January 2005 JOSH - 10/11 ET - 20 picas - commentary #1 - males05.jan
What Do Student Drug Use Surveys Really Mean?
Mike A. Males
Mike A. Males, PhD, Lecturer, Sociology Dept., University of California, Santa Cruz, 214 College Eight, Santa Cruz, CA 95064; (email@example.com).
Federally funded projects such as Monitoring the Future and PRIDE, as well as numerous state and local activities devote considerable resources to survey junior and senior high school students’ drug use.1 Monitoring the Future, a detailed annual survey of high school students, began in 1975 for 12th graders and 1991 for eighth and 10th graders, on issues such as drug, alcohol, and tobacco use, and other personal behaviors and attitudes. PRIDE, a congressionally authorized annual survey by the Parents’ Resource Institute for Drug Education, surveys students in grades 6-12 on drug, alcohol, and tobacco use and related issues. These surveys generate the main, and often the sole, means by which drug education policies and programs are evaluated. Student surveys also represent the main measure of national drug policy design and evaluation, as noted by their prominence in the 2003 National Drug Control Strategy report.2
The Office of National Drug Control Policy (ONDCP) Reauthorization Act of 1998 designated its main objective as “reduction of adolescent unlawful drug use (as measured in terms of illicit drug use during the past 30 days by the Monitoring the Future Survey of the University of Michigan or the National PRIDE Survey conducted by the National Parents' Resource Institute for Drug Education)...”3
Lobbies seeking to reform drug policies and to legalize marijuana also preoccupy themselves with adolescent drug use. For example, the Marijuana Policy Project refers constantly to drug use surveys to argue that ONDCP’s “War on Drugs” has failed because “the prohibition of marijuana in the United States has not curtailed adolescent marijuana use.”4
A reported increase in student drug use can mean the end for drug education programs, as shown by the abandonment of the 1970s pharmacological education,5,6 and the 1990s impetus to curtail Drug Awareness and Resistance Education (DARE) programs.7 Conversely, even a small decline in student drug use is cited as evidence of policy success. For example, the claim that random drug testing of students by school authorities results in less drug use (a point on which the few existing studies yield mixed results)8 provides the main basis for proposals by the ONDCP to promote testing.2
Yet, a larger question remains: Why are we concerned about student drug use? Does survey-reported legal and illicit drug use constitute a valid measure of student well-being and justify the importance attached to its levels and trends? Those who attach overriding important to survey findings argue that drug use by youth is associated with serious problems, such as delinquency, school failure and dropout, early pregnancy, greater odds of injury, suicide, violence, and other anti-social behaviors, as well as future drug abuse in adulthood.2 Other studies suggest correlations between drug use and unhealthy outcomes largely disappear when the relatively small number of frequent drug abusers with serious problems is evaluated separately from the larger number of moderate drug users whose behaviors resemble nonusers.9,10 Monitoring the Future finds students who only use marijuana report behaviors and attitudes similar to those who report using no drugs.11 The National Household Survey of Drug Use and Health reports that many Americans whose illegal drug use only consisted of marijuana do not contribute to hard-drug abuse.12
This commentary addresses specific questions: Does student drug use as reported on surveys display external validity? Do trends in, and levels of, students reporting use of marijuana or other illegal drugs correlate with or predict trends for other problems, including those most often said to be associated with drug abuse?
Monitoring the Future (MTF), a project of the Institute for Social Research at the University of Michigan, has surveyed representative samples of high school seniors in schools nationwide for 29 years (1975 through 2003). MTF surveys examine type, frequency, duration, and circumstances of student drug use, as well as other topics including satisfaction with life, social attitudes, school safety, future plans, delinquency, and victimization.13 Consistent questions used over time provide a basis to examine associations between student drug use and other behaviors by year. MTF’s continuity and comprehensiveness render it the benchmark survey cited by ONDCP and other authorities.
Several other studies also examined the levels and trends of student drug use from 1975 to 2002. The National Center for Health Statistics provided annual information on number and rate for births14 and violent deaths15 involving older teens (aged 15-19) and young adults (aged 20-29). The Federal Bureau of Investigation reported annual criminal arrests by crime for youths by age.16 The National Center for Education Statistics provided high school graduation rates and teacher-student ratios by year.17 The Bureau of the Census provided annual figures for median incomes of families with children, children living in poverty, divorce, and teenage unemployment.18 The premise for this paper was addressed in three ways.
To examine immediate associations with drug use, MTF’s annual reports of the percentages of high school seniors using drugs were correlated with the corresponding percentages of delinquency, victimization, and school safety by seniors reported annually by MTF in the Bureau of Justice Statistics’ Sourcebook of Criminal Justice Statistics.13 High school seniors’ drug use percentages by year also were correlated with rates of birth, violent death, and criminal arrest involving teenagers. In this latter analysis, high school students comprised about 85% of the larger population whose vital and crime rates were measured, introducing a possibility of selection bias.
To examine whether high school students’ drug use rates affect their behavior as young adults, MTF’s high school drug use rates were compared to self-reported drug use rates among 18-29 year-olds six years later. That is, high school seniors’ annual drug use rates for 1975 through 1996 were compared to young-adult drug use rates for 1981 through 2002. Similarly, high school seniors’ drug use rates were compared, using the same six-year time delay, to vital and criminal arrest statistics for those aged 18-29. Six years was chosen as the time delay on the assumption that the average high school senior is aged 17-18, and the average 18-29 year-old is 23-24.
A regression analysis compared the most commonly cited drug-use measure, percentage of students using illicit drugs the previous month, with the same outcome measures in the correlation analysis. Stepwise multiple regression controls for external variables that might affect drug use and other behaviors, such as annual percentage of youth in poverty, median income of families with children, divorce rate, teenage unemployment rate, teacher-student ratio, and time period (year). The criterion set for statistical significance, p < .01, was stricter than that used in most studies (p < .05).
Variations in student self-reported drug use by class (year) were large and fluctuated rapidly. In 1978 and 1979, for example, 39% of high school seniors reported using an illegal drug the previous month. By 1992, fewer than half as many (14%) did so. Five years later, in 1997, the proportion nearly doubled, to 26%. Other student behavior, as well as vital and crime statistics, also showed large fluctuations over the 28-year period. If high rates of drug use contributed significantly to other student problems, or if low rates were associated with amelioration of such problems, one would expect the effects on classes with sharply differing drug use rates to be quite noticeable.
In the results from the correlational analysis, measures of drug use by students were strongly internally consistent. Complete results of the correlational analysis are available in a table posted by the author at http://home.earthlink.net/~mmales/table2.htm. Classes that reported high rates of drug use in general, or of a particular drug, reported high rates of using other drugs, as well as alcohol and, less consistently, cigarettes. The exception was heroin, whose trends were not correlated with those of other drugs.
Student drug use trends displayed external validity as well, but not in the expected direction. Trends in student drug use consistently were negatively correlated with other undesirable behaviors tracked in the MTF survey. Compared to students in classes that report low drug use rates, students in classes that report high rates were significantly less likely to report having been in a serious fight, injuring someone seriously, having frequent fights with parents, being in a gang fight, stealing a car, committing armed robbery, committing arson, or being victimized by a major or minor theft at school.
Analysis of national vital statistics and crime figures showed results consistent with MTF’s self-reported behaviors. Classes with high rates of drug use had significantly lower rates of birth, arrest for homicide, arrest for violent offenses, death by homicide, death by suicide and related accidents, and death by firearms.
The analysis generated 315 combinations of the nine categories of drug use and 35 other student behavior, vital, educational, and victimization variables, including nine variables that showed no significant relationship to any kind of drug use. Of the total number, 117 showed significant negative correlations, 195 showed no significant relationship, and only three showed a significant positive relationship. Two-thirds of 195 nonsignificant associations were negative and were considerably stronger in direction than the smaller number of positive associations. Of the six external (background) variables examined, only teacher-student ratio showed a consistently significant association. The higher the ratio, the higher the rate of student drug use.
Rates of drug use by high school seniors correlated with rates of drug use by young adults aged 18-29. Heroin was the only exception; high school classes with high rates of overall drug use reported significantly lower rates of heroin use both in high school and as young adults. High school drug use rates did not predict later problems among young adults in violent death, suicide, homicide, firearm fatalities, criminal arrest, or college enrollment at physical or online universities. All 90 correlations were nonsignficant, and most were negative.
Results from the regression analysis, available from the author, generally supported an association between higher rates of student drug use and lower rates of other problems. Weighted regression coefficients showed higher rates of student drug use were strongly and significantly associated with 18 positive behavior outcomes, including lower rates of self-reported crime and violence by, and victimization of, students in and out of school; homicide, violent, and property crime arrests; homicide, suicide, and firearm deaths; and rate of birth by teenage mothers. Seventeen associations were not significant, and 14 were negative. None of the associations was significantly positive; only three were nonsignificantly positive.
More illegal drug use reported by American high school senior classes on Monitoring the Future surveys was not associated with increased risk for negative behaviors or outcomes. The preponderance of associations between higher rates of student drug use and lower rates of delinquency, victimization, criminal arrest, violent death, and other unwanted outcomes, one-half of which were statistically significant even by the strict standards, was striking and unexpected. Consistently greater student safety and better behavior in classes where drug use was more common persisted even when major background variables such as poverty, income, divorce, unemployment, time period, and teacher-student ratio were controlled. Conversely, greater drug use was rarely associated with worse behavior and outcomes.
Are students, then, better off when they use more drugs? The question addressed here was not whether drug use is good or bad for students, but whether drug use as measured on self-reporting surveys provides a valid indicator of student well-being and thus a viable basis for policy.
Surveys can fail as valid bases for policy development for several reasons. First, they may be inaccurate. Their numbers may reflect the willingness of students to report using drugs, not actual use of drugs. The National Research Council concluded in 2001 that, “inaccurate response is particularly acute in surveys of drug abuse, since illegal drug use is a stigmatized behavior and respondents are reluctant to report it accurately.”19 Strident anti-drug campaigns, such as “just say no” in the late 1980s or drug-testing regimes in schools today, could produce lower survey numbers not by reducing drug use, but by making students less willing to report it. Lack of correlation [(in fact, a near-significant negative correlation) between rates of student drug use and of teenage drug overdose deaths over time (T = 1.57, p < 0.13, for monthly drug use)] indicated that surveys provide flawed measures of drug abuse.
Large and officially sanctioned national surveys of the same school populations at the same time can yield contradictory results. From 2001 to 2002, MTF reported illicit drug use in the past month fell by 11% for eighth graders and 8% for seniors. Conversely, PRIDE reported eighth graders’ drug use rose by 41% and seniors’ increased 8%.1 ONDCP officials ignored PRIDE and touted recent MTF figures as validating their “aggressive national goals to reduce youth drug use.”20 The Marijuana Policy Project, an opponent of the War on Drugs and advocate of “realistic” drug education, ignored MTF and seized on PRIDE’s finding as “demonstrat(ing) that ONDCP's past attempts to reduce teen drug use have been complete failures.”21 Neither side mentioned the sharp increase in teenage and adult drug-related hospital emergency treatments and drug abuse deaths past five years,15,22 indicating how extensively the political debate over teenage drug use numbers has strayed from realistic concern about teenage health.
Even if surveys are accurate, levels of and changes in drug use they report may signify less than they appear. Intuitively, one would expect that just as higher rates of drug use by an individual often accompany personal problems with school, employer, family, police, and health, higher rates of drug use in a population would signal other generational problems. This does not appear to be the case. Up or down shifts in the student population’s overall drug use were dominated by moderate and occasional use of milder drugs such as marijuana, which are the most prevalent styles and also those least likely to be connected to other risks such as violence or crime. Conversely, only small percentages of students report heavier drug use (such as daily marijuana use) or use of more dangerous drugs (such as heroin), which are the styles most often connected to other problems.1,9,10
A paradox may apply to the abilities of different generations to handle drugs. For example, the teenage cohort of 1970, though reporting sparing use of illegal drugs by survey measures,23 suffered drug overdose death rates more than double those of the teenage cohort of 1980, which used drugs at a rate 250% higher. Overreliance on drug use surveys sabotaged potentially promising pharmacological education approaches of the 1970s, which were abandoned after student drug use increased, even though drug overdose deaths declined by 62%, among teens.
The contradictory trends continued from the early 1980s to 1992, when high school seniors’ illicit drug use plummeted by 65%, while most teenage risks (violence, crime, violent death, school problems) increased sharply. Then both trends reversed again. From 1992 to 1997, student drug use rates nearly doubled, while nearly all other teen risk measures plunged. To complete the baffling trends, high school seniors’ drug use generally fell from 1997 to 2002, as did most other youth risk measures, but teenage drug overdose deaths doubled.1,15,16,22
Overreliance on surveys promotes increasingly intrusive efforts to stop all student drug use, and this trend may be counterproductive. MTF researchers’ updated 2003 analysis of school drug-testing programs found that the most extensive, random testing regimens may slightly discourage casual marijuana use but promote significantly more use of less-detectable harder drugs.8,25
Drug-use surveys may be inaccurate; they may provide irrelevant or trivial data compared to more important influences affecting student behavior; or they may obscure the fact that higher rates of drug use are connected to student well-being in ways not yet fully understood. In any case, the conclusion is the same: policy makers, school administrators, substance abuse programs, and the news media attach too much importance to surveys. Students in the years in which 40% reported using drugs were no worse off, and often significantly better off by most important measures, than were students in years when 15% report using drugs. Thus, do policies that focus primarily on reducing numbers on self-report surveys best serve school health objectives?
1. Monitoring the Future. Ann Arbor, Mich: Institute for Social Research, University of Michigan; Available at: http://www.monitoringthefuture.org. PRIDE. Parents’ Resource Institute for Drug Education. Available at: http://www.pridesurveys.com/.
2. Office of National Drug Control Policy. National Drug Control Strategy. Washington, DC: The White House; 2003:1-4.
3.105th Congress, 2nd Session. HR 4328: Office of National Drug Control Policy Reauthorization Act of 1998. Div. C Title VII Sec. 706 (a)(4)(B). Washington, DC: House of Representatives; 1998.
4. Marijuana Policy Project. Marijuana prohibition has not curtailed marijuana use by adolescents. 2004. Available at: http://www.mpp.org/adolescents.html.
5. Goodstadt MS. School-based drug education in North America: what is wrong? what can be done? J Stud Alcohol. 1986;56(3):278-288.
6. Stuart RB. Teaching facts about drugs: pushing or preventing? J Educ Psychol. 1974,37(1):98-101.
7. Lynam D, Milch R, Clayton R. Project DARE: no effects at 10-year follow-up. J Consult Clin Psychol. 1999;67(4):590-593.
8. Yamaguchi R, Johnston LD, O’Malley PM. Relationship between student illicit drug use and school drug-testing policies. J Sch Health. 2003;73(4):159-164.
9. Newcomb MD, Bentler PM. Impact of adolescent drug use and social support on problems of young adults: a longitudinal study. J Abnorm Psychol. 1988;67(1):64-75.
10. Shedler J, Block J. Adolescent drug use and psychological health. Am Psychologist. 1990(5):612-630.
11. Johnston LD, Bachman JG, O'Malley PM. Monitoring the Future: Questionnaire responses from the nation's high school seniors, 2001. Ann Arbor, Mich: Institute for Social Research; 2003.
12. Substance Abuse and Mental Health Services Administration. Preliminary estimates from the 1995 National Household Survey on Drug Abuse. Rockville, Md: US Dept of Health and Human Services; 1996:31.
13. Johnston et al. Results also available in: Maguire K, Pastore AL, eds. Sourcebook of Criminal Justice Statistics 2002.Washington, DC: Bureau of Justice Statistics, US Dept of Justice; 2004. Tables 3.40, 3.43, 3.46, and corresponding previous annual tables, 1975-2001.
14. Martin JA, et al. Births: Final data for 2002. National Vital Statistics Report. Hyattsville, Md: National Center for Health Statistics; December 17, 2003;52:10;Table 4.
15. Kochanek KD, Smith BL. Deaths: Provisional data for 2002. National Vital Statistics Report. Hyattsville, Md: National Center for Health Statistics; February 11, 2004;52:13;Tables 1-3 and previous annual, 1970-2001.
16. Federal Bureau of Investigation. Uniform Crime Reports for the United States 2002. Washington, DC: US Dept of Justice; 2003. Table 38, and corresponding previous annual tables, 1975-2001.
17. National Center for Education Statistics. Digest of Education Statistics 2002. Washington, DC: US Dept of Education; 2004:Chapter 2. Available at: http://nces.ed.gov/programs/digest/d02_tf.asp
18. Bureau of the Census. 2003 Statistical Abstract of the United States. Washington, DC: US Government Printing Office; 2004:Tables 126 (divorce), 621 (unemployment), 684 (family income), and 701 (child poverty), and corresponding previous annual tables, 1975-2002.
19. Manski CF, Pepper JV, Petrie CV, eds. Informing America’s Policy on Illegal Drugs: What We Don’t Know Keeps Hurting Us. Washington, DC: National Academy Press; 2001:8.
20. National Institute on Drug Abuse. Teen drug abuse declines across wide front.
Washington, DC: December 19, 2003. Available at: http://www.nida.nih.gov/Newsroom/03/NR12-19.html.
21. Marijuana Policy Project. ONDCP ads increase teen drug use. Washington, DC: September 16, 2003. Available at: http://www.mpp.org/releases/nr091603ondcp.html.
22. Drug Abuse Warning Network (1980-2003). Annual Emergency Department Data. Annual Medical Examiner Data. Washington, DC: Substance Abuse and Mental Health Services Administration, US Dept of Health and Human Services.
23. Johnston L. Drugs and American Youth. Ann Arbor, Mich: Institute for Social Research, University of Michigan; 1973.
24. Yamaguchi R, Johnston LD, O’Malley PM. Drug testing in schools: Policies, practices, and association with student drug use. Youth, Education, & Society Occasional Paper 2. Ann Arbor, Mich: Institute for Social Research, University of Michigan; 2003.