USING THE CUMULATIVE STRATEGIES MODEL FOR DRUG ABUSE PREVENTION: A SMALL GROUP ANALYSIS OF THE CHOICES PROGRAM.

Abstract: IA Cumulative Strategies Model for drug abuse prevention employs a process by which two or more drug abuse prevention strategies are used conjointly, such that each strategy targets two or more factors that place children at high risk for drug abuse. This model was examined using an innovative multiple component program called CHOICES. CHOICES emphasizes the reduction of risk factors that are characteristic of a high risk population of children of alcoholics and predictive of drug abuse. Using a repeated measures multiple baseline design, the results small group analysis indicate that the Cumulative Strategies Model inherent to the CHOICES program is an effective risk reduction strategy for a high risk population of children of alcoholics. Children who received two components of the program showed greater improvement in self esteem, social skills, and attachment to school than children who received one component or no intervention.

It is widely supported in the literature that children affected by familial alcohol abuse are more likely than other children to experiment with, use, or abuse alcohol, tobacco, or other drugs (ATOD) early in life (Ackerman, 1994; Hawkins, Catalano, & Miller, 1992; Sher, Walitzer, Wood, & Brent, 1990; Wilson & Blocher, 1990). Often these children develop a range of affective, behavioral, cognitive, and social problems, many of which are directly associated with an increased risk for drug abuse (Hawkins et al., 1992; Elmquist, Bell, & Associates, 1995; Noll, Zucker, & Fitzgerald, 1992). For example, these children may experience diminished levels of self esteem and negative self-attitudes, and use isolation or withdrawal as a means of coping (Anderson & Quast, 1983; Botvin et al., 1989; Brake, 1988; Goldman & Rossland, 1992; Johnson et al., 1990). Many of these children present poor school readiness and performance, low bonding and attachment to school, and low involvement in extracurricular activities (Hawkins & Catalano, 1992; Gottfredson, 1981; Johnson et al., 1990). Additionally, the dysfunction associated with familial alcohol abuse often results in poor interpersonal and social skills, low social attachment, and favorable attitudes toward ATOD use or abuse among children (Hawkins & Catalano, 1992; Weintraub, 1991; Wilson & Blocher, 1990).

Paralleling a trend toward substance abuse prevention that addresses multiple risk factors in multiple domains, a Cumulative Strategies Model (Horn & Kolbo, 1995; Horn & Kolbo, in press) was developed. This model is grounded in a number of specific principles intended to guide prevention program development. The four principles of intervention that are central to this model include: (1) early vs. later intervention; (2) multiple vs. single risk factors; (3) comprehensive vs. isolated strategies; and (4) multiple vs. single risk-focused strategies. More specifically, successful prevention strategies will: (1) focus on younger children; (2) offer specific risk-focused rather than generalized interventions; (3) address a range of behavioral, psychological, and cognitive risk factors that are common to children from ATOD abusing families and predictive of ATOD abuse, and (4) consist of multiple comprehensive components that target two or more risk factors conjointly. A detailed description of the principle of the Model can be found elsewhere (see Horn & Kolbo, in press).

An evaluation of the Cumulative Strategies Model was possible through the examination of an intervention called CHOICES (Children Having Opportunities in Courage, Esteem, and Success). CHOICES is a multicomponent prevention program (see Horn, 1996; 1997a; 1997b; Horn & Kolbo, 2000 for detailed program information) designed for a high risk population of third and fourth grade children who are affected by familial alcohol abuse. Following the principles and guidelines of the Cumulative Strategies Model, CHOICES is school-based, with intensive intervention directed toward high risk children who have needs that exceed traditional drug abuse prevention programs. While the long term goal of the CHOICES is to prevent the onset of ATOD, there is an equally important investment in the immediate process of promoting happier and healthier childhoods. Thus, the immediate and long term objectives of the program are to: (1) improve self esteem; (2) increase social skills; (3) promote attitudes that are unfavorable to ATOD abuse; and (4) increase attachment to school.

CHOICES is comprised of three components: Structured Support Groups (SSG), Healthy Lifestyle Peer Mentors (HeLP'M), and Academic Channels to Ensure Success (ACES). Only the SSG and HeLP'M components are pertinent to this current study. All components are described in detail elsewhere (see Horn & Kolbo, 2000). The CHOICES SSG curriculum is comprised of 11 sessions. SSGs meet at least 1 hour each week for approximately 11 weeks. Specific sessions include: (1) All About CHOICES; (2) Learning About Feelings; (3)Recognizing Feelings; (4) Expressing Feelings; (5) Knowing Me; (6) Knowing My Family; (7) About Smoking; (8) About Smokeless Tobacco; (9) About Alcohol and Other Drugs; (10) Coping the Best I Can; (11) Deciding for Me. Led by school counselors, meetings take place outside of the general classroom in a place that affords privacy, such as the school counselor's office. The HeLP'M component matches third grade children who are participating in the on-going SSG with fourth grade children who: (1) complete at least nine sessions of the SSG; (2) attend at least three follow up/training sessions; and (3) want to be mentors. HeLP'M provides fourth grade children with an opportunity to enter positive mentoring relationships with third grade peers who are also affected by familial alcoholism. Mentors are trained (by counselors) to provide a positive peer interaction that is comprised of five stages: (1) Passwords; (2) Exchange of Greetings; (3) Feelings Check In; (4) Peer's Choice and (5) Closing. Once they are matched, mentors and mentees are expected to interact for 30 minutes once a week. Meetings are arranged by the children and are approved by their teachers and the school counselor.

Since the assumption of the Cumulative Strategies Model is that two or more strategies are more effective than one in reducing risk, it seems logical that children who receive two or more components of CHOICES experience greater risk reduction than children who only receive one component or none. As such, the study described herein examined the differential effects of two of the components (SSG and HeLP'M) on the four targeted risks factors. The primary hypothesis was that, compared with the other study groups, children in the SSG + HeLP'M training group would demonstrate greater improvement in: (1) self-esteem; (2) social skills;(3) attitudes toward ATOD use (i.e., less favorable attitudes toward use); and (4) attachment to school. The secondary hypotheses were that: (1) children in the SSG would show greater improvement (in 1-4 above) than children who received peer mentoring only and those who received no intervention; and (2) children in the mentor only group would show greater improvement than children who received no intervention.

METHOD

PARTICIPANTS

In this study, 180 third and fourth graders were screened from one elementary school in a WV county. Children were included in the study if they: a) were in third or fourth grade; b) were 8 to 11 years-old; c) attended a group screening session, an interview with the school counselor, and met selection criteria (described in a following section); d) signed an assent form; and e) had parental consent to participate. A total of 44 children (24.5%) screened met the criteria for this study. Parental consent/assent was obtained for 17 children. One child was excluded due to temporary foster care placement and extended absence from school Sixteen (n=16) children completed the study. Individual information data revealed that 81.3% of the children were Caucasian, 6.3% were African American, and 12.5% were biracial. The ages ranged between 8 and 10 years, with an mean age of 6.8 years; 43.8% were male and 56.2% were female. All children resided and attended school in the same county.

At the time of the study, the school had an enrollment of approximately 750 students (K-5) of which about 195 were third (n=95) and fourth graders (n=100). The school's student population encompassed both suburban and rural areas of the county, including varying compositions of race and SES.

PROCEDURE

Screening and selection for target population. Children were screened for CHOICES in the selected school through the use of a trigger video. As part of the school's general drug education unit, all third and fourth grade students in attendance viewed the video. The video, which portrays two young children who live in a family where a parent struggles with alcoholism, was shown and monitored by the school counselor. After the video, children were asked to respond in writing to two screening questions. Question #1 was: "Do you believe that someone close to you has a problem like this?" Question #2 was: "Would you like to be a part of a special program that helps children learn more about the ways alcohol affects kids and families?" Using the screening form, children had the option of circling YES, NO, or MAYBE for each question. Children returned their responses to a "special box" provided by the counselor. Prior to submitting their responses, they were informed that their responses would be reviewed by the counselor later. Children's answers to the questions were not discussed in the classroom. Instead, if a child responded YES or MAYBE to either question, the counselor met privately with the child to further assess the his/her appropriateness for CHOICES and rule out false positive selection. This private interview was held within 7 days of the screening. Consent for participation in the program was obtained by the counselor. A letter and a institutionally approved consent form were sent home to the parent of each child. The letter stated the goals and objectives of the program and the study in a sensitive and unaccusing manner.

DESIGN

A repeated measures multiple baseline small group design (see Figure 1) was employed. Children were randomly assigned to one of four groups: (1) SSG + HeLP'M training (to become mentors) (n=4); (2) SSG only (n=4); (3) peer mentor-matching (PMM) (n=5); and (4) delayed intervention (DI) (n=3). Risk factor scores were measured over three points in time. Study procedures occurred between the months of December, 1996 and June, 1997. See Figure 1.

Figure 1. Repeated measures multiple baseline design, with a coded summary of the baseline, implementation and measurement procedures used for the study.

                12/96     1/97      2/97      3/97

Group 1         01                            02
                SSG       SSG       SSG       SSG

Group 2         01                            02
                --        --        --        SSG2

Group 3         01                            02
                --        --        --        --

Group 4         01                            02
                --        --        --        --

                4/97      5/97      6/97

Group 1                             03
                HeLP'M    HeLP'M    HeLP'M

Group 2                             03
                SSG2      SSG2      SSG2

Group 3                             03
                --        PMM       PMM

Group 4                             03
                --        --        --

01      Observation 1 represents the initial baseline measurement
        point for all groups. At this time, children were
        administered the IIF and primary teachers were administered
        a battery of instruments: ILTS, SSRS-T, and WAIL.
SSG     Approximately 2 weeks after 01, children in group 1 began
        the SSG (11 hour-long weekly sessions).
02      Observation 2 represents the second measurement point for
        all groups. For group 1, this measurement point occurred 2
        weeks after completion of the SSG. Primary teachers were
        administered a battery of instruments: ILTS, SSRS-T, and
        WAIL.
HeLP'M  Group 1 participated in the HeLP'M training (4 hour-long
        sessions) and subsequently began mentoring children in
        group 3.
SSG2    Group 2 began the SSG component (11 hour-long weekly
        sessions) approximately 2 weeks after02.
PMM     Group 3 participants were matched with a mentor (from group
        1).
03      Observation 3 represents the third and final measurement
        point for all groups. Children's primary teachers were
        readministered a battery of instruments at this time: ILTS,
        SSRS-T, and WAIL.
--      This denoted baseline B that is, groups did not receive any
        CHOICES intervention at that time.

DEPENDENT VARIABLES

The four risk factors were examined in this study were assessed using four measures completed by participants' primary teachers. Children completed an individual information form which provided socieodemographic data, specifically participant's gender, age, grade, race, living arrangements, and location of residence. As supported by previous research (Achenbach, McConaughy, & Howell, 1987; Epkins, 1995), this study relied on teacher reports as they were viewed to be more valid and reliable than child reports for targeting and understanding appropriate intervention.

I'd Like to Say (ILTS). This instrument was created specifically for the evaluation of CHOICES and was administered to children's primary teachers at three points across the academic school year (refer to figure 1). The ILTS is a 5-option Likert rating scale intended to help teachers report their observations and perceptions of change in each of the targeted risk factors. ILTS assesses teachers' general perceptions of changes in children from first contact to present time. This instrument includes 18 items with risk factor subscales of: self-esteem, social skills, ATOD attitudes, and attachment to school. Some items relate to more than one risk factor. An extensive review of the drug prevention literature and the expert judgement of prevention specialists suggest that the content of this instrument is highly valid for the purposes of this study. The alpha coefficient for the full scale was .95 on administration. The ILTS: Attitude includes 2 items (range = 0-8) and revealed an alpha coefficient of .82 upon administration. ILTS: Attachment includes 6 items (range 0-24) with a reliability of .90 in the current study. These subscales were used independently to assess ATOD attitudes and attachment to school. The ILTS: Self Esteem subscale includes 8 items, with a range of 0-32. The ILTS: Social Skills subscale includes 8 items, with a range of 0-32. These subscales were used with measures discussed below to obtain aggregate scores of self esteem and social skills.

What Am I Like (WAIL). WAIL (Harter, 1985) is a global self-worth scale. The WAIL (teacher form) allows teachers to report on children's: scholastic competence; social acceptance; athletic competence; physical appearance; behavioral conduct; and global self-worth. The full scale includes 15 items with a range of 0-45. Previously reported reliabilities by Harter (1985) ranged from .71 to .86. In the current study, the WAIL (15 items) and ILTS: Self Esteem (8 items) were combined to obtain an aggregate self esteem score. The aggregate scale included 23 items, with a range between 0-77, and an alpha coefficent of .89 upon administration.

Social Skills Rating System-Teacher (SSRS-T). SSRS-T measures four social skills subdomains: Cooperation, Assertion, Self-Control, and Empathy. Flanagan et al. (1996) report extensive efforts to assure the content validity of the instrument, including factor analysis, correlations with other instruments, and extensive survey of the literature on social skills assessment. SSRS-T includes 57 items. The author does not report the reliability of the instrument. The subscale used in this study included 30 items, with a score range of 0-60. To obtain an aggregate score of social skills, the SSRS subscale (30 items) was combined with the ILTS: Social Skills subscale (8 items). The aggregate scale had 38 items with a range between 0-92, and an alpha coefficient of .91 on administration.

RESULTS

Consistent with recommendations for multiple baseline small group designs (Barlow, Hayes, & Nelson, 1984; Kadzin, 1982), the primary method of data evaluation involved visual inspection. Visual inspection was used to: (1) interpret data patterns; (2) facilitate decisions about whether the data reflected systematic intervention effects; and (3) determine the consistency (i.e., trends and levels) of the effects of the intervention. Statistical analyses were used to supplement visual inspection. Two-way mixed model repeated measures analyses of variance (ANOVA) were performed on participant risk factor scores. Further, consistent with the recommendations of Kazdin (1982), separate t tests were conducted to assess changes in trends and levels of data patterns between any two observation points for each group over time (i.e., O1, O2, and O3) even if no significant interaction was found. Using an alpha of .05, these analyses were used to determine main effects, systematic intervention effects, and differences in data (i.e., trends and levels) patterns over time. Although it is not common practice to use ANOVAs and t-tests together, a combination of these analyes complemented the exploratory nature of this study. All data was analyzed using SPSS? statistical software.

BASELINE DATA

At baseline, children in the study sample were comparable on variables of age (mean=8.8), gender (43.8% male and 56.2% female) and race (81.3% Caucasian). All children resided in the same county; 50% lived with both parents; 31.3% lived with their mother only; and 18.8% lived with their mother and step-father. Upon entering the study, all children reported being drug-free. None of the children had ever tried cigarettes, smokeless tobacco, beer, or wine. All four groups were compared on risk factor scores at baseline. Children's scores on self-esteem, social skills, attitudes toward drug use, and attachment to school were similar upon entering the study.

SELF-ESTEEM

Participant self-esteem scores in Groups 1, 3, and 4 were similar at baseline, with means ranging between 36 and 38 points. See Table 1. The mean score for Group 2 was higher than other groups (M=50.25). Although 75% of the children (12 out of 16) showed some improvement in self-esteem over time, Group 1 participants demonstrated the greatest point gain. Increases in self-esteem were seen at the onset of intervention (after O1) and continued over time. The most striking differences in participants self esteem scores were between Group 1 (M =17.75) and Group 4 (M=4.67). See Figures 2 and 3.

[Figures 2-3 ILLUSTRATION OMITTED]

Table 1 Confirmation of Study Hypothesis

3-Part Hypothesis                             Risk Factor

                                      Self Esteem   Social Skills

SSG + HeLP'M group demonstrated           Yes            Yes
greater improvement than all other
groups

SSG group will show greater               No             No
improvement than PMM and DI
groups

PMM group showed greater                  Yes            Yes
improvement than the DI group

3-Part Hypothesis                             Risk Factor

                                       Attitudes      Attachment

SSG + HeLP'M group demonstrated           No             Yes
greater improvement than all other
groups

SSG group will show greater               No             Yes
improvement than PMM and DI
groups

PMM group showed greater                  Yes            Yes
improvement than the DI group

A 4 x 3 (group x time) mixed model ANOVA was calculated, with the between variable represented by group and the within variable represented by time of administration. A significant main effect was found, F (2,22)= 14.38, p [is less than] .05 for time. A significant group x time interaction, F (6,22) = 4.92, p [is less than] .05, was also found. That is, Group 1 experienced significant gains in self-esteem between O1 and O2 while other groups did not. Paired t tests were also conducted. Group 1 data revealed that significant changes in the trend and level of self-esteem scores occurred between O1 and O2, (t [3] = 3.02, p [is less than] .05), as well as between 02 and 03, Ct [3] = 5.66, p [is less than] .05). Self-esteem scores for Group 3 increased modestly upon intervention between O2 and O3 (t [4] = 3.89, p .05). No statistical significance was found for the data patterns of Group 2 or Group 4.

SOCIAL SKILLS

Data revealed improved social skills scores among study most participants. Specifically, 94% (15 out of 16) of the children showed at least some improvement from O1 to O3. Group mean data indicated that Group 1 demonstrated greater point gain than all other groups. The mean difference score for Group 1 (M= 23.75) was almost three times higher than that of Group 4 (M=8.0). The score increases for Group 1 were followed by Group 3, then by Group 4, and Group 2. See Figures 4 and 5.

[Figures 4-5 ILLUSTRATION OMITTED]

A 4 x 3 (group x time) mixed model ANOVA revealed a main effect for time, F (1,11) = 13.02, p [is less than] .05. No significant group x time interaction was found (F [6,22] = 2.25, p =.07). Upon further exploration of data patterns, Group 1 participant scores increased over time, with the most dramatic change in the level and trend occurring between O2 and O3 (t [3] = 3.30, p? .05). Although slight increases in trends are visually detectable after O2 for Groups 2 and 4, no statistical significance was found. Mean data from Group 3 show a significant increase in trend and level from O1 to O2 (t [3] = 6.55, p [is less than] .05) as well as between O2 to O3 (t [3] = 3.60, p [is less than] .05).

ATTITUDES

At O1, children's attitude scores were comparable, ranging between 2.25 and 3.5 points. Children's scores increased regardless of group over time. Precisely, improvement was observed for 94% (15 out of 16) of the children. Children in Groups 1, 2, and 3 demonstrated similar improvement, with mean difference scores ranging from between 3.5 and 4.1. Participants who received some type of intervention (Groups 1, 2, and 3) demonstrated greater improvement than children in the non-intervention group. Among the 15 out of 16 children who improved, the most potent shift in attitude scores occurred between O1 and O2. Additionally, between O2 and O3, 44% (7 out 16) of the children continued to show improvement. See Figures 6 and 7.

[Figures 6-7 ILLUSTRATION OMITTED]

Group mean data suggest that the scores of children in intervention Groups 1, 2, and 3 accelerated between O1 and O2, then leveled off between O2 and O3. Group 4 scores rose steadily over time, but the mean score difference between O1 and O3 was not as high as other groups. A 4 x 3 (group x time) mixed model ANOVA revealed a significant main effect for time, F (2,22)=76.16, p ? .001. No significant interaction was found. Positive changes in the trends and levels of data between O1 and O3 were significant for Group 1 (t [3] = 7.00, p [is less than] .05), Group 2 (t [3] = 12.12, p [is less than] .05), and Group 3 (t [3] = 17.00, p [is less than] .05). While visual inspection showed a slight gradual improvement for Group 4, the difference was not statistically significant.

ATTACHMENT

Children in groups that received some type of intervention showed improvement on scores of attachment. Group 1 scores rose steadily over time. The mean difference score for Group 1 ranged between three to six times higher than all other groups. See Figures 8 and 9. A two-way mixed model ANOVA revealed a significant main effect for time, F (2,22) = 9.36, p [is less than] .05. A significant group x time interaction, F (6,22) = 9.36, p [is less than] .05 was also found. Most strikingly, data patterns revealed a significant positive change in trend and level from O1 to O2 for participants Group 1, (t [3] = 4.52, p? .02) and Group 3 (t [3] = 3.58, p? .04). Paired t tests found no statistical significance for changes in attachment scores of Group 2 and Group 4.

[Figures 8-9 ILLUSTRATION OMITTED]

DISCUSSION

SELF-ESTEEM

Teacher reports indicated that most children showed improvement in global self-esteem across the academic year. As expected, however, participants in SSG+HeLP'M intervention showed greater improvement than all other participants. Also consistent with study predictions, children in the other intervention groups remained stable or dropped in self-esteem during baseline and demonstrated some improvement upon receiving intervention. Participants who received no intervention did not change significantly on self-esteem over time. Overall, findings suggest that SSG+HeLP'M intervention was more effective in improving self-esteem than all other interventions. Contrary to the secondary hypotheses, results indicated that the second most effective intervention for self-esteem was the PMM, followed by the DI and SSG groups, respectively. Interestingly, a slight loss in self-esteem was observed for the SSG Group 2 upon intervention. Although this loss was marginal (-2 points) and not statistically significant, a decrease in self-esteem warrants discussion. A loss in self-esteem upon intervention has several explanations. First, as a result of intervention, children may have developed an increased awareness of the external stressors in their home environments. One objective of the program was to help children understand that they cannot cause or control the dysfunction in their homes rendered by familial alcoholism. A loss in self-esteem may occur as children come to terms with this concept.

An additional explanation involves the composition of the SSG group. Despite random assignment, the SSG (Group 2) was comprised of all females. Research shows that girls more often than boys experience decreased self-esteem, particularly as they move toward adolescence (Bolognini et al., 1996). Although the SSG intervention did not increase self-esteem among these girls, it may have prevented the event of decreased functioning and/or the emergence of additional risk factors. It is possible that the intervention served as protection against a decrease in self-esteem for the girls in the SSG group. This notion is consistent with theory on psychosocial competence (Noll et al., 1992; Rutter, 1979), but requires further research.

SOCIAL SKILLS

The SSG+HeLP'M group, as predicted, showed greater improvement than all other groups on social skills scores. Although the scores of the SSG+HeLP'M children increased steadily over time, results suggest that the greatest improvement occurred during the time that children received the SSG component. Also, consistent with expectations, the social skills scores of children in the other groups remained stable or decreased slightly during the baseline period. Social skills were also positively impacted by the PMM intervention. Unexpectedly, no differential effects on social skills were found between the SSG intervention and the DI group. These findings are consistent with other cognitive-behavioral training skills programs for children (Botvin et al., 1995; Schinke & Gilchrist, 1985) that resulted in improvement on measures of social skills (e.g., problem solving, decision making, and assertiveness). Moreover, an examination of the Life Skills Training model used by Botvin and colleagues (1995; 1980) showed that children who experienced improved scores of social skills were less likely to smoke.

ATTITUDES

Results imply that children's attitude toward drug use became increasingly unfavorable over time. Children who received some type of intervention (SSG+HeLP'M, SSG, or PMM) showed greater improvement than children who received no intervention. Although the ATOD attitudes of children in the intensive intervention showed positive changes over time, changes were not significantly different from changes observed in other children, including the wait-list control (DI) group. Different from reports of other prevention programs (Hansen, 1992), differences between groups were not visually or statistically potent enough to determine that the SSG+HeLP'M or any other level of intervention significantly affected children's attitudes toward ATOD.

Historically, main-stream prevention programs were largely didactic and used scare tactics to alter children's attitudes and dissuade them from using drugs. In recent years, prevention efforts have moved toward changing attitudes using normative beliefs about ATOD consumption (Hawkins et al., 1992). For instance, current comprehensive school-based prevention programs and curricula emphasize normative information, with a major objective of dispelling the myth that drug use is prevalent and acceptable among youth. Children begin receiving these general prevention messages in early elementary school. It was evident upon entering the study that children already held moderately unfavorable attitudes toward drug use. Although the CHOICES program includes normative information, it was not clear that CHOICES had a significant impact on ATOD attitudes. Nonetheless, given high scores at baseline, steady improvement, and constancy among groups over the course of the study, it appeared that children had received effective normative information and general education on some domain.

ATTACHMENT TO SCHOOL

Children who received some type of intervention showed improved attachment to school. As anticipated, children who received the SSG+HeLP'M intervention experienced the most impressive gain in attachment. Also, as expected during the baseline period the attachment scores for other groups remained stable. Consistent with the hypotheses, other groups experienced improved scores after intervention, except the DI group which remained stable over time. Specifically, for attachment, the most effective intervention was the SSG+HeLP'M, followed by the SSG, PMM, and DI respectively. These findings support the foundation of the Cumulative Strategies Model and are consistent with findings from other prevention program evaluations (Hawkins et al., 1992; Hansen, 1992; Elias et al., 1991). A focus on school attachment in prevention programs is crucial (Johnson et al., 1988). In fact, a recent study reported that a connection to school protects youth from cigarette, alcohol, and marijuana use (Blum & Rinehart, 1997).

INTERVENTION COMPARISONS

Using teacher reports as the primary measurement source, results suggest that children who received two intervention strategies (SSG+HeLP'M) showed greater risk reduction than children who received one intervention (SSG or PMM) or none. Not surprisingly, the most striking differences were between the SSG+HeLP'M and DI groups. Specifically, compared with other children, those who received the more intensive intervention (SSG+HeLP'M) showed the highest gains in self-esteem, social skills, and attachment to school. Although children's ATOD attitudes improved, changes did not appear to be the result of any component of the program. Because the study hypothesis has multiple components, a table format is used to simplify the confirmation of each part of the hypothesis. See Table 1.

Contrary to expectations, children who received the PMM intervention experienced higher point gain on all risk factors than children in the SSG group. Similar to children in the intensive SSG+HeLP'M intervention, children who were matched with a peer mentor (PMM) showed significant gain in self-esteem, social skills, and attachment to school. These findings imply that the PMM intervention, although not as effective as SSG+HeLP'M, may be an effective intervention in its own right or perhaps more effective than the SSG alone.

Interestingly, upon close examination of the data, the common factor in the two most effective interventions (SSG+HeLP'M and PMM) in this study is the involvement of peers, suggesting that both mentors and mentees benefitted from a peer approach. This finding is consistent with other studies (Benard, 1990; Botvin et al., 1989:Tobler, 1986) that suggest the use of peers as a powerful drug abuse prevention tool. To illustrate, Botvin et al. (1989) found that children who participated in a peer led intervention were significantly different from controls on tobacco, alcohol, and marijuana use. Despite that fact that the PMM intervention was not as intensive as the SSG+HeLP'M in duration, session length, or use of resources, risk factor scores were significantly effected. Similarly, Tobler (1986) found programs involving peers were dramatically more effective, even at lower intensities. Benard (1990) reported that when intensity or duration is higher, peer program effects are more pronounced. Although the findings of the current study do not confirm the isolated effects of the HeLP'M or PMM components on mentor or mentees, the above rationales suggest that the impact of peers on prevention programming needs further examination.

Although the argument for peers is powerful, several additional issues should be considered when making comparisons of the effectiveness of the SSG and the PMM interventions. First, the SSG used in Group 2 was not carried out with the same fidelity as in Group 1 (SSG+HeLP'M). That is, time constraints and scheduling difficulties limited the group to 9 versus the regularly scheduled 11 sessions. Thus, some might argue that children in the Group 2 SSG did not receive the full impact of the program. Second, as noted previously, the participants in Group 2 were all females. Overall, this group did not show as much improvement as other children who received intervention. In fact, for social skills and self-esteem, their point gains were lower than the DI group. While systematic gender effects were not detected in this study, the female composition of Group 2 along with their inconsistent performance cannot go unmentioned, It might be possible that the performance of the girls in the SSG group had more to do with the facilitator than the intervention itself. For instance, a group comprised of all females may not respond optimally when an intervention is administered by a male facilitator. To illustrate, a study by Bolognini et al. (1996) found that self-esteem was most impacted for girls when intervention was delivered by a same sex peer.

Another important and more plausible issue involves outlying scores. One participant in Group 2 displayed outlying scores on self-esteem and social skills. Another participant had extreme scores of self-esteem, social skills, and attachment. More precisely, these participants had baseline scores higher than other participants in the SSG group (and all other groups). High scores created a ceiling effect, leaving little room for improvement for these two participants. Further, because of the outlying scores the group means may not accurately reflect the impact of the SSG, particularly for self-esteem and social skills.

STUDY LIMITATIONS THREATS TO INTERNAL VALIDITY

Several steps were taken to counter the threats to internal validity. First, children were randomly assigned to intervention groups. Second, groups were administered measures and intervention in separate settings, and, finally, contact between groups was minimized. As with any study, however, it is possible that factors other than the intervention potentially attributed to the results of the study. Upon interpretation of the data for this study one threat to internal validity warrants further discussion.

Selection Biases. The CHOICES program has undergone four years of pilot evaluation. As a result, the screening and selection processes have been refined and standardized such that it is extremely sensitive to the appropriate selection of the target population. Nonetheless, because the initial selection is based on children's self-report, the event of false positive identification (i.e., selecting unaffected children for the program) must be considered. Precautions were taken to reduce the event of false positives. Children who said that (1) alcohol was not a problem in their families and (2) they did not want to be in the program were not selected. Further, if children expressed that (1) alcohol was not a problem in their families, but (2) they wanted to be in the program, they were not selected. All maybe responses were closely assessed by the counselor in a private interview and eventually classified as responses of either yes or no based on the operational definition of "affected by familial alcoholism" as used in this study. Previous pilot evaluation of CHOICES revealed that false positive identification is an unlikely event and the same is expected for children who participated in this study. However, as is the case with any self-report data, there are no guarantees that selection procedures were 100% accurate.

Another selection issue involves the parental consent. The study sample reflects only children who had parental consent to participate in an intervention. Thus, some experts might argue that the sample is biased toward children who have cooperative parents or parents who are emotionally ready to confront the issue of familial alcoholism. On the other hand, it is important to entertain the possibility that non-consenting parents did not want their children to be a part of a "research study." Parents may have been more responsive if CHOICES were offered as a "service" rather than research.

THREATS TO EXTERNAL VALIDITY

Fidelity of implementation. In effort to increase the fidelity of implementation and overall generalizablity of the program, several standards are inherent to the CHOICES program. First, each component of CHOICES has a highly scripted curriculum for facilitators (i.e., a session-by-session guide). Second, program standards require that facilitators receive extensive training (a 2-day training minimum). Third, it is also recommended that the program be implemented in school settings by school counselors. Every effort was made to assure that these program standards were met in this study. Despite these efforts, the fidelity implementation of the SSG in Group 2 was not fully obtained.

According to the process information collected from the facilitator (school counselor) who participated in this study, the SSG+HeLP'M component as well as the PMM component were delivered as suggested in the curriculum. The SSG used in Group 2, on the other hand, was not delivered in entirety. Because of time limitations and scheduling difficulties, children in Group 2 SSG received only 9 out of the recommended 11 sessions. Therefore, children in the Group 2 SSG may not have received the full impact of the program and results should be interpreted accordingly. As discussed previously, children in this group did not respond as positively as children in other intervention groups.

Sampling. Another issue of external validity involves sampling. Because the study sample was drawn from one elementary school, limited in number (n=16), and largely Caucasian (81.3%), it is not clear whether the sample represents US children affected by familial alcoholism. Although previous feasibility evaluation of the program (1993-1995) with children in at least 11 additional schools suggested similar data patterns, future research is needed to ascertain the generalizablity and reliability of the intervention across other conditions and subgroups, including larger samples of suburban non-Caucasian populations of varying SES.

CONCLUSION

The current study entailed the first exploratory evaluation of the Cumulative Strategies Model, employing a process by which two or more drug prevention strategies are used conjointly in multiple domains, such that each strategy targets two or more factors that place children at high risk of drug abuse (Horn & Kolbo, 1995; Horn & Kolbo, 1998). Given the presence of multiple and synergistic risks, the Cumulative Strategies Model as demonstrated by the study of CHOICES may be an effective drug abuse prevention approach for a high risk population of children of alcoholics.

Results from this study are promising. Following the Cumulative Strategies Model, children who participated in two components of CHOICES (compared with those who received one or none) showed significant improvement in self-esteem, social skills, and attachment to school. Gains in these areas may dramatically reduce the likelihood that these at-risk children will enter patterns of ATOD abuse later in life. Although drug abuse prevention efforts continue to improve on numerous fronts, many gaps remain in knowledge and practice. Future research in needed to explore: (1) the optimal age at which to deliver intervention; (2) the optimum change agent (e.g., counselors, peers, teachers, parents) in program delivery; (3) the characteristics of the optimum change agent; (4) the involvement of systematic family intervention; and (5) the impact of gender, ethnicity, and class on intervention effectiveness. Most importantly, long term tracking of children who receive intervention is needed to determine whether or not drug use is actually prevented or delayed.

For the Cumulative Strategies Model and CHOICES in particular, additional research is needed to: (1) explore program effectiveness with a larger heterogenous sample (including suburban and rural populations); (2) determine the differential effectiveness of three components of CHOICES (SSG, HeLP'M, and ACES); and (3) examine the role of peers as change agents. Additionally, programs other than CHOICES need to be examined to further understand and substantiate the Cumulative Strategies Model. Further exploration of this may provide important information about "what, when, and how" of prevention service delivery in practice settings. Drug abuse prevention is a complex and challenging task, but it is an extremely worthwhile investment of time and resources. Despite the fact that most children do not develop addictions to alcohol, tobacco, or other drugs, those who do often experience serious psychological and behavioral consequences in childhood and throughout life. Programmatic efforts to prevent or at least delay the onset of drug abuse among children are crucial if we are to improve the overall health, safety, and well being of US children.

REFERENCES

Achenbach, T.M., McConaughy, S.H., & Howell, C.T. (1987). Child/adolescent behavioral and emotional problems: Implication of cross-informant correlations for situational specificity. Psychological Bulletin, 101, 213-232.

Ackerman, R. (1994, May). Promoting resiliency in children and adolescents: Strategies for preventing alcohol and tobacco use. In Piecing Together the Future conference in Morgantown, WV under Cooperative Agreement Number 5207A30171, US Department of Education. Prevention Connection, Bowling Green State University, Bowling Green, OH.

Anderson, E., & Quast, W. (1988). Young children in alcoholic families: A mental health needs-assessment and an intervention/prevention strategy. Journal of Primary Prevention, 3(3), 174-187.

Barlow, D.H., Hayes, S.C., & Nelson, R.O. (1984). The scientist practitioner: Research accountability in clinical and educational settings. Elmsford, NY: Pergamon Press.

Benard, B. (1990). The case for peers. Louisville, KY: Southeast Regional Center for Drug-Free Schools and Communities.

Blum, R.W., & Rinehart, P.M. (1997). Reducing the risk: Connections that make a difference in the lives of youth. (Available from the Division of General Pediatrics and Adolescent Health, University of Minnesota, Box 721, 420 Delaware St., S.E., Minneapolis, MN, 55455).

Bolognini, M., Plancherel, B., Bettschart, W., & Halfon, O. (1996). Self-esteem and mental health in early adolescence: Development and gender differences. Journal of Adolescence, 19, 233-245.

Botvin, G.J., Batson, H.W., Witts-Vitale S., Bess, V., Baker, E. & Dusenbury, L. (1989). A psychosocial approach to smoking prevention for urban Black youth. Public Health Reports, 104, 573-582.

Botvin, G.J., Baker, E., Dusenbury, L., Botvin, E. & Diaz, T. (1995). Long-term follow-up results of a randomized drug abuse prevention trial in a white middle-class population. Journal of the American Medical Association, 273 (14), 1106-1112.

Botvin, G.J., Eng A., & Williams, C.L. (1980). Preventing the onset of cigarette smoking through life skills training. Preventive Medicine, 9, 135-1433

Brake, K. J. (1988). Counseling young children of alcoholics. Elementary School Guidance and Counseling, 23, 106-111.

Dielman, T. E., Butchart, A. T., & Shoppe, J. T. (1993). Structured equation model tests of patterns of family interaction, peer alcohol use, and intrapersonal predictors of adolescent alcohol use and misuses. Journal of Drug Education, 23, 273-316.

Elias, M.J., Gara, M.A., Thomas, F., Schuyler, M.A., Branden-Muller, L.R., & Sayette, M.A. (1991). The promotion of social competence: Longitudinal study of a preventive school-based program. American Journal of Orthopsychiatry, 61 (3), 409-417.

Elmquist, D.L., Bell, T. & Associates (1995). Alcohol and other drug use prevention for youths at high risk and their parents. Education and Treatment of Children, 18 (1),65-88.

Epkins, C.C. (1995). Teachers' ratings of inpatient children's depression, anxiety, and aggression: A preliminary comparison between inpatient-facility and community-based teachers' ratings and their correspondence with children's self reports. Journal of Clinical and Consulting Psychology, 24 (1), 63-70.

Flanagan, D.P., Alfonso, V.C., Primavera, L.H., Povall, L., & Higgins, D. (1996). Convergent validity of the BASC and SSRS. Psychology in the Schools, 33, 13-23.

Goldman, B.M., & Rossland, S. (1992). Young children of alcoholics: A group treatment model. Social Work in Health Care, 16 (3) 53-56.

Gottfredson, G.D. (1981). Schooling and delinquency. In S.E. Martin, L.B. Sechrest, & R. Redner (Eds.), New Directions in the Rehabilitation of Criminal Offenders. Washington, DC: National Academy Press.

Hansen, W.B. (1992). School-based substance abuse prevention programs: A review of the state of the art in curriculum, 1980-1990. Health Education Research, 7,403-430.

Harter, S. (1985). The Self-Perception Profile for Children: Revision of the Perceived Scale for Children. University of Denver: Author.

Hawkins, D.J., Arthur, M., & Catalano, R.F. (1994). Preventing substance abuse. JOB: Crime & Justice, 532,197-278.

Hawkins, D.J. & Catalano, R.F. (1992). Communities that Care. San Francisco, CA: Jossey-Bass Publishers.

Hawkins, D.J., Catalano, R.F. & Miller, J.Y. (1992). Risk and protective factors for alcohol and other drug problems in adolescent and early adulthood: Implications for substance abuse prevention. Psychological Bulletin, 112 (1), 64-105.

Hawkins, W.E., Clarke, G.N., & Seely, J.R. (1993). Application of social learning theory to the primary prevention of depression in adolescents. Health Values, 17, 31-39.

Horn, K. (1996). CHOICES: A structured support group curriculum for children at risk for alcohol, tobacco, or other drug abuse. Morgantown, WV: Pepper Press, Inc. (Available from the Prevention Research Center, West Virginia University, PO Box 9005, Morgantown, WV, 26506).

Horn, K. (1997a). Academic Channels to Ensure Success (ACES): A Guide for Teachers Working With High Risk Children. West Virginia University Printing and Publishing: Morgantown, WV. (Available from the Prevention Research Center, West Virginia University, PO Box 9005, Morgantown, WV, 26506).

Horn, K. (1997b). Training Children At Risk for Drug Abuse As Healthy Lifestyle Peer Mentors (HeLP'M): A Facilitators Guide. West Virginia University Printing and Publishing: Morgantown, WV. (Available from the Prevention Research Center, West Virginia University, PO Box 9005, Morgantown, WV, 26506).

Horn, K., & Kolbo, J. (2000). Application of a Cumulative Strategies Model for drug abuse prevention: Exploring CHOICES for high risk children. Journal of Drug Education Volume 30(3), 289-310.

Horn, K., & Kolbo, J. (1995). Risk based primary prevention strategies for elementary age children of alcoholics: A resource for social workers in the school setting. School Social Work Journal, 19 (Spring), 1-14.

Johnson, C., Pentz, M., Weber, M., Dwyer, N., Baer, N., MacKinnon, D.P., &Hansen, W.B. (1990). Relative effectiveness of comprehensive community programming for drug abuse prevention with high-risk and low-risk adolescents. Journal of Consulting and Clinical Psychology, 58 (4), 447-456.

Johnson, E., Amatetti, S., Funkhouser, J.E., & Johnson, S. (1988), Theories and models supporting prevention approaches to alcohol problems among youth. Public Health Reports, 103 (6), 578-585

Kazdin, A.E. (1982). Single case research designs: Methods for clinical and applied settings. Oxford, NY: Oxford University Press.

Noll, R.B., Zucker, R.A., & Fitzgerald, H.E. (1992). Cognitive and motoric functioning of sons of alcoholic fathers and controls: The early childhood years. Developmental Psychology,

Rutter, M. (1979). Protective factors in children's responses to stress and disadvantage. In M.W. Kent and J.E. Roll (Eds.), Primary Prevention of Psychopathology, (pp. 49-74). Hanover, NH: University Press of New England.

Schinke, S.P., & Gilchrist, L.D. (1985). Preventing substance abuse with children of alcoholics. Journal of Consulting and Clinical Psychology, 53, 596-602.

Sher, K.J., Walitzer, K.S., Wood, P.K., & Brent, E.E. (1990). Characteristics of children of alcoholics: Putative risk factors, substance use and abuse, and psychopathology. Journal of Abnormal Psychology, 100 (4), 427-448.

Tobler, N. (1986). Meta-analysis of 143 adolescent drug prevention programs: Quantitative outcome results of program participants compared to control or comparison group. Journal of Drug Issues, 18 (4), 537-567.

Weintraub, S. A. (1991). Children and adolescents at risk for substance abuse and psychopathology. The International Journal of Addictions, 25 (4a), 481-494.

Wilson, J., & Blocher, L. (1990). The counselor's role in assisting children of alcoholics. Elementary Guidance and Counseling, 23, 98-105.

Kimberly A. Horn, Ed.D., Research Assistant Professor, West Virginia University, Department of Community Medicine and Prevention Research Center, [e-mail]. Jerome R. Kolbo, University of Southern Mississippi.