); E. Laumann, J. Gagnon, R. Michael, S. Michaels, Social Organization of Sexuality (Univ. of Chicago Press, Chicago, IL, 1994); A. Johnson, J. Wadsworth, K. Wellings, J. Field, S. Bradshaw, Sexual Attitudes and Lifestyles (Blackwell, London, 1994); P. B. Davis, R. L. Yee, J. Chetwynd, N. McMillan, AIDS 7, 1509 (1994); K. Tanfer, Fam. Plann. Perspect. 25, 83 (1993).
6. F. L. Sonenstein, J. H. Pleck, L. Ku, Fam. Plann. Perspect. 21, 132 (1989); J. A. Catania et al., Science 258, 1101 (1992); E. Laumann, J. Gagnon, R. Michael, S. Michaels, Social Organization of Sexuality (Univ. of Chicago Press, Chicago, IL, 1994); A. Johnson, J. Wadsworth, K. Wellings, J. Field, S. Bradshaw, Sexual Attitudes and Lifestyles (Blackwell, London, 1994); P. B. Davis, R. L. Yee, J. Chetwynd, N. McMillan, AIDS 7, 1509 (1994); K. Tanfer, Fam. Plann. Perspect. 25, 83 (1993).
7. See, for example, H. G. Miller, C. F. Turner, L. E. Moses, Eds., AIDS: The Second Decade (Report of the National Academy of Sciences-National Research Council Committee on AIDS Research and the Behavioral, Social, and Statistical Sciences; National Academy Press, Washington, DC, 1990), pp. 147-152.
8. This situation appears to be unique to the United States. Among young males 15 to 24 years of age, the United States has the highest homicide rate among 22 industrialized nations [M. L. Rosenberg, J. Health Care Poor Underserved 6, 102 (1995); H. Saner and P. Ellickson, J. Adolesc. Health 19, 94 (1994); J. M. Bergstein, D. Hemenway, B. Kennedy, S. Quaday, R. Ander, J. Trauma 41, 794 (1996); P. F. Adams, C. A. Schoenborn, A. M. Moss, C. W. Warren, L. Kann, Vital Health Star. 10, 1 (1992)].
9. This situation appears to be unique to the United States. Among young males 15 to 24 years of age, the United States has the highest homicide rate among 22 industrialized nations [M. L. Rosenberg, J. Health Care Poor Underserved 6, 102 (1995); H. Saner and P. Ellickson, J. Adolesc. Health 19, 94 (1994); J. M. Bergstein, D. Hemenway, B. Kennedy, S. Quaday, R. Ander, J. Trauma 41, 794 (1996); P. F. Adams, C. A. Schoenborn, A. M. Moss, C. W. Warren, L. Kann, Vital Health Star. 10, 1 (1992)].
10. This situation appears to be unique to the United States. Among young males 15 to 24 years of age, the United States has the highest homicide rate among 22 industrialized nations [M. L. Rosenberg, J. Health Care Poor Underserved 6, 102 (1995); H. Saner and P. Ellickson, J. Adolesc. Health 19, 94 (1994); J. M. Bergstein, D. Hemenway, B. Kennedy, S. Quaday, R. Ander, J. Trauma 41, 794 (1996); P. F. Adams, C. A. Schoenborn, A. M. Moss, C. W. Warren, L. Kann, Vital Health Star. 10, 1 (1992)].
11. This situation appears to be unique to the United States. Among young males 15 to 24 years of age, the United States has the highest homicide rate among 22 industrialized nations [M. L. Rosenberg, J. Health Care Poor Underserved 6, 102 (1995); H. Saner and P. Ellickson, J. Adolesc. Health 19, 94 (1994); J. M. Bergstein, D. Hemenway, B. Kennedy, S. Quaday, R. Ander, J. Trauma 41, 794 (1996); P. F. Adams, C. A. Schoenborn, A. M. Moss, C. W. Warren, L. Kann, Vital Health Star. 10, 1 (1992)].
12. R. C. Lewontin, N.Y. Rev. Books, 20 April 1995, p. 24.
13. E. F. Jones and J. D. Forrest, Demography 29, 113 (1992).
14. C. F. Turner, R. D. Danella, S. M. Rogers, Sex. Transm. Dis. 22, 173 (1995).
15. C. F. Turner, J. T. Lessler, J. Devore, in Survey Measurement of Drug Use: Methodological Issues, C. F. Turner, J. T. Lessler, J. C. Gfroerer, Eds. (Government Printing Office, Washington, DC, 1992) pp. 177-220.
16. See, for example, C. Jenkins, paper presented at the meeting of the American Association for Public Opinion Research, Norfolk, VA, 15 to 18 May 1997.
17. G. Johnston, demonstration for a seminar at the National Center for Health Statistics, Hyattsville, MD, 28 January 1992; J. M. O'Reilly and C. F. Turner, presentation to the Washington Statistical Society, 18 March 1992; J. M. O'Reilly, M. Hubbard, J. Lessler, P. Biemer, C. F. Turner, J. Off. Stat. 10, 197 (1994).
18. G. Johnston, demonstration for a seminar at the National Center for Health Statistics, Hyattsville, MD, 28 January 1992; J. M. O'Reilly and C. F. Turner, presentation to the Washington Statistical Society, 18 March 1992; J. M. O'Reilly, M. Hubbard, J. Lessler, P. Biemer, C. F. Turner, J. Off. Stat. 10, 197 (1994).
19. C. F. Turner et al., in Survey Research Using Automated Data Collection Systems, M. Couper et al., Eds. (Wiley, New York, in press); C. F. Turner, H. G. Miller, S. M. Rogers, in Researching Sexual Behavior, J. Bancroft, Ed. (Indiana Univ. Press, Bloomington, in press); C. F. Turner, S. M. Rogers, T. P. Hendershot, H. G. Miller, J. P. Thornberry, Public Health Rep. 111, 276 (1996).
20. C. F. Turner et al., in Survey Research Using Automated Data Collection Systems, M. Couper et al., Eds. (Wiley, New York, in press); C. F. Turner, H. G. Miller, S. M. Rogers, in Researching Sexual Behavior, J. Bancroft, Ed. (Indiana Univ. Press, Bloomington, in press); C. F. Turner, S. M. Rogers, T. P. Hendershot, H. G. Miller, J. P. Thornberry, Public Health Rep. 111, 276 (1996).
21. C. F. Turner et al., in Survey Research Using Automated Data Collection Systems, M. Couper et al., Eds. (Wiley, New York, in press); C. F. Turner, H. G. Miller, S. M. Rogers, in Researching Sexual Behavior, J. Bancroft, Ed. (Indiana Univ. Press, Bloomington, in press); C. F. Turner, S. M. Rogers, T. P. Hendershot, H. G. Miller, J. P. Thornberry, Public Health Rep. 111, 276 (1996).
22. Multilingual audio-CASI administration was not used in NSAM but has been tested in preliminary studies [T. P. Hendershot, S. M. Rogers, J. P. Thornberry, H. G. Miller, C. F. Turner, in Health Survey Research Methods, R. Warnecke, Ed. (National Center for Health Statistics, Hyattsville, MD, 1996), pp. 165-169].
23. L. Ku, F. L. Sonenstein, J. H. Pleck, Fam. Plann. Perspect. 26, 246 (1994); Am. J. Public Health 83, 1609 (1993); J. H. Pleck, F. L. Sonenstein, L. Ku, Fam. Plann. Perspect. 25, 106 (1993); L. Ku, F. L. Sonenstein, J. H. Pleck, ibid. 24, 100 (1993).
24. L. Ku, F. L. Sonenstein, J. H. Pleck, Fam. Plann. Perspect. 26, 246 (1994); Am. J. Public Health 83, 1609 (1993); J. H. Pleck, F. L. Sonenstein, L. Ku, Fam. Plann. Perspect. 25, 106 (1993); L. Ku, F. L. Sonenstein, J. H. Pleck, ibid. 24, 100 (1993).
25. L. Ku, F. L. Sonenstein, J. H. Pleck, Fam. Plann. Perspect. 26, 246 (1994); Am. J. Public Health 83, 1609 (1993); J. H. Pleck, F. L. Sonenstein, L. Ku, Fam. Plann. Perspect. 25, 106 (1993); L. Ku, F. L. Sonenstein, J. H. Pleck, ibid. 24, 100 (1993).
26. L. Ku, F. L. Sonenstein, J. H. Pleck, Fam. Plann. Perspect. 26, 246 (1994); Am. J. Public Health 83, 1609 (1993); J. H. Pleck, F. L. Sonenstein, L. Ku, Fam. Plann. Perspect. 25, 106 (1993); L. Ku, F. L. Sonenstein, J. H. Pleck, ibid. 24, 100 (1993).
27. The 1992 National Health and Social Life Survey (NHSLS), for example, found that 9.1% of U.S. men in 1992 reported having had male-male contacts after puberty but only 4.9% reported such contacts after age 18. Those results imply that 4.2% of men had mate-male contacts that were restricted to adolescence and some portion of the remaining 4.9% would have begun such contacts during adolescence. In a 1970 Kinsey Institute survey of a representative sample of U.S. men, 81% of the men reporting some history of male-male contact said that their first male-male experiences occurred before age 19 (6). It should also be noted that other analyses of the 1970 Kinsey Institute survey indicated that 8.4% of men reported some male-male contact that did not persist beyond age 14 [R. E. Fay, C. F. Turner, A. D. Klassen, J. H. Gagnon, Science 243, 338 (1989)]. The 1992 NHSLS did not assess prepubertal contacts.
28. The design of the survey was changed in 1991 to make it more gender neutral in discussing sexual contacts and included a SAQ on male-male sexual contacts that was given before asking about male-female contacts. This change was made so that respondents would not infer that we expected only heterosexual contacts to be reported.
29. The sample was drawn as a probability subsample of an area probability sample selected for another study conducted in 1992. Details of the multistage sampling procedures used in selecting the NSAM sample can be found in Research Triangle Institute, National Survey of Adolescent Males: Field Report (Research Triangle Institute, Research Triangle Park, NC, 1995), pp. 3.2-3.6. (Our analysis is based on a final data set that includes a small number of NSAM cases that were received after the field report was completed.)
30. An additional 12 units (0.02%) were excluded because of mismatches between the original sample frame information and interviewer determination as to whether the unit was a household or group quarters.
31. Bureau of the Census, Preliminary Projections: Non-institutional Population by Age, Sex. Race and Hispanic Origin (U.S. Department of Commerce, Washington, DC, May 1995).
32. B. Shah, B. G. Barnwell, P. N. Hunt, L. M. LaVange, Software for Survey Data Analysis (SUDAAN) (Research Triangle Institute, Research Triangle Park, NC, 1996).
33. After completing the self-interview portion of the survey, respondents who were age 18 and older were asked to provide a urine sample to be tested for chlamydia and gonorrhea. The results of those tests are not discussed here.
34. Nonresponse to individual questions may reduce the N for individual items. However, item nonresponse rates were quite low, averaging 1.3% for the measurements shown in Tables 2 through 5. Nonrespondents to individual items have been excluded from our analyses. In addition, it should be noted that some analyses purposely restrict the population base. So, for example, our estimate of the percentage of males using "condoms at last sex" (Table 2) is restricted to those males who reported having had vaginal, oral, or anal sex with a female. Population restrictions are noted in the tables.
35. L. A. Goodman, J. Am. Stat. Assoc. 63, 1091 (1968); Analyzing Qualitative/Categorical Data (Haberman, Cambridge, 1978); Analysis of Categorical Data (Academic Press, New York, 1978); Y. Bishop, S. Feinberg, P. Holland, Discrete Multivariate Analysis: Theory and Practice (MIT Press, Cambridge, 1975).
36. L. A. Goodman, J. Am. Stat. Assoc. 63, 1091 (1968); Analyzing Qualitative/Categorical Data (Haberman, Cambridge, 1978); Analysis of Categorical Data (Academic Press, New York, 1978); Y. Bishop, S. Feinberg, P. Holland, Discrete Multivariate Analysis: Theory and Practice (MIT Press, Cambridge, 1975).
37. L. A. Goodman, J. Am. Stat. Assoc. 63, 1091 (1968); Analyzing Qualitative/Categorical Data (Haberman, Cambridge, 1978); Analysis of Categorical Data (Academic Press, New York, 1978); Y. Bishop, S. Feinberg, P. Holland, Discrete Multivariate Analysis: Theory and Practice (MIT Press, Cambridge, 1975).
38. L. A. Goodman, J. Am. Stat. Assoc. 63, 1091 (1968); Analyzing Qualitative/Categorical Data (Haberman, Cambridge, 1978); Analysis of Categorical Data (Academic Press, New York, 1978); Y. Bishop, S. Feinberg, P. Holland, Discrete Multivariate Analysis: Theory and Practice (MIT Press, Cambridge, 1975).
39. An alternative approach would be to treat the sample of 1672 respondents as a closed population that had been randomly assigned to one of two interviewing conditions. In that approach, neither weights nor special statistical algorithms would be required. Statistical tests using that approach would provide an appropriate answer to the question of whether audio-CASI had had a significant effect on that particular sample of respondents. The approach is reasonable, but it does not allow one to generalize the results to the national population.
40. 2 statistic reported by SUDAAN (78).
41. As a check on the adequacy of the randomization, we compared the response distributions of 62 predominantly yes-no nonattitude questions measured in the prior interviewer-administered section of the questionnaire for respondents assigned to the two experimental conditions. We were initially perplexed to find that people who were assigned to the paper SAQ condition were more likely than those in the audio-CASI condition to report in the interviewer-administered section of the survey that they were virgins. They did not, however, show significant differences on demographic variables or other sexual behaviors (for example, employment during the past year, timing of last heterosexual intercourse). Our examination indicated that this result appears to have been a random occurrence. Of the 62 comparisons, 2 were found to be significant at the P < 0.05 level. On the basis of probability theory, we would expect random assignment to produce (on average) at least three significant (P < 0.05) results in 62 independent tests. We thus conclude that the randomization was successfully executed. The adjusted odds ratios shown in Tables 2 through 5 provide statistical adjustments for observed random differences in the composition of the sample used in the two experimental conditions. These adjusted ratios were derived from logistic regression models that used the following as control variables: race (black, white, other), respondent's age, whether the respondent was attending school, whether the respondent reported during the earlier interviewer-administered section of the survey that he had ever had sex with a female, and whether the respondent was covered by health insurance. The first three characteristics were selected to take account of differences among subpopulations in our targeted behaviors; the last two variables were chosen on the basis of preliminary statistical analyses that indicated random variations between the experimental and control conditions for the two characteristics (of 62 that were examined). Use of the adjusted rather than crude odds ratios does not markedly affect our conclusions (Tables 2 to 5).
42. In a few instances noted in Table 2, respondents in the experiment were asked again about heterosexual contacts that they had previously reported on in the interviewer-administered portion of the survey. Because respondents may have felt compelled to answer consistently rather than "honestly," responses to the questions that were asked again could be subject to a consistency bias that might have attenuated the effect of the interview mode. Moreover, there is a weak suggestion in these results that audio-CASI may have diminished "overreporting," which is sometimes thought to afflict measurements of highly normative behaviors, for example, reports by males of some male-female sexual experience. That effect, however, is not statistically significant and is weak compared with the effects observed on reporting of male-male sex and the other behaviors discussed [see also S. Newcomer and J. R. Udry, J. Adolesc. Res. 2. 419 (1988)].
43. Since the original tabulation of estimates for Tables 2 to 5 in this article, we detected a minor data anomaly that has been corrected in the public use data set. For four variables, the value 999 was an allowable (albeit unusual) response to questions on: (i) the number of times in past 12 months someone pulled a gun, knife, or razor on respondent; (ii) the number of times the respondent threatened to hurt someone during the last 12 months; (iii) the respondent's total number of lifetime female sexual partners; and (iv) the respondent's total number of female sex partners during the past 12 months. Similarly, the value 99 was an allowable response to a fifth question on the number of times the respondent was in a fight during the last year. Seven respondents gave one or more of these answers (999 or 99) to these questions. In our original tabulations, these extreme values were treated as missing data codes and excluded from the analysis. Correcting this error alters the reported percentages for two mode comparisons by more than 0.1 percentage points. For reporting of five or more lifetime (female) sex partners (Table 2), the results are 15.8% (SAQ) and 19.0% for audio-CASI (versus 15.8% and 18.8%, if the value 999 were treated as missing). Similarly, for respondents' reports of having threatened to hurt someone in the past year (Table 4), the results are 17.1% for SAQ and 26.3 % for audio-CASI (versus 17.1% and 26.1%, if the value 999 were treated as missing). In manual editing of paper SAQs for data entry, all values of 400 or more for these variables were treated as invalid and coded as "missing data." Thus comparison of paper SAQ estimates to audio-CASI estimates in which 999 codes are treated as missing values - as was done in Tables 2 through 5 - is appropriate. If the additional values 400 to 998 were treated as missing values for the audio-CASI measurements, the audio-CASI estimates would not change by more than 0.1%.
44. As noted in Table 2, it is possible that the prostitute was male. It should also be noted that a suggestive mode effect (P < 0.15) was found in Table 2 for respondents who reported that they had been paid to have sex.
45. It should also be noted that the only internally consistent reports of injection drug use in the past year were obtained from nine males in the audio-CASI mode. One male in the paper SAQ mode gave an inconsistent report of injection drug use. Respondents in the paper SAQ condition were asked a series of questions on injecting drugs. First, they were asked if they had ever taken street drugs by using a needle. Next, they were asked how frequently they had injected drugs during the past year, whether they had ever shared a needle, and how often they had sterilized previously used needles. In the audio-CASI condition, respondents who reported they had never injected drugs were skipped out of the remaining questions. However, one respondent completing the paper questionnaire first reported never injecting drugs but later reported doing so "a few times" during the past year.
46. The series of questions on commingling of sex and alcohol or drug use was preceded by a gate question asking "Have you ever had sexual intercourse, including vaginal, oral, or anal intercourse, with a female?" Respondents who answered "no" to that question were skipped past the series of questions on drug and alcohol use during sex. Analyses comparing responses to this question with those to questions asked separately about oral, vaginal, and anal sex suggest that some respondents may have interpreted the gate question as asking about the co-occurrence of all three types of sexual behavior. Thus, only 60.6% of paper SAQ and 54.0% of audio-CASI respondents replied "yes" to this question. In comparison, in response to four questions asking separately about vaginal sex, insertive and receptive oral sex, and anal sex with females, 68.1% of paper SAQ and 63.9% of audio-CASI respondents indicated that they had engaged in one (or more) types of sex.
47. This finding is not likely to reflect the carrying of guns for hunting: 13.4% of males living in urban areas reported that they had carried a gun during the prior 30 days compared with 11.8% of males living in rural areas.
48. Exploratory analyses were conducted with weights that had been normalized so that the weighted N equaled the sample N. The effect of the complex sample design was not taken into account in these initial exploratory analyses that fit a series of hierarchical log-linear models. Subsequently, odds ratios were recalculated for results that evidenced substantial disparities across schooling levels in the estimated effect of interview mode on survey response. These recalculated odds ratios and associated P values (Table 5) used algorithms that took account of the complex sample design.
49. More precisely, the four categories were Hispanics, non-Hispanic whites, non-Hispanic blacks, and other.
50. There was also a less substantial difference in the effect of mode on reporting that someone had pulled a gun on the respondent in the past year. For respondents who were at grade level, there was an insignificant mode effect (crude OR = 1.06; 17.5% in paper SAQ versus 18.4% in audio-CASI), whereas a significant (P < 0.05) mode effect in the same direction was found for respondents who were not at grade level (crude OR = 2.22; 15.3% in paper SAQ versus 28.5% in audio-CASI).
51. The National Opinion Research Center (NORC) General Social Survey asked the following question of national samples drawn annually between 1988 and 1994: "What about sexual relations between two adults of the same sex. Do you think it is always wrong, almost always wrong, wrong only sometimes, or not wrong at all?" Among males aged 18 and older (N = 2877), 86% of those who had not completed high school responded "always wrong" compared with 83% of those who left school after 12th grade, 72% of those who had completed 13 to 15 years of school, and 57% of those who had completed 16 or more years of education. (Estimated percentages are weighted to account for varying probabilities of selection into the sample.) Data were extracted and tabulated from the NORC cumulative data file [J. A. Davis and T. W. Smith, General Social Surveys, 1972-1994 (machine-readable data file) (NORC, Chicago, IL, 1972-1994)].
52. S. M. Rogers and C. F. Turner, J. Sex Res. 28, 491 (1991); M. Witt, J. Pantula, R. Folsom. B. Cox, in (7), pp. 85-108.
53. In particular, we note that the question on injection drug use was worded: "Have you ever taken street drugs, like heroin or cocaine, using a needle? This includes 'shooting up' and 'skin popping'" (emphasis in original). It is possible that in formulating a response, audio-CASI respondents may have focused on the initial clause "ever taken street drugs, like heroin or cocaine" in the audio format of this question, whereas respondents in the paper SAQ condition may have been more likely to notice the subsequent "using a needle" clause and the explanatory phrases "shooting up" and "skin popping" when reading the paper version of this question.
54. These estimates are based on weighted tabulations from the public use data set for the 1995 National Household Survey on Drug Abuse (NHSDA). The NHSDA is sponsored by the Substance Abuse and Mental Health Services Administration of the Department of Health and Human Services. In 1995, the NHSDA surveyed a population sample of more than 17,000 household residents in the United States. Questions on illicit drug use were administered with paper SAQs. Our tabulations are based on the sample of 1624 males ages 15 to 19 who were included in the NHSDA survey.
55. See, for example, N. Bradbum, S. Sudman, and Associates, Improving Interview Method and Questionnaire Design (Bass, San Francisco, 1979); J. J. Waterton and J. C. Duffy, Int. Stat. Rev. 52, 173 (1984).
56. See, for example, N. Bradbum, S. Sudman, and Associates, Improving Interview Method and Questionnaire Design (Bass, San Francisco, 1979); J. J. Waterton and J. C. Duffy, Int. Stat. Rev. 52, 173 (1984).
57. We would also note that a parallel study of the effect of audio-CASI on abortion reporting found that with this technology the known underreporting of this procedure decreased. In that instance, there was an external standard (reports by abortion providers) to provide assurance that increased reporting was more accurate reporting [H. G. Miller, S. M. Rogers, J. Gribble, C. F. Turner, in The Science of Self Report (provisional title), A. Stone, Ed. (Lawrence Erlbaum Associates, Mahwah, NJ, in press)].
58. R. Tourangeau and T. Smith, Public Opin. Q. 60, 275 (1995);
59. C. F. Turner, H. G. Miller, T. K. Smith, P. C. Cooley, S. M. Rogers, in Survey and Statistical Computing 1996, R. Banks, J. Fairgrieve, L. Gerrard, Eds. (Association for Survey Computing, Chesham, Bucks, UK, 1996).
60. This research was supported by NIH grant R01-HD30861 with funding from the National Institute of Child Health and Human Development (NIH-NICHD), the National Institute of Mental Health (NIH-NIMH), the Office of Population Affairs, and the Centers for Disease Control and Prevention. We thank H. Zelon, F. Mierzwa, J. Chromy, D. Alien, B. Forsyth, and S. Williams for their contributions and A. Harrell, H. Miller, R. Folsom, and J. Gfroerer for helpful comments.