Reducing The Risk of Drug Involvement Among Early Adolescents: An Evaluation of Drug Abuse Resistance Education (DARE)

This research was supported in part by a grant from the Office of Educational Research and Improvement, U.S. Department of Education, and the Center for Research on Effective Schooling for Disadvantaged Students at the Johns Hopkins University. Partial support was also provided by the Charleston County School District in South Carolina. I would like to thank the following people for their technical support and assistance: The Charleston County School District staff (especially Candice Bates), the Charleston County DARE officers, and Lois Hybl and Gary Gottfredson at the Johns Hopkins University. I am also grateful for comments provided by Denise Gottfredson on earlier drafts of this paper.

This paper examines the effectiveness of the DARE (Drug Abuse Resistance Education) program in Charleston County, South Carolina by comparing 341 fifth grade DARE students to 367 nonDARE students. Significant differences were found in the predicted direction for alcohol use in the last year, belief in prosocial norms, association with drug using peers, positive peer association, attitudes against substance use, and assertiveness. No differences were found on cigarette, tobacco, or marijuana use in the last year, frequency of any drug use in the past month, attitudes about police, coping strategies, attachment and commitment to school, rebellious behavior, and self-esteem.

The adolescent drug use epidemic in the United States dates back over 20 years. Beginning in the 1960's when much of the nation's youth began to use psychoactive drugs such as LSD and PCP, the drug epidemic created public concern as it continued into the 70's. The 1980's showed much of the same with drug use on the rise and new drugs such as MDMA (XTC), ice, and crack suddenly appearing in every major city.

This paper begins with an examination of the adolescent drug use problem in the United States and Charleston, South Carolina (where the current study takes place). Possible solutions to this problem are briefly discussed and a summary of prior studies of Drug Abuse Resistance Education (DARE) is provided. The current study is then presented followed by a brief discussion and recommendations for future research.

Much of what is known about adolescent drug use is a result of the annual High School Senior Survey conducted by the Institute for Social Research at the University of Michigan (Johnston, 1973). Data from a recent report examining drug use (Johnston, Bachman, & O'Malley, 1991) show a gradual decline for all types of drugs since 1975. However, the current levels of drug use in the United States imply a large number of adolescents are still using drugs. For example, in 1990, 90 percent of U.S. seniors reported drinking alcohol at some time in their lives, while 64 percent said they had smoked cigarettes.

Adolescent drug use in Charleston, South Carolina, where the current study takes place, is similar to national use.

During the 1989-90 school year all students (223,663) in grades 7-12 in South Carolina's 91 public school districts were surveyed to collect a variety of information on current and past drug use (South Carolina Department of Education and South Carolina Commission on Alcohol and Drug Abuse, 1990). Many of the survey questions were modeled after the annual High School Senior Survey (Johnston, 1973).

The main findings for grade 12 from the South Carolina survey for Charleston County are presented in Table 1. Relevant national data are also shown for comparison. Information collected from the Charleston survey includes lifetime, annual, and 30-day prevalence rates.

Lifetime prevalence rates show the U.S. percentages generally larger than those in Charleston. For example, almost 90% of the U.S. seniors compared to 77% of the Charleston seniors said they had drunk alcohol at least once in their lives. In addition, 8% more U.S. seniors than Charleston seniors said they had used marijuana and 17% more U.S.seniors said they had smoked cigarettes.

In contrast to the lifetime prevalence rates, the 30-day prevalence rates for Charleston and the U.S. are quite similar. Very small differences exist with about half favoring the U.S. and half favoring Charleston.

The South Carolina and National Youth Survey data give a useful picture of the extent of drug use in the United States and Charleston, South Carolina. However, it should be noted that many youths leave school before their senior year. Obviously, the youths who have dropped out of school before their senior year are not included in either the annual High School Senior Survey or the South Carolina Youth Survey. Since drug use is higher for high school drop outs than it is for those who stay in-school (Anhalt & Klein 1976; Johnston, 1973) reported senior drug use rates are most likely underestimates for all adolescents.

Even with dropouts excluded from the survey data, the amount of reported drug use in the United States, and Charleston, South Carolina, is high. Although national data show reported drug involvement slowly declining, the current high levels of drug use display a grim picture of adolescents today. Furthermore, the drug epidemic is far from over and the goal of drug free youth in America is still very distant. Answers to the question of what can be done to stop the drug use epidemic still escapes practitioners, law enforcement personnel, health professionals and social scientists.

The past two decades have led to a variety of strategies aimed at combatting the drug problem. Polich, Ellickson, Reuter, & Kahan (1984) suggest the three most widely used attempts to combat or control drug use are supply reduction, treatment, and prevention.

Efforts to limit or control the supplies of drugs have been carried out via laws unfavorable towards drug use and corresponding law enforcement activities. Law enforcement agencies have directed efforts at reducing the production, import, distribution, and retail sales of illegal "street" drugs. The hope remains that by targeting these major market areas the quantity of drugs entering the country will decrease, trafficking and selling drugs will become more risky, shortages of drugs in the illicit market will take place, and the price of drugs to consumers will increase, ultimately reducing consumption.

As Hawkins, Catalano, and Miller (1992) point out, manipulating illegal drug supplies by increasing drug interdiction and drug dealer arrests should lead to positive outcomes such as raising the price of street drugs to the user, thus reducing the demand for drugs.

However, contrary evidence is cited by Polich, et al. (1984). They conclude doubling drug interdiction, and/or increasing arrests and imprisonment of drug dealers would affect neither retail prices nor the availability of illegal drugs. Essentially, the point is made that large drug quantities will always be available to take the place of any quantity confiscated. Increasing arrests would do little, they argue, because prison overcrowding forces the least violent to become paroled and often times these types of prisoners are low level street dealers that end up back on the street. Even if lower level dealers are arrested and kept in jail, there are many more that will take their place. Finally, because there is an immense amount of competition on the streets, dealers are forced to keep their prices down to stay in business. Therefore, supply reduction used alone as a means for reducing society's drug problem appears ineffective.

Similar to supply reduction, millions of dollars are spent every year on treatment as a means of curtailing drug use. And much like supply reduction strategies, treatment also shows little promise for eliminating drug use, particularly among adolescents.

Much of the drug treatment literature suggests treatment for adolescents is ineffective (Hubbard, Cavanaugh, Graddock, & Rachal, 1983; Miller, 1973; Stein & Davis, 1982). Treatment effectiveness is most often measured by continued abstinence from drugs.

Research on adolescent treatment programs suggests treatment, especially for adolescents, requires a lifestyle adjustment. This is to say that for most adolescents drug abuse is not a problem of physiological dependence. Rather, the problem stems from adolescent "life problems." Many researchers suggest that attention to these types of problems should be first and foremost (Bennett, 1983; Coupey & Schonberg, 1982). Researchers in the medical field agree adolescent drug abuse cannot be treated apart from family, school, and peer related problems (Macdonald & Newton, 1981; Mackenzie, 1982; Monopolis & Savage, 1982). This implies adolescent drug abusers are treated for the wrong problem since most programs are designed to deal with physical drug dependence. Research supports this argument by showing traditional drug treatment programs are ineffective in treating adolescent clients (Hubbard et al., 1983; Sells & Simpson, 1979).

A review of treatment programs produces mixed results with no clear, conclusive evidence and studies plagued by methodological flaws. For example, Ogborne (1978) claims treatment is not effective, NIDA (1981) reports it is effective, and Einstein (1981) says a general evaluation cannot be made. However, Polich et al. (1984) and Beschner (1989) caution that few, if any, are backed by true scientific evaluations.

Polich and his colleges (1984) reviewed several small scale treatment studies applicable to youthful drug abusers, none of which produced any large desired effects. A few studies did show some evidence of success. However, these results can most often be explained by rival hypotheses. Many studies found length of treatment stay to effect treatment success. Since many youth drop out of treatment programs it is not known whether these results are due to the increased benefit of treatment or to client self- selection. The lack of control group in the typical study also makes it difficult to evaluate the self-selection threat.

Although adolescent treatment efforts in general have not demonstrated desired effects, this is not to say treatment should be abandoned. Instead, steps should be taken to restructure adolescent treatment programs to deal with general adolescent life problems. Perhaps then, treatment programs will show more promise as a strategy for reducing or eliminating adolescent drug use.

Prevention holds more promise for controlling adolescent drug use than supply reduction or treatment. Reasons for promise include the timing of prevention programs and their focus on "gateway" substances - alcohol, tobacco, and marijuana. National data show youth initiating alcohol use as early as age 11 and marijuana and other illicit drugs at age 12 (Elliot & Huizinga, 1984). Because drug use often begins at such an early age, prevention programs must target youth before they come in contact with drugs. Currently, many drug prevention programs do in fact target youth while they are still in elementary school. This is especially important in light of the fact that the earlier an individual uses drugs the more likely they are to engage in greater and more persistent use of dangerous drugs (Flemming, Kellam, & Brown, 1982; Robins & Przybeck, 1985).

Many studies indicate drug use begins with one of the "gateway" substances and follows a logical progression to experimentation with other drugs (Hamburg, Braemer, & Jahnke, 1975; Kandel, 1978; Loeber & Le Blanc, 1990; Richards, 1980). Prevention programs show promise because most, if not all, drug prevention programs focus on "gateway" drugs.

Prevention efforts have not always been as promising, however. In the past, evaluations of many different prevention programs showed little or no effectiveness. Early studies were also methodologically weak. Since then there have been several "waves" of drug prevention programs each building on what was previously learned. More recent approaches have proven effective in reducing "gateway" drug use with studies demonstrating an increase in methodological rigor.

Traditional prevention approaches include information dissemination, affective education, and alternative activities. These efforts are based on a misunderstanding about why adolescents engaged in drug use. For example, information dissemination programs assume adolescents use drugs simply because they lack information about such drugs. Information dissemination approaches provided adolescents with facts about the pharmacology of drugs, the uses of various drugs types, and the consequences of drug use. Fear arousal and moral persuasion are two variations of the information dissemination, or health education model that provide similar information, adding either scare tactics or moral appeals. Affective education focuses on clarifying values and increasing self-esteem assuming individuals lacking these attributes will use drugs. Similarly, alternative activities try to relieve boredom and provide adolescents with stimulating alternatives hoping they will engage in these activities instead of turning to drug use.

Research clearly demonstrates the first generation of drug prevention programs has little or no impact on deterring adolescent drug use (Berberin, Gross, Lovejoy, & Paparella, 1976; Hanson, 1980; Kinder, Pape, & Walfish, 1980; Malvin, Moskowitz, Schaps, & Schaeffer, 1985; Schaps, Bartolo, Moskowitz, Palley, & Churgin, 1981). In fact, some programs are associated with an increase in drug use (Gordon & McAlister, 1982; Swisher & Hoffman, 1975).

The second generation of drug prevention efforts has proven more effective in reducing adolescent drug use. Psychosocial approaches such as psychological inoculation, resistance skills training and, personal and social skills training target research-based risk factors for adolescent drug use. All of these programs focus on increasing an individual's personal and social competence through skill acquisition (Arkin, Roemhild, Johnson, Luepker, & Murray, 1981; Botvin & Dusenbury, 1987; Schinke & Gilchrist, 1985; Hansen, Johnson, Flay, Graham, & Sobel, 1988; Telch, Killen, McAlister, Perry, & Maccoby, 1982). Most programs teach personal and social skills such as problem-solving, decision-making, coping, resisting peer pressure, and assertiveness. Of the prevention efforts reviewed, the literature suggests continued psychosocial efforts be employed with emphasis placed on resistance skill training and personal and social skill training approaches. Follow-up, or booster sessions are recommended, however, since there is some evidence initial effects may decline (Botvin, Eng, & Williams, 1983).

DARE (Drug Abuse Resistance Education) is a drug abuse prevention program that focuses on teaching students skills for recognizing and resisting social pressures to use drugs. DARE lessons also focus on the development of self-esteem, coping, assertiveness, communications skills, risk assessment and decision making skills, and the identification of positive alternatives to drug use.

Taught by a uniformed police officer, the program consists of 17 lessons offered once a week for 45 to 50 minutes. The DARE curriculum can only be taught by police officers who attend an intensive two-week, 80 hour, training. The DARE program calls for a wide range of teaching activities including question and answer sessions, group discussion, role play, and workbook exercises.

The DARE curriculum was created by Dr. Ruth Rich, a curriculum specialist with the Los Angeles Unified School District, from a "second generation" curriculum known as Project SMART (Self-Management and Resistance Training) (Hansen, et al., 1988). DARE was piloted in fifty Los Angeles elementary schools with over 8,000 fifth and sixth grade students during the 1983-84 school year. Two years later, all 345 elementary schools under the Los Angels Police Department's jurisdiction had a DARE officer assigned to teach the curriculum. The program, which originally targeted fifth and sixth grade students, was then expanded to include a junior high school curriculum and a much briefer orientation for students in kindergarten through fourth grade.

DARE is one of, if not the most, wide spread drug prevention programs in the United States. In 1989, over three million children in 80,000 classrooms were exposed to DARE ("Project DARE", 1990). Currently, there are DARE programs in every state in the United States and some counties have mandated DARE as part of the school health curriculum. It has also been implemented in several other countries including Canada, England, Australia, and New Zealand. In addition, it has been adopted by many reservation schools operated by the Bureau of Indian Affairs, and by the worldwide network of U.S. Defense Department schools for children of military personnel. There is a Spanish version and a Braille translation of the student workbook. Efforts are also under way to develop strategies for teaching DARE to hearing impaired and other special needs students.

Several DARE evaluations have been conducted over the last eight years in at least seven states and Canada (Agopian & Becker, 1990; Aniskiewicz & Wysong, 1987; Clayton, Cattarello, Day, & Walden, 1991; Clayton, Cattarello, & Walden, in press; DeJong, 1987; Earle, 1987; Evaluation and Training Institute, 1990; Faine, 1989; Faine & Bohlander, 1988, 1989; Manos, Kameoka, & Tanja, 1986; Nyre, 1985, 1986; Nyre & Rose, 1987; Ringwalt, Ennett, & Holt, 1991; Walker, 1990). Some show positive results, some show negative results, and most have serious methodological flaws.

Most of the DARE studies conclude that DARE is a "success". However, success has various meanings. For some evaluations it means teachers and other school administrators surveyed said "DARE was a success". In other evaluations it means students responded they liked the DARE program. Still others claim success if teachers and students rate DARE as "useful" or "valuable". For the most part, success is based on the finding that students are more able to generate "appropriate" responses to a widely used 19 item questionnaire about drug facts and attitudes after the DARE program than before. In these last instances, almost all had no control group. Several of the studies above contain such severe methodology problems that any results, if cited, should be questioned. In a review of several of these studies, Clayton et al. (1991, p. 300) labels most of them as "at best pilot and/or descriptive in nature" and does not bother mentioning any of their findings.

Methodological flaws contained in most of the DARE evaluations include one or more of the following problems: 1) no control group, 2) post-test only, 3) poorly operationalized measures, 4) low alpha levels for scales ( For their experiment in North Carolina, Ringwalt and his colleges (1991) evaluated the DARE program using 1270 fifth and sixth grade students as subjects. They randomly assigned 10 schools to receive the DARE program and 10 schools to serve as controls.

All students were pre-tested before the program began using a questionnaire designed to measure the following variables: self-report drug use, intentions to use drugs in the next year, attitudes towards drugs, perceived costs and benefits of drug use, perceived peer attitudes toward drug use, perceived media influences on drug use, self-esteem, and assertiveness. The reported internal reliability of all scales was favorable (.60 to .90)

Significant pre-treatment differences were found on measures of race, sex, self-report alcohol use, general attitudes towards drugs, perceived peer attitudes towards drugs, costs of alcohol use, and perceived media influences.

Controlling on pre-treatment differences, the dependent variable at time 1 (pre-test), and school type, it was concluded DARE met some of its immediate objectives. Significant differences between the experimental and control group include general attitudes towards drugs, attitudes toward specific drugs (beer, wine coolers, wine, cigarettes, and inhalants), perceptions of peers attitudes towards drug use, assertiveness, recognizing media influences to use drugs, and the costs associated with drug use. However, no statistically significant effects were found for self-reported drug use, future intentions to use drugs, perceived benefits of drug use (alcohol and cigarettes) or self-esteem.

Ringwalt et al. (1991) conducted an evaluation study showing the DARE program had effects on some of the immediate outcome objectives. However, because the experimental and control groups were quite different to begin with, it could be argued that even though statistical controls were employed the groups probably differed on other variables not measured by the pre-test. These unmeasured pre-treatment differences could account for the observed post-test differences.

A second point about the study should also be mentioned. The initial pages of the study explain the fact that methodological shortcomings have existed in drug program evaluations but that the current study improves upon one of those problems by performing statistical analyses appropriate for the research design. Continuing, in the results section the authors note prior evaluations have conducted the analysis at the wrong level. They make the argument that some studies have used individuals as the unit of analysis when schools have been assigned to treatment and control conditions. They immediately go on to say that in order to guard against any contamination of the results by school differences in their study, analysis of covariance, with school as a covariate, is employed. While the authors succeed at controlling for post-test differences associated with school membership, they still perform the analysis at a different level than the assignment, thus inflating the degrees of freedom.

A second DARE evaluation also demonstrating methodological strength over previous studies is that of Faine and Bohlander (1988). The authors not only compared DARE to nonDARE students in the fifth grade but also looked at four school types in Frankfort, Kentucky - rural, parochial, inner- city, and suburban. Eight schools were randomly assigned to receive DARE and six were randomly assigned to the control condition in the Fall and Spring of the 1987-88 school year. Two additional control group schools were selected on the basis of school type to match the school characteristics of the experimental group. The randomization and selection process resulted in 451 experimental students and 332 control students.

The six outcome variables measured were self-esteem, knowledge of drugs, attitudes towards drugs and alcohol, peer resistance, perceived external control and attitudes toward the police. There were no reported interaction effects between DARE and school type on any of the outcome measures. It should be noted that self-reported drug use was not examined. Comparing DARE to control students, they found significant differences in the expected direction for all six measures which included self-esteem (p Faine and Bohlander (1989) extended their original evaluation by conducting two phases of a one-year follow up study. However, severe methodological problems prohibit drawing any conclusions. The first phase design involves testing the control and experimental cohort at the end of the 1988-89 school year in order to assess the long term effectiveness of DARE. However, after one year the control group had also received DARE. In this situation, any observable differences cannot confidently be attributed to the DARE program. This is especially true in light of the fact the authors reported the shift from the first to the second year meant the majority of students moved from an elementary school to a junior high school. The change in school structure alone could have influenced the results, not to mention other possibilities such as a maturation effect.

Unfortunately, the second phase of the follow-up is just as methodologically flawed as the first. Because all students in the original DARE evaluation had received DARE by the end of the 1988-89 school year, a control group from two additional counties was sought out in order to make comparisons. Since the additional control counties had not been pre-tested, there is no way of knowing if any pre-treatment differences existed between the control and experimental students before the experimental students were exposed to DARE. Although Faine and Bohlander's (1988) initial DARE evaluation produced convincing results, too many rival hypothesis exist to draw conclusions about the long term follow-up study.

The last DARE study worth mentioning took place in Lexington, Kentucky (Clayton et al., 1991). During the 1987-1988 school year, the first of a five year longitudinal study, 23 schools were randomly assigned to the treatment (DARE) condition and 8 schools were randomly assigned as controls. The control group received the standard health curriculum which contained a drug education unit. The initial cohort was made up of 2,091 sixth grade students.

Initial equivalency tests indicate the treatment group had significantly more white students and significantly more positive attitudes towards drugs than the control group. The treatment group also reported significantly more lifetime, last year, and last month alcohol use.

The authors used analysis of variance to compare the treatment and control group outcomes. However, they only controlled on race despite other pre- treatment differences. Statistically significant (p A two-year follow-up study (Clayton, in press) examined the same cohort of 6th grade students using two follow-up questionnaires after the initial post-test. The first follow- up questionnaire was given during the 1988-1989 school year when the cohort was in the 7th grade and the second follow-up questionnaire was administered during the 1989-1990 school year when the cohort was in the 8th grade. Attrition rates over the two years did not differ significantly between the two groups.

The long-term effects of DARE prove to be minimal in terms of past year alcohol, cigarette, and marijuana use. The only statistically significant difference occurred at the first follow-up for last year marijuana use. Unfortunately, this finding occurred in the opposite direction than that expected. Significantly more marijuana use was reported by the DARE students than nonDARE students. Otherwise, no significant effects were found at any other time for any other drug type.

The long-term effectiveness of DARE was not demonstrated in the Lexington evaluation. However, Clayton and his colleagues (in press) suggest an alternative explanation for the lack of significant findings. They propose the lack of any long-term effects may be due to the fact that the control group was not in a no-treatment condition. Since it is not specified what the standard health curriculum (drug unit) entails, it is certainly possible the control students received similar education and training as that provided by the DARE program.

Recent DARE evaluations demonstrate an improvement in methodology over earlier studies. The three DARE studies described above all use respectable research methodology. Summarizing the results of these studies is somewhat difficult given each one utilizes unique outcome measures such as recognizing media influences and costs and benefits of drug use (Ringwalt, et al., 1991) external locus of control and attitudes towards police (Faine & Bohlander, 1988) and peer relations (popularity among one's peers) (Clayton et al., 1991). However, all three studies do measure drug attitudes, self-esteem, and peer resistance (assertiveness) providing inconsistent results with respect to self-esteem and peer resistance (assertiveness). Findings from Ringwalt et al. (1991), Faine and Bohlander (1988) and Clayton et al.(1991) agree that DARE has an effect on drug attitudes. In all three cases, the treatment (DARE) group had significantly less positive attitudes towards drugs compared to the control group. There is a lack of agreement among all other outcome variables measured.

Although other long-term studies have been attempted, the only one demonstrating adequate methodology is the Lexington study (Clayton et al., in press). Possibly confounded by the lack of a true "no treatment" control group, the results do not warrant program success.

In short, studies of the DARE program have produced mixed results and DARE evaluations up to this point are inconclusive. Further replications are necessary in order make more confident conclusions about the effects of the DARE program.

Several aspects of the DARE program make it a likely candidate for success. First, the program is offered to students just before the age when they are likely to experiment with drugs. Second, although there is little research on the effectiveness of law enforcement personnel as classroom instructors, uniformed police officers serve as teachers of the DARE curriculum in hopes of increasing favorable attitudes towards the law and law enforcement personnel. Third, the DARE program seeks to prevent the use of "gateway drugs" (i.e., alcohol, cigarettes, and marijuana), thereby decreasing the probability of subsequent heavier, more serious, drug use. Fourth, the DARE program draws upon several aspects of effective drug prevention efforts from the "second generation" such as the development and practice of life skills (coping, assertiveness, and decision making).

Although DARE shows promise as a drug prevention strategy, more evaluation efforts need to take place before forming an overall conclusion about the program. This is especially important considering the fact that millions of government dollars are spent on this one particular drug prevention program every year and its dissemination continues to spread rapidly throughout the United States - all without any conclusive evidence concerning its effectiveness.

The purpose of the current study is to evaluate the effectiveness of the DARE program in Charleston County, South Carolina. Specific aims of the program include the stated DARE objectives - increasing self-esteem, assertiveness, coping skills, and decreasing positive attitudes towards drugs, actual drug use, and association with drug using peers. The study also examines the program's effectiveness for reducing other known risk factors associated with adolescent drug use such as social integration, commitment and attachment to school, and rebellious behavior.

The current study uses a nonequivalent control group quasi-experimental design (Campbell & Stanley, 1963) to determine if participating in the DARE program has any affect on the measured outcome variables compared to a similar group that did not receive the program.

The DARE program took place during the Fall and Spring semesters of the 1989-90 school year. A student self-report questionnaire was used to measure the outcome variables. All pre- and post-tests were administered approximately 20 weeks apart.

The survey administration was conducted by the school alcohol and drug contact person. The administration was conducted in such a way as to preserved the confidentiality of the students. All students were assigned identification numbers prior to the time of the pre-test. The identification number was used to link the pre- and post-test questionnaire responses. A questionnaire was distributed in an envelope with the student's name in the top right hand corner. Each name was printed on a removable label which the students tore off and threw away. The administrator read the cover page of the survey informing the students there was a number on the survey booklet which may be used to match their responses with questions asked later. The administrator also informed the students they had the right not to answer any or all of the questions.

Response rates for the sample were high. Table 2 shows pre-test rates range from 79.3% to 98.7%, with an average response rate of 93.7% for the DARE students and 93.7% for the comparison students. An average of 90% of the DARE students and 86.4% of the comparison students completed the post- test. The pre- and post-test (combined) response rates were similar for both groups; 86.7% (295) of the treatment and 83.7% (307) of the comparison students completed both surveys. Statistical analysis procedures were performed to examine the differences between the DARE and nonDARE students. To begin, Analysis of Variance procedures were employed. This type of analysis enables pre- treatment differences on demographic or dependent measures to be detected and subsequently controlled for in later analysis. Controlling for any pre-treatment differences between the two groups and the measured dependent variable on the pre-test, the Analysis of Covariance procedure was used to detect significant differences at the time of the post-test.

Seven hundred eight fifth grade students from eleven elementary schools in Charleston County, South Carolina participated in the present study. Students came from five schools receiving the DARE program and six that did not. Of the 708 students involved in the study, 341 received the treatment (DARE), and 367 served as comparison students. The students came from schools representing a cross section of those found in the Charleston County School District. Three schools are urban, six suburban, and two rural.

Each of the DARE schools was paired with a comparison school based on the following characteristics: Number of students, percent of students receiving free or reduced lunch, percent white, percent male, percent never retained, and percent meeting BSAP (Basic Skills Assessment Program) reading and math standards.

The You and Your School questionnaire was used to measure DARE objectives and other factors associated with later drug use. You and Your School was a preliminary version of What About You? (Gottfredson, 1990), a questionnaire designed to measure drug involvement and risk factors for later drug use.

You and Your School consists of 10 scales and 4 sets of individual questions designed to measure the dependent variables. The ten scales used in the study are: 1) Belief in Prosocial Norms, 2) Social Integration, 3) Commitment to School, 4) Rebellious Behavior, 5) Peer Drug Modeling, 6) Attitudes Against Substance Use, 7) Attachment to School, 8) Self- Esteem, 9) Assertiveness, and 10) Positive Peer Modeling. Sets of individual variables include questions on attitudes about police, coping strategies, and drug use in the last year and last month. Appendix A shows the contents of each scale and the individual items used in the survey.

Scale reliabilities were determined using Cronbach's alpha. Table 4 shows the number of items in each scale and the corresponding reliability coefficients. Reliability coefficients range from .58 for Assertiveness to .85 for Social Integration. Each scale was calculated so that a high score indicates a high level of the factor. For all scales, the items were recoded so that the responses were in the same direction and averaged.

Comparisons were made between the DARE and nonDARE groups to assess initial equivalence on the demographic and outcome variables (see Table 5). Statistically significant pre-treatment differences were found for two of the three demographic measures. The DARE group had significantly more female students (p Three other measures were also shown to be significantly different for DARE and nonDARE students at the time of the pre-test. Before the DARE program began, a higher percentage of the DARE students reported smoking cigarettes in the last year. The DARE group was also found to be less attached to school and believe less in prosocial norms than the comparison group.

Initial analyses compared the DARE and comparison groups on each outcome measure without applying statistical controls for known pre- treatment differences (see Table 6). These analyses revealed differences between only two variables, peer drug modeling and attitudes against substance use, both at the p Other findings demonstrated no effect. DARE and nonDARE students did not differ significantly on the percent reporting cigarette, tobacco, or marijuana use in the last year or frequency of any drug use in the past month. Items targeting coping strategies and attitudes about police were also no different between the two groups. Finally, social integration, commitment and attachment to school, rebellious behavior, and self-esteem scale scores were not significantly different for those in the DARE program than for those not in the program.

In summary, the evidence shows DARE students had more beliefs in prosocial norms, more attitudes against substance use, more assertiveness, and more positive peer associations than the comparison group. The DARE students also reported less association with drug using peers and less alcohol use in the last year. However, the DARE students were equivalent to the nonDARE students on social integration, commitment and attachment to school, rebellious behavior, coping strategies, attitudes about the police, self-esteem, and last year and last month drug use (with the exception of last year alcohol use).

Several factors limit the present study. These are different units of analysis, selection threat due to lack of randomization, and multiple comparisons.

The problem with the unit of analysis is that the treatment and comparison groups were determined by matching schools on specified school characteristics, the program was delivered to classrooms of students, and the analysis was performed at the individual level. The best solution to this problem would have been to randomly assign students to classrooms within schools where some classrooms would receive DARE and others classrooms would not. Assuming this could be done with many classrooms (at least 50), in several different schools, the analysis could be performed using the DARE and control classroom means. This was not possible since the evaluation was conducted after student assignment to classrooms and DARE assignment to schools had already taken place.

Given the random assignment of children into classrooms was not possible, it then would have been better if schools were randomly assigned to receive the DARE program or serve as control schools. This would have decreased a selection threat since currently the argument could be made that the treatment and comparison schools were different to begin with on school characteristics other than those used for matching. As previously mentioned, the decision about which schools received DARE and which did not was determined prior to the beginning of the evaluation.

The last issue worth mentioning is that of multiple comparisons. It is possible the significant outcome effects are overestimated due to the fact that the statistical tests performed were not independent but were treated as such.

The current DARE evaluation demonstrates the program's effectiveness on some of the measured outcome variables but not on others. The current study shows DARE does have an impact on several of the program objectives. Among these are attitudes against substance use, assertiveness, positive peer association, association with drug using peers, and alcohol use within the last year.

It should be noted several of the variables showing no difference between the treatment and control groups are not specifically targeted by DARE (although they are shown to be correlated with adolescent drug use). Among these are social integration, attachment and commitment to school, and rebellious behavior. It could also be argued the DARE program does not specifically aim to change attitudes towards police officers, although this may be a tacit objective. Since the program does not target these outcomes specifically, it may not be surprising there were no differences found between the DARE and nonDARE groups. It was hypothesized the DARE program may impact factors relating to later adolescent drug use even if those factors were not specific aims of the program but this hypothesis did not hold true. In a sense this is evidence that helps to reject the selection argument. If the positive results were due to selection, they would not be found only for the outcomes targeted by DARE.

Much like the three previously reviewed DARE evaluations, the current study adds to the mixed results produced thus far with one exception. Across all studies using a pre-post comparison group design, DARE students' attitudes against drug use have consistently been shown to increase and differ significantly from the control students. Since favorable attitudes towards drug use has been shown to predict or correlate with later adolescent drug use (Kandel, Kessler, & Margulies, 1978) this finding provides some of the most convincing evidence that DARE shows promise as a drug prevention strategy.

On the other hand, there are no other consistent findings for assertiveness (resisting peer pressure), self-esteem, or attitudes towards police. The current study found an increase in assertiveness among the DARE students as compared to the nonDARE students. Ringwalt et al. (1991) and Faine and Bohlander (1988) also found this to be true but Clayton et al. (1991) did not. Effects on self-esteem were not demonstrated in the present DARE evaluation nor were they in Clayton's (Clayton et al., 1991) or Ringwalt's (Ringwalt et al., 1991). However, significant differences in self- esteem were seen for the DARE participants over the controls in Faine and Bohlander's (1988) study. Thus, the Charleston study helps to increase the consistency of the assertiveness and self-esteem results.

Faine and Bohlander's (1988) study also showed positive attitudes towards police were significantly greater for the treatment group than the control group but the present study did not replicate such findings. However, the difference found between these two studies may be due to the measures used. The current DARE study uses only two single item questions to assess students' attitudes about the police whereas Faine and Bohlander (1988) used an 11-item scale. Moderate to high factor loadings (.27 to .82) were reported for each item in the scale, and although the overall reliability was not reported, Faine and Bohlander's (1988) measure of police attitudes is likely to be more valid.

With reference to drug use, all of the stronger DARE evaluations found no effects with the exception of the current study which found a significant difference on last year alcohol use. Clayton's follow-up evaluation showed only one significant difference in the wrong direction on the first of two follow-up post-tests (Clayton et al., in press). As Clayton et al. (in press) points out, the lack of short-term drug use differences may be due to low base rates and thus, should not be interpreted to mean DARE has no effect on adolescent drug involvement.

Replication studies of the evaluation of the DARE program should be continued since mixed evidence exists about the program's overall effectiveness. Conducting randomized experiments would certainly be best for drawing more confident conclusions about DARE program outcomes. Longitudinal studies would also aid in assessing the long-term program goal of deterring adolescent drug use.

There is one large problem with recommending a long-term study on a drug prevention program that is conducted in schools in the United States. The problem involves finding a true "no treatment" control group. Almost every school in the nation has some type of drug education component embodied in the school curriculum which is often mandated by the state. Therefore, it is likely the control group will receive some form of drug education. This problem has been documented as Clayton's (Clayton et al., in press) study used a comparison group that received the school drug education unit and ETI (Evaluation and Training Institute) had to discontinue their 5-year longitudinal study because the entire control group had essentially become a treatment group (Criminal Justice Statistics Association, 1990).

In the future, it may be possible only to compare student's receiving some specified drug prevention program with the school system's drug education unit. However, this appears acceptable if the school system simply requires a unit session on factual drug information or a similar low level intervention since prevention efforts such as this have consistently been shown to have no positive effects (Berberin et al., 1976; Kinder et al., 1980; Schaps, et al., 1981; Tobler, 1986).

Should evaluations of the DARE program continue, it is suggested one national survey instrument be developed and used for all outcome evaluations. Currently, it is difficult to assess whether or not DARE is actually a success since different researchers use different survey instruments to examine a variety of outcome measures. Measuring DARE program objectives and other risk factors associated with later drug use with one survey would enable researchers to compare results across evaluations conducted in U.S. cities and other parts of the world.

Additional recommendations include employing peer leaders (i.e., high school students) as instructors instead of police officers. There are two reasons for this suggestion. First, it has not been consistently demonstrated that attitudes towards police become more positive upon receiving the DARE program, and second, there has been some evidence supporting the use of peer leaders as primary program providers (Arkin, et al., 1981; Botvin & Eng, 1982; Botvin, Baker, Renick, Filazzola, & Botvin, 1984; Perry, Killen, Slinkard, & McAlister, 1980).

It would be not only interesting, but informative, to compare DARE program outcomes utilizing peer leaders vs. police officers as instructors. Should peer leaders provide equal or better outcomes, DARE programming costs would be considerably less and police officers would be more readily available to respond to citizen calls.

It is further recommended that DARE be restructured to incorporate components shown more consistently to be effective such as those found in "second generation" approaches. Although DARE aims to increase resistance skills, coping, and decision-making, the lessons specifically targeting these factors do so in the context of drug use only. As previously mentioned adolescents engaging in drug use behavior are often involved in other problem behaviors (Jessor & Jessor, 1977). It would seem most practical and beneficial to target all of these behaviors utilizing one program as Botvin (1982) and Swisher (1979) have suggested. The DARE program could serve as this one program assuming several changes were implemented.

First, existing components would have to be expanded and additional components added in order to target more broad based adolescent life problems such as family struggles, peer acceptance, sexual involvement, intimate relationships, and effective communication (expressing ideas, listening). Additional sessions should include components from "second generation" programs such as setting goals, solving problems, and anticipating obstacles (Botvin, et al., 1983; Schinke & Gilchrist, 1985).

Second, skill acquisition is said to come about only through practice and reinforcement (Bandura, 1977). It is proposed that any new skills taught, such as problem solving, be reinforced with "real life" homework where students practice these skills in the context of the "real world" rather than simply role playing them in the classroom.

The last recommendation is applicable not only to the DARE program but any drug prevention effort. It involves the addition of booster sessions following the prevention program. Since adolescence is a time of growth, individual attitudes and behaviors may continue to change and develop as the youth is maturing. While short-term evidence of program effectiveness is encouraging, there is no guarantee a youth will continue to practice those same behaviors or hold those same beliefs years, or even months, after the program has ended. In fact, follow-up studies have documented the eroding effects of drug prevention programs (Botvin & Eng, 1980; Botvin & Eng, 1982) and the superior effects of booster session (Botvin et al., 1983; Botvin, et al., 1984). For these reasons, DARE, or any other drug prevention program targeting adolescents, should include a series of follow- up sessions in order to increase the likelihood of sustaining any positive effects.

Charleston County and U.S. High School Seniors' Drug Use - Prevalence Rates(a) for the Class of 1990

	Lifetime		Annual		30 day
Drug	Charleston	US	Charleston	US	Charleston	US
Alcohol	77.2	89.5	68.0	80.6	54.4	57.1
Cigarettes	47.1	64.4	30.4	NA	22.5	29.4
Marijuana	30.6	40.7	22.1	27.0	15.9	14.0
Cocaine	8.7	9.4	6.3	5.3	3.7	1.9
Crack	1.5	3.5	1.0	1.9	0.9	0.7
Hallucinogens	9.0	9.4	7.5/td>	5.9	4.4	2.2
Amphetamines	5.0	17.5	3.4	9.1	2.3	3.7
Sedatives	2.8	5.3	2.0	2.5	1.5	1.0

Reducing The Risk of Drug Involvement Among Early Adolescents: An Evaluation of Drug Abuse Resistance Education (DARE)

by Michele Alicia Harmon, April 1993 Institute of Criminal Justice and Criminology University of Maryland, College Park, MD 20742

by Michele Alicia Harmon, April 1993
Institute of Criminal Justice and Criminology
University of Maryland, College Park, MD 20742