If It's Offered, Will They Come? Influences on Parents' Participation in a Community-Based Conduct Problems Prevention Program

Abstract

This study examined influences on the rate and quality of parent participation in the Fast Track Program, a multi-system, longitudinal preventive intervention for children who are at risk for conduct problems. A theoretical model of the relations among family coordinator characteristics, parent characteristics, the therapeutic engagement between family coordinator and parent, and rate and quality of parent participation was the basis for this study. “Family coordinators” are the Fast Track program personnel who conduct group-based parent-training sessions and home visits. Participants in this study included 12 family coordinators (42% were African American, 58% European American) and 87 parents (55% were African American, 45% European American). The level of therapeutic engagement between the parent and the family coordinator was positively associated with the rate of parent attendance at group training sessions. The extent of family coordinator-parent racial and socioeconomic similarity and the extent of the family coordinator's relevant life experiences were highly associated with the level of therapeutic engagement. The quality, but not the rate, of participation was lower for African American parents. Implications of these findings for preventive intervention with this population are discussed.

Researchers have come to view the etiology of conduct problems as a multiply determined developmental process (e.g., Greenberg, Spelz, & DeKlyen, 1993), with ineffective parenting as a prominent cumulative negative influence (e.g., Patterson, 1982; Patterson, Reid, & Dishion, 1991). Accordingly, interventions for children at risk for conduct problems have become more comprehensive and longitudinal, and now include group-based parent-skills training and home visits to promote parent empowerment (e.g., Conduct Problems Prevention Research Group, 1992; Miller & Prinz, 1990). In these interventions, the active participation of parents in parent skills training is necessary for successful child outcomes (Patterson & Chamberlain, 1988). However, parent drop-out rates are as high as 50%, posing a major obstacle to the success of family-centered interventions. Thus, the nature of and influences on parent participation in interventions for childhood conduct problems must be understood. As Prinz and Miller (1996) have noted, parent participation comprises not only rate (i.e., frequency of attendance and dropout), but also quality (i.e., parent engagement with the group and group leader during training sessions). This paper reports on a study of influences on parent participation in an intervention for children at risk for conduct problems.

Influences on the rate and quality of parent participation in community-based interventions may derive from several sources. Aside from situational constraints (e.g., unreliable transportation, competing obligations), other important influences are theorized to be: (a) the nature of the relationship between therapist and parent; (b) parent and family characteristics (e.g., Prinz & Miller, 1996); and relatedly, (c) the extent to which the parent likes the program and finds relevance in its content. The significance of these potential influences on parent participation is likely heightened for prevention programs in which children are selected via school-based screening, services are delivered in a school/community setting, and, most important, parents do not initiate the request for help themselves. Thus, prevention programs must include services that will minimize the impact of situational constraints and maximize parent participation. Indeed, prevention research efforts offer ancillary services such as transportation, child care, and compensation for attendance (e.g., Capaldi & Patterson, 1987; CPPRG, 1992). Primary features of these prevention projects include home visits to facilitate parent empowerment and parent-group training sessions to reduce coercive parent-child interaction cycles. These services are typically delivered by personnel who have therapeutic and case management experience with high-risk populations.

In the Fast Track program, the preventive intervention on which the present study is based, these personnel, called “family coordinators,” provide a mixed model of service delivery, conducting both parent-group training sessions and home visits (CPPRG, 1992). In this mixed model, parent-group sessions typically involve six to eight parents who meet weekly with the family coordinator to learn new parenting alternatives. The family coordinator also develops an individual relationship with the parent via home visits that focus on parent empowerment and consolidation of parenting skills learned in parent-group training sessions.

The quality of the therapeutic relationship that develops between the family coordinator and the parent may be a primary influence on the parent's attendance of group sessions as well as the parent's level of interest and involvement with the group and family coordinator during these training sessions. Indeed, this is suggested by research in traditional individual therapy which shows that the quality of the therapist–client working alliance predicts client participation (e.g., Bordin, 1979; Horvath, 1994; Strupp, 1973a, 1973b).

Although research on the working alliance in individual therapy has been complicated by several challenges, including construct definition, timing of measurement, and choice of respondent (Horvath, 1994), three common definitional elements have emerged: (a) the therapist's contribution (e.g., nonspecific qualities such as respect, positive regard, and specific qualities such as intervention technique or experience; (b) the client's contribution (e.g., participation or resistance); and (c) a collaborative element (e.g., Gomes-Schwartz, 1978; Marziali, 1984; Kokotovic & Tracey, 1990). The literature suggests that, in individual therapy, these three elements contribute equally to a successful working alliance (e.g., Bordin, 1979). However, this is unlikely to be the case for the therapeutic relationship between the family coordinator and the parent in community-based interventions, such as Fast Track, because these interventions differ from individual therapy in important ways. These differences include: (a) the target population—parents of children who are at risk for conduct problems tend to face multiple ongoing life stressors such as unemployment, single parenthood, living in high-risk neighborhoods; (b) the referral source— these parents typically do not self-refer, and in fact may be recruited explicitly for the intervention; (c) the identified client—the parent is not the identified client; (d) the therapeutic modality—services are delivered in a mixed-model combining home visits and parent–group sessions; (e) the therapeutic services–the intervention contains a substantial didactic component; (f) the therapeutic goals—the therapeutic goals are not mutually agreed upon but are prescribed by the project; and (g) the time frame—the intervention may extend over several years.

It is argued that, in comparison to the therapist's contribution to the working alliance in individual therapy, the family coordinator is faced with a more daunting challenge and charged with greater responsibility for the establishment and maintenance of an effective therapeutic relationship. In fact, based on their individual and family work with parents whose children are at risk for conduct problems, Patterson, Chamberlain, and colleagues strongly suggest that the therapist's contribution to the relationship may be the most critical component in the rate, and perhaps also the quality, of parent participation (e.g., Patterson & Chamberlain, 1988; Chamberlain, Patterson, Reid, Kavanagh, & Forgatch, 1984; Patterson & Forgatch, 1985; Chamberlain & Baldwin, 1987; Patterson & Chamberlain, 1994).

It appears that this particular population—parents of children at risk for conduct problems—presents its own unique challenges to the therapist or family coordinator. Patterson and others have shown that, in samples similar to the Fast Track sample, three out of four parents of children with conduct problems exhibit significant resistance to the intervention (e.g., Patterson & Chamberlain, 1988; Chamberlain et al., 1984; O'Dell, 1982). Acknowledging that resistance may be a natural part of the change process, Patterson and his colleagues emphasize that resistance in this population should be of particular concern for several reasons. First, the incidence and valence of resistance of these parents exceed what is typical. Second, parents' resistance is an inverse curvilinear function of the intervention, with greatest resistance occurring during the most critical phase of the intervention: when the therapist is teaching parents the very skills that are necessary to effect change in their antisocial children's behavior (Patterson & Forgatch, 1985; Patterson & Chamberlain, 1988, 1994). It is during this phase of treatment that parents evince the characteristic “I can't” and “I won't” symptoms of resistance (Patterson & Chamberlain, 1988, p. 203) that tend to undermine therapists' motivation and effort to remain engaged. When therapists' disengagement results, it is manifested in decreased liking and support for the parent, withdrawal of effort to teach new skills and to confront the parent about his or her resistance, and loss of interest in the parent's progress and even in whether or not the parent keeps appointments, thereby “letting the case fail, or letting the case drop out of treatment” (Patterson & Chamberlain, 1988, p. 194). If therapists disengage, parents will not learn the requisite parenting skills and the coercive family patterns that characterize the development of children's conduct problems will persist. In sum, if therapists disengage, the intervention will fail to produce change in the target child population (Patterson & Chamberlain, 1988).

The Family Coordinators' Level of Engagement in The Therapeutic Relationship

In the present study, it was hypothesized that family coordinators' level of engagement with parents during the period of the intervention when parent resistance was likely to be greatest (herein referred to as “family coordinators' therapeutic engagement”) would predict the rate of parents' attendance (e.g., Patterson & Forgatch, 1985; Chamberlain & Baldwin, 1987; Patterson & Chamberlain, 1988, 1994) as well as parents' interest and involvement with the group and family coordinator during training sessions. The work by Patterson and colleagues suggests, moreover, that family coordinators' therapeutic engagement would comprise several factors. According to Patterson and Chamberlain (1988, 1994), family coordinators are at risk of interpreting parents' resistance as personal rejection. However, if they recognize that parents' resistance is a normative part of the change process and retain their objectivity toward parents, they will more likely continue to serve parents effectively. Thus, one important component of family coordinators' therapeutic engagement is their beliefs about the extent to which they are respected, liked, and trusted by parents.

Patterson and Forgatch (1985) also suggest that if family coordinators remain empathic in the face of parents' resistance, resistance is attenuated and family coordinators' therapeutic engagement is enhanced. Thus, a second component in family coordinators' level of therapeutic engagement is the difficulty they experience in relating to parents and in remaining supportive of parents during the period of greatest resistance.

As has been noted earlier, parents' resistance escalates when therapists are teaching them new skills and confronting them about maintaining consistency in their use of these skills (Chamberlain & Baldwin, 1987). This is the period when therapist disengagement is most likely to occur. Thus, a third component in family coordinators' therapeutic engagement is their ability to remain effective—despite parents' resistance—both in teaching parents the requisite skills to effect change in their children's antisocial behavior and in helping parents to maintain a fundamental sense of competence.

Family coordinators' level of therapeutic engagement was one of several influences posited to predict parents' participation. It was also hypothesized that individual characteristics of family coordinators and of parents would be associated with the rate and quality of parents' participation in the intervention. Moreover, it was expected that specific demographic characteristics, as described below, would predict, in part, the level of family coordinators' therapeutic engagement with parents.

Determinants of Family Coordinators' Level of Therapeutic Engagement

The following factors were hypothesized to influence family coordinators' level of therapeutic engagement: (a) the demographic similarity between the family coordinator and the parent; and (b) the family coordinator's life experiences that are relevant to the contextual circumstances of the target population. Although early work on cross-cultural counseling pointed to the client's demographic background as an influence on the therapeutic relationship (e.g., Padilla, Ruiz, & Alvarez, 1975), the critical factor more likely was the demographic match between client and therapist (e.g., S. Sue, Allen, & Conaway, 1975).

More recently, scholars of cross-cultural counseling (e.g., Pinderhughes, 1989; D. Sue & D. W. Sue, 1990) note that the clinician's sensitivity to the client's cultural background is critical to therapeutic effectiveness. Contributing to this sensitivity may be direct exposure to the types of life circumstances that characterize the contexts for high-risk parents, or indirect exposure to a variety of experiences, including reading books, viewing films or undergoing formal training. Although both types of experiences contribute to clinician sensitivity, direct exposure to relevant life experiences may be a more salient facilitator of sensitivity. Furthermore, as researchers of racial and cultural identity note, one's own identity development as a racial or cultural being influences clinician effectiveness with clients who are culturally different (e.g., Helms, 1984; D. Sue & D. W. Sue, 1990). Critical influences in cultural identity development are the cross-cultural experiences that one has had and the degree to which they have been incorporated into one's sense of self and awareness of others (Hardiman, 1982; Helms, 1984; Ponterotto, 1988). Thus, family coordinators' life experiences that are relevant to the population of parents whose children display conduct problems warrant examination.

In this study, both demographic similarity and a history of relevant life experiences were examined for their association with family coordinators' level of therapeutic engagement and with parents' participation. Specifically, it was hypothesized that this similarity between the family coordinator and parent, as well as the family coordinator's relevant life experiences, would enhance the family coordinator's level of therapeutic engagement which, in turn, would positively influence the parent's participation.

Parent Characteristics

Parent characteristics that may predict participation rate include parental stress, parent report of child problem behaviors, parent sense of competence and parent demographics. Parental stress is negatively associated with rate of participation in treatment. Parents with high levels of stress fail to continue in clinic-based services (Kazdin, 1990; Kazdin & Mazurick, 1994; Kazdin, Mazurick, & Bass, 1993). However, the influence of parental stress on quality of participation has not been reported in the literature. In keeping with findings on stress and participation rate, it may be that parents who report high levels of stress and who attend intervention sessions participate minimally in those sessions. If this is the case, high levels of stress preclude active interpersonal engagement as well as involvement in group activities. Alternatively, parents who have high levels of stress may be motivated to engage in parent-training as a way of minimizing their level of stress. Thus, the relation between parental stress and quality of participation needs study.

Parents' perception of child problem behaviors and their sense of parenting competence also were considered as influences on participation. Parents of children with more severe behavior problems are more likely to terminate clinic-based services prematurely (Kazdin et al., 1993; Kazdin, 1990). Interrelations among parental stress, parent sense of competence, and parent perception of child problem behaviors are suggested in the literature. For example, parental stress has been found to be correlated with decreased parent role satisfaction (Koeske & Koeske, 1990; Mouton & Tuma, 1988). In addition, parents who report high levels of stress also report higher rates of child problem behaviors (Mouton & Tuma, 1988). Further, parents' perception of child problem behaviors have been found to be inversely correlated with parent role satisfaction (Johnston & Mash, 1989) and parent self-efficacy (Teti & Gelfand, 1991), two components of parents' sense of competence.

Finally, parents' own demographic profile may be associated with participation. Kazdin (Kazdin, 1990; Kazdin et al., 1993) found that premature termination from clinic-based services for conduct problems was associated with being a member of a minority group, a young mother, a single parent, and greater socioeconomic disadvantage. Although these parental characteristics are all associated with premature termination, processes leading to participation outcomes may differ. African American parents might participate less frequently either due to a longer history of societal stressors which could undermine their ability to benefit from the intervention (Dumas & Wahler, 1983; Hopps, Pinderhughes & Shankar, 1995), or a history of negative experiences with social service agencies, leading them to avoid or minimize their contact with an intervention (Hopps, Pinderhughes, & Shankar, 1995). Younger mothers might not have developed a level of individual competence prior to facing parenthood (Colon, 1980; Hines, 1988); thus they may lack sufficient individual competencies that enable one to continue in an intervention that targets improvements in parenting skills. Single parents may be less likely to participate because they face additional stressors and lack support from intimate partners. Low-SES parents must contend with greater income related stressors, making participation in an intervention unlikely. Thus, it is hypothesized that parents' race, age, marital status and SES will directly predict participation in the Fast Track intervention.

Hypothesized Model

Figure 1 presents the model of hypothesized relations among parent characteristics, family coordinator characteristics, family coordinators' therapeutic engagement, and parents' participation that guided this study. The similarity between the family coordinator's and parent's racial and socioeconomic characteristics and the extent of the family coordinator's relevant life experiences were hypothesized to influence the family coordinator's level of therapeutic engagement that, in turn, was hypothesized to influence rate and quality of parent participation. Furthermore, the family coordinator's therapeutic engagement as a proximal influence was hypothesized to mediate the relations between the distal influences, parent-family coordinator demographic and experiential similarity, and rate and quality of parent participation. Parent demographic characteristics were expected to influence participation directly whereas the influence of parents' level of stress was expected to be mediated by the proximal influences of their perception of child problem behaviors and their sense of parenting competence. Specifically, it was hypothesized that parents' race, age, and whether a male partner lived in the home would predict differential rates and quality of participation. African American, young, and single parents were expected to attend less frequently and to engage in sessions less actively. In addition, high levels of parent stress were expected to be associated with higher levels of child problem behaviors and a lower sense of parent competence. Child problem behaviors were expected to be associated with a lower rate of participation and less active involvement in sessions.

Fig. 1.

Distal influences, Proximal influences.

It was also hypothesized that higher levels of parents' sense of competence would be associated with higher rate and quality of participation. Parents with a low sense of parenting competence likely would be unable to mobilize themselves to attend training sessions or to actively participate in sessions. On the other hand, services provided in parent–group training sessions are designed to meet the needs of parents with a low sense of parenting competence, thus possibly facilitating their participation.

Method

Participants

All participants in this study were part of the ongoing Fast Track program, a multisystem, longitudinal preventive intervention for children who are at risk for conduct problems (CPPRG, 1992; Lochman & CPPRG, 1995). The project began when the children were in kindergarten; the oldest are now in middle school. A three-stage screening process (see Lochman & CPPRG, 1995) was used to identify these children as at risk: (a) schools serving communities with a high incidence of conduct problems were identified and randomly assigned to intervention or control conditions; (b) in these schools, kindergarten teachers' rated the children's disruptive behavior and poor peer relations using a revised version of the Teacher Observation of Classroom Adaptation-Revised (Werthamer-Larsson, Kellam, & Wheeler, 1991); and (c) parents were interviewed about their children's noncompliance and aggression at home. Children whose combined ratings from the teacher and parent were in the top 10% of problem behaviors were selected. Parents of children attending intervention schools were then recruited to participate in the intervention. To conservatively assess the future impact of the intervention, all parents and children recruited were considered part of the project, regardless of their subsequent decisions to participate in the intervention. Three successive cohorts were recruited, totaling 160, 160, and 135 children and their parents, respectively. Participants receive intervention services in four geographically distinct regions of the United States (approximately 40 children per site): a rural eastern area, a large city in the mid-south, a moderate-sized southeastern city, and a large metropolitan city in the northwest.

Participants in this study included only the primary caregivers (herein referred to as parents) of children from three sites in the second cohort and their family coordinators.⁴ Family coordinators are primarily involved in the parent-focused components of the intervention. Parents who were in the caseload of participating family coordinators were included in the study. Family coordinators were assigned to families based on family coordinator familiarity with school and/or community and randomness.

This recruitment strategy yielded a parent sample of 87 (72%) of a possible 120. These parents were all female; 86 were mothers and one was a grandmother.⁵ At the time of this phase of the intervention (i.e., when the target children were entering first grade, the parents ranged in age from 23 to 67, with mean and median age of 31 and 32, respectively.⁶ Forty four of the 87 families (51%) had adult male partners in the home, 57% of whom were biological fathers of the target child.⁷ Table 1 contains other demographic characteristics.

Table I. Participant Characteristics.

	%	N
Parents (N = 87)
Race
African American	55%	48
European American	45%	39
SES
Low to lower middle	66%	57
Middle	19%	16
Middle to upper-middle	15%	13
Upper	<1%	1
Education
Elementary school	16%	14
Attended high school	23%	20
Completed high school	35%	30
Attended college	23%	20
Completed college	3%	3
Marital status
Never married	44%	38
Married	27%	24
Separated or divorced	27%	24
Widowed		1
Family coordinators (N = 12)
Race
African American	42%	5
European American	58%	7
Childhood SES
Low	8%	1
Lower middle to middle	84%	10
Upper	8%	1
Education
Completed college	33%	4
Completed graduate traininga	67%	8
Marital status
Never married	17%	2
Married	75%	9
Separated or divorced	8%	1
Parental status
Parent	75%	9
Non parent	25%	3

^aTraining was completed in fields related to social services.

Family coordinators who completed measures on their level of engagement in the therapeutic relationship with parents were included as participants. Of 18 family coordinators who delivered the intervention to cohort two families at the three sites, 12 (67%) participated. Of the six nonparticipating family coordinators, three had left the project prior to data collection for this study. The remaining nonparticipating family coordinators did not meet study deadlines. There were no site differences as two family coordinators from each site did not participate. No differences in participation were found due to race, marital status, and parenthood status. All participating family coordinators were female, and their ages ranged from 24 to 40, with a mean age of 31. Table I contains other demographic characteristics.

Procedures

Parents

Parents were interviewed at home during the summer of 1992, after the child's kindergarten year and before the start of first grade (and the beginning of the intervention). To control for literacy levels, questionnaires were administered orally in face-to-face interviews with the parents. Data collected from parents included demographic information, type and severity of stressful life events they experienced over the course of the child's life, ratings of the behavior problems exhibited by the child, and ratings of feelings of satisfaction and efficacy in the parenting role.

The parent–group component of the intervention, one of several components involving the participation of parents, was the primary parent-training forum. Parents were encouraged to attend parent groups held at their child's school. In the first year of intervention, weekly parent–groups were offered and in the second year, twice monthly groups were offered during the 9-month school year. To promote attendance, transportation and child care were provided and each parent was paid $15 per group session attended. Each group was composed of no more than eight parents.

The stated objectives of the Fast Track program's parent group were “to foster the development of: (a) a positive family-school relationship; (b) parental self-control; (c) reasonable and developmentally appropriate expectations for the child's behavior; and (d) parenting skills to increase positive parent–child interaction and to decrease the occurrence of acting-out behaviors” (McMahon, Slough, & the CPPRG, 1996, p.8).

Family Coordinators

One aspect of the role of the family coordinator was to lead these parent groups. Family coordinators recorded parents' attendance and rated the quality of parents' participation in the group. At the end of the first year of the intervention, comprising 22 weeks of parent group sessions, family coordinators completed questionnaires to assess the level of their engagement in the therapeutic relationship with each parent to whom they were assigned.

Measurement

Parent Variables

The parent age variable was measured as age in years at the time of the interview. Race information was coded as a binary variable, with one indicating an African-American mother and zero indicating any other racial group (although almost all others were European Americans). The presence of a male partner in the home was also coded as a binary variable, with one indicating a male partner in the home and zero otherwise.

Type and severity of stressful life events experienced by parents over the course of their child's life were assessed by asking the parent to indicate whether 16 common stressful events (e.g., death of an important person, divorce, legal problems) had occurred during the current year or any time in the past; parents then rated the event on a 3-point scale of level of severity. Scores ranged from zero (stressor did not occur) to two (stressors that had major impact on the parent). The 16 stressful event ratings were summed into two separate scores: one for the current year and another for all previous years. The current and previous scores were summed into a single stress score (possible range = 0 to 64) that ranged from 1 to 33, with a mean of 11.3 (SD = 6.8) and median of 10. The stress score was positively skewed, but was not improved by square-root or logarithmic transformation. The raw scores were used in all analyses.

Parent perceptions of child behavior problems were assessed using the Child Behavior Checklist [CBCL] (Achenbach, 1991), and three additional items that were created for the Fast Track program: On these three items, parents globally rated their reactions to their children's behavior. The CBCL is a comprehensive list of child behavior problems that uses a 3-point scale for the frequency at which children exhibit each behavior. Normed CBCL Externalizing (M = 61.10, SD = 9.25) and Internalizing (M= 54.8, SD = 10.49) behavior subscale scores (T-scores) were used, with higher scores indicating a greater number of problem behaviors. The three parent perception ratings of children were reported on a 1 to 5 scale, with higher scores indicating more positive parent reactions to her child (“In general, how satisfied are you with your child's behavior?,” “How pleasant has it been to raise your child?,” “How difficult is it to be patient with your child?”—the last item was reverse-scored). The three ratings were summed into a single score of parent perception of one's child (alpha = .50). This score had good distributional characteristics, with a mean of 10.9 and an SD of 1.4.

The Parenting Sense of Competence Scale (Gibaud-Wallston & Wandersman, 1978; cited in Johnston & Mash, 1989), was adapted for use in face-to-face interviews in this study. Twelve of the original 17 items were used: 6 about their feelings of parenting efficacy and 6 about their satisfaction with their parenting role. Seven of these 12 items were rewritten to be more readable for this parent population. Parents responded to each item on a l-to-7 scale. The mean of the 6 parenting efficacy items (α = .72) was taken as a parenting efficacy subscale score. The score ranged from 1 to 7, with a mean of 5.5 and an SD of 0.8; higher scores indicated greater feelings of parenting efficacy. The mean of the 6 parent role satisfaction items was taken as a parent role satisfaction subscale score (α = .77). The score ranged from 1 to 7, with a mean of 3.8 and an SD of 1.3.

Family Coordinator Variables

The family coordinator variables, Race-Match and SES-Difference, were designed to indicate the degree of similarity between the family coordinators' and parents' personal characteristics and life experiences. One binary variable was created to indicate matching race of the family coordinators and parents; each assumed a value of one given a positive match. Socioeconomic status of the parents was computed first by applying the 4-factor method suggested by Hollingshead (1979). During the postkindergarten summer interview, the mother provided information on her occupation and years of education as well as information on the occupation and education of the male head of household. If there was no male head of household, the mother's information was double-weighted (as recommended by Hollingshead). From this information, one categorical group score was generated with ranges from 1 to 5; with 1 indicating the highest SES class and 5 indicating the lowest SES class. The parent's current categorical SES score then was subtracted from a similar 5-point categorical self-rating of the family coordinator's childhood SES to create the SES-Difference variable. Low scores reflected greater socioeconomic similarity. Family coordinators' childhood SES, and not their current SES, was used as the match variable because family coordinators' childhood SES was conceptualized to be an index of the experiences of their formative years, experiences that likely contributed to their cultural sensitivity. Family coordinators' current SES could not logically be used as a match variable, given that they were employed by Fast Track largely because of their educational background. Thus, their employment provided a lower bound on occupation.

Family coordinators were asked to describe personal experiences that they thought facilitated their ability to relate to the needs of the parents served by the Fast Track program. They were given examples of such personal experiences including financial hardship, experiencing oppression, exposure to different cultural backgrounds, and single parenthood. In open-ended responses, family coordinators identified these and other important experiences: living in a low SES community, exposure to substance abuse, exposure to severe behavior problems, and family dysfunction. In total, eight relevant life experiences were culled from the responses.

Two sets of eight binary variables then were coded from these open-ended responses: one set indicating whether the family coordinator mentioned having each experience in childhood and another set indicating each experience in adulthood. In recognition that childhood experiences are very potent formative influences on an individual's development, they were weighted by two and added to adult experiences. Of the eight life experiences, six were theorized to be exceptionally important as influences on a family coordinator's sensitivity to parents served by the Fast Track program. Four of these experiences, individual financial hardship, single parenthood, severe behavior problems, and family dysfunction were viewed as very similar to the characteristics of families in the target population. Two others, exposure to different cultures and experiences of oppression were theorized to sensitize family coordinators to parents who may be of different cultural background. Thus, these six scores were weighted by two. Substance abuse and living in a low SES community scores were unit-weighted. The eight scores were summed into a Relevant Life Experiences indicator that had a possible range of 0 to 42 (actual range 0 to 21, M = 11.57, SD = 6.03). Higher scores indicated that the family coordinator had a greater number of relevant life experiences that were similar to the overall parent population.

The second score, Family Coordinators' Level of Therapeutic Engagement, was designed to measure the extent to which family coordinators remained engaged in the therapeutic relationship with parents and pursued the Fast Track program's intervention goals. Fourteen items measuring three components of family coordinators' therapeutic engagement were developed for this study. The Therapeutic Engagement score was derived by summing component subscale scores: (1) Family Coordinator Beliefs about Parent Response was composed of three items (α = .91) that included the family coordinator's ratings of how much the parent liked and trusted her (e.g., “do you think this parent respects what you have to offer?”); (2) Family Coordinator Ability to Remain Empathic and Supportive was composed of six items (α = .85) designed to measure the family coordinator's ability to remain empathic and supportive toward each parent during the period of greatest resistance (e.g., “how difficult is it for you to maintain a friendly and receptive attitude toward this parent?”); and (3) Family Coordinator Ability to Remain Effective in Teaching and Confronting Parents included five items (α = .86) measuring the family coordinator's ability to remain effective in the delivery of the intervention services to the parent (e.g., “how effective do you think you have been in providing emotional support to this parent?”). All elemental items comprised by these subscales had a 1 to 4 scale. Subscale scores were calculated by taking the mean of the items in a subscale, so that the three subscale scores had the same one to four possible range. These three intercorrelated (.66 to .80) subscale scores were summed to produce a composite Therapeutic Engagement score (possible range = 4 to 16, actual range = 4.2 to 12, M = 9.48, SD = 1.77). Higher scores indicated a higher level of therapeutic engagement.

Outcome Variables

The outcome variable for parent-group participation rate was each parent's attendance at the twice monthly group sessions held during the second intervention year. The total number of sessions offered to parents varied somewhat, with an average of 16.5 (SD = 2.7; mode = 16-18). To account for the different number of sessions offered, a participation rate proportion score was calculated for each caregiver by dividing the number of sessions attended by the number of sessions offered.

Quality of participation was assessed by a measure comprising six items that were rated by the family coordinator on the level of interest and involvement shown by the parent during parent–group sessions in the second intervention year. All six items had a 1 to 4 scale, therefore the sum score had a possible range of 6 to 24, with higher scores indicating greater interest and involvement. (Sample ratings include: Degree of participation in the parent group: 1 = parent is silent throughout most sessions; 4 = parent participates enthusiastically and in a positive way during most sessions in terms of both discussion and role play.) Cronbach's alpha for the scale was .89. The quality-of-participation score used in the analyses was the mean of two highly correlated (.63) scores: one from ratings of the first 11 weeks of parent groups and another from the second 11 weeks of parent groups (M = 18.11 SD = 4.54).

Results

Parents attended, on average, 56.3 % (SD = 33%) of the parent group sessions offered. Table II contains the percentages of the sample who received different quartiles of dosage. Nine parents attended no group sessions, and eight parents attended all sessions that were offered to them.

Table II. Parent Participation in Sessions Offered to Them by Quartiles of Dosage.

Group	% of Sample
25% of sessions or less	26.4
26–50% of sessions	15
51–75% of sessions	21.8
Over 75% of sessions	36.8

Table III shows the means and standard deviations of all variables included in this analysis, and their intercorrelations. A series of regression models were specified to test the mediation hypothesis according to the guidelines of Baron & Kenny (1986). These regression models were carried out in the following order: (a) the proximal influence-outcome relation; (b) distal influence-proximal influence relation; (c) distal influence-outcome relation; (d) distal influence-outcome relation, controlling for proximal influence. These results will be presented first for therapeutic engagement, and then for parent characteristics.

Table III. Intercorrelations Among All Variables.

	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16
Mean	0.56	16.6	0.7	0.2	11.6	9.5	11.3	5.5	3.8	10.2	61.1	54.8	31.9		3.9
SD	0.33	6.4	0.4	1.4	6.1	1.8	6.8	.8	1.3	1.7	9.3	1 .5	7.0		1.1
1. Attendance yr 2	—
2. Parent-gp. qual. y 2	.62****	—
3. Race-match	−.01	.18	—
4. SES-difference	−.07	−.15	−.16	—
5. Life experiences	.06	−.05	.02	−.36***	—
6. Therapeutic engagement	.39***	.43****	.20+	−.27*	.21*	—
7. Parent stress	−.12	.03	−.17	−.06	−.10	−.05	—
8. Parent efficacy	.12	−.15	−.19	.03	.21*	.03	−.02	—
9. Parent satisfaction	.01	.13	.01	−.26*	.09	−.04	−.03	−.02	—
10. Child prob. beh.	−.06	.06	.08	−.18	.04	.10	−.19	.10	−.41***	—
11. CBCL externalizing	.18	.19	.11	.10	.07	.13	.19	−.04	.22*	−.38***	—
12. CBCL internalizing	.07	.19	.11	.15	.02	.05	.10	−.06	.09	.28**	.61****	—
13. Parent age (yrs.)	.05	.18	.19	−.25*	−.11	.24*	.03	−.16	−.25*	.19	−.10	.12	—
14. Afr American (0–1)	−.18	−.35**	−.28**	−.03	.36***	−.15	−.11	.20	.01	.07	−.24*	−.23*	−.17	—
15. Parent SES	−.01	−.14	−.14	.78****	.11	−.14	−.11	.18	−.31**	−.13	.20	.28	−.35***	.16	–
16. Male Household (0-1)	.13	.20	.03	−.03	−.28**	.10	−.07	−.26*	−.12	−.06	−.13	−.04	.24*	−.37***	.21*	—

Therapeutic Engagement and Participation

As hypothesized, Therapeutic Engagement was significantly and positively related to the rate of parent participation (r = .39, p < .001): As the degree of engagement invested by family coordinators increased, parent participation in the following year increased. Therapeutic Engagement also was significantly and positively related to quality of participation in the second year of the intervention (r = .43, p < .0001). When the family coordinator's engagement with the parent was high, parents showed more interest and actively participated in sessions. Because outcome variables were assessed in the second year of the intervention, rate and quality of parent participation may have reflected a stable pattern from the first year of intervention. Thus, two follow-up regression models—one each for rate and quality—which controlled for Year 1 participation were run. Results of the regression predicting year two participation rate and controlling for Year 1 rate revealed that Therapeutic Engagement remained marginally significant, with a SD b = .182, p < .10. The regression predicting Year 2 participation quality after controlling for Year 1 quality indicated that Therapeutic Engagement no longer was a significant predictor (SD b = .018).

Family Coordinator–Parent Similarity and Therapeutic Engagement

Bivariate analyses indicated significant relations between SES-Difference and Therapeutic Engagement (r = −.27, p < .05), and between Relevant Life Experiences and Therapeutic Engagement (r = .21, p < .05). Race-Match correlated with Therapeutic Engagement at p < .06, r=.20. The set of determinants were regressed on Therapeutic Engagement, yielding a significant relation, R² = .12, F(3, 82) = 3.79, p < .01. Race-Match and SES-Difference had SD b coefficients = .18, t = 1.73 (p < .10), and .20 t = 1.72 (p < .10), respectively, showing that racial and socioeconomic similarity tended to be associated with more positive therapeutic engagement.

Family Coordinator-Parent Similarity and Parent Participation

As Table III shows, Race-Match, SES-Difference and Relevant Life Events were not significantly correlated with participation rate or quality. The regression model testing relations between this set of distal predictors and participation was not significant. Thus, the role of therapeutic engagement as a mediator of these relations could not be tested.

Parent Characteristics and Parent Participation

Parent Perceived Competence and Child Functioning

Correlations of individual measures of parent competence and parent perception of the child and parent report of child functioning with participation rate or quality were not significant. The regression models testing relations between this set of five proximal predictors and participation rate and quality were not significant. Bivariate analyses revealed that parent stress was unrelated to parent satisfaction, parent efficacy or parent perceptions of child and parent report of child externalizing behaviors. Bivariate correlations between parent stress and participation rate and quality were nonsignificant. Thus, none of the links of the hypothesized model of parent stress-parent perception of child-parent participation were significant.

Of the four parent demographic characteristics hypothesized to relate to participation, only caregiver race significantly correlated with participation quality (r = −.35). No parent demographics significantly correlated with participation rate.⁸ The regression model testing relations between this set of distal predictors and participation quality was significant (R₂ = .15, F(4, 70) = 3.16, p < .05). Caregiver race uniquely predicted quality of participation, (std. b = −.27, p < .01): As predicted, African American parents participated less actively in sessions than did European American parents.

Test of Full Model

Finally, a last regression was run to test the theoretical model presented in Fig. 1 positing that the family coordinator's level of engagement with the parent and parent demographics each would independently predict parent participation when in the context of other possible influences. A modified test was conducted, using only the significant variables from prior analyses of outcome variables. This modified model, rather than the full model, was used to reduce capitalization on chance, given the sample size. Therapeutic Engagement and parent race were entered simultaneously in a regression model of quality, R² = .26, F(2,75) = 12.96, p < .001. Therapeutic Engagement was significantly related to quality in this model, std. b = .37, t = 3.67, p < .001. Parent race also was significantly related to quality, std. b = −.27, t = −2.68, p < .001.

Exploratory Analyses of Determinants of Therapeutic Engagement

Given the bivariate relations between Race-Match, SES-Difference and Relevant Life Events and Therapeutic Engagement, and the findings that when all three predictors were regressed on Therapeutic Engagement, Race-Match and SES-Difference were significant only at p < .10, further exploratory analyses were run to clarify possible influences on therapeutic engagement. Regression analyses revealed that Relevant Life Events (SD b = .20) and Race-Match (std. b = .20) each predicted therapeutic engagement at p < .06 (R² = .08, p < .05). Only SES-Difference was a significant predictor (SD b = .28, p < .05) when combined with Relevant Life Events (SD b = .13, p = .25), R² = 09, p < .05. The combination of Race-Match and SES-Difference yielded std. bs of .17 (p < .10) and −.25 (p < .05), respectively (R² = .10, p < .01). Thus, it appears that socioeconomic similarity offered unique contributions to therapeutic engagement when paired with either Relevant Life Events or Race-Match.

Exploration of race differences in the relation between Race-Match, SES-Difference and Relevant Life Events and Therapeutic Engagement yielded interesting results. Table IV contains the mean Therapeutic Engagement scores for the race and Race-Match interaction. As can be seen, mean Therapeutic Engagement scores were lowest for African American parents who did not share the same race as the family coordinator; all other groups had comparable scores. An ANOVA of these means did not reveal significant differences, probably due to insufficient power to test the interaction. However, these findings suggest that similarity of racial background may differentially matter to therapeutic engagement as a function of race. No race differences were found in relations between Relevant Life Events and Therapeutic Engagement or between SES-Difference and Therapeutic Engagement. Exploratory analysis of Relevant Life Events and Therapeutic Engagement for low SES caregivers revealed a modest correlation (r = .31), indicating higher levels of engagement for those family coordinators who had more experiences that were similar to the sample. However, this correlation was not significantly different from the correlation between Relevant Life Events and Therapeutic Engagement among middle-SES caregivers (r = −.03).

Table IV. Therapeutic Engagement Mean scores by Race-Match and Race of Caregiver.

	Race Match with FC
	Yes	No
African American	9.61 (SD = 1.93) n = 30	8.63 (SD = 1.47) n = 18
European American	9.77 (SD = 1.64) n = 34	9.80 (SD = 2.23) n = 5

In summary, Therapeutic Engagement was found to be highly and positively associated with parent group participation rate and quality. Greater similarity between the parents' and family coordinators' race, SES, and life experiences was associated with higher levels of Therapeutic Engagement. No parent characteristics were related to participation rate, and race was the only parent characteristic found to be related to quality. Exploratory analyses of determinants of Therapeutic Engagement suggested two interesting possibilities. First, socioeconomic similarity uniquely predicted Therapeutic Engagement when paired with either Race-Match or Relevant Life Events. Second, although not a significant finding, European American family coordinator—African American parent matches appeared to yield lower Therapeutic Engagement scores than did other matches.

Discussion

The model of hypothesized relations among parent characteristics, family coordinator characteristics, the family coordinator's therapeutic engagement with the parent, and parent participation (see Fig. 1) was supported only in part. Foremost among the findings was that the degree to which family coordinators remained engaged in the therapeutic relationship emerged to have important associations with rate and quality of parent participation in this type of intervention.

A bivariate relation between therapeutic engagement and rate of participation revealed that when the family coordinator's therapeutic engagement with the parent was high, the parent attended parent group sessions more frequently in the following year. After controlling for parent participation in the first year of intervention, this relation remained marginally significant, indicating that when the stability of parent attendance was considered, therapeutic engagement tended to remain a contributor to the rate of parent participation.

The relation between therapeutic engagement and quality of participation, however, seems to be more complex. The bivariate relation between engagement and quality of participation was significant, suggesting that when therapist engagement was high, the parent also exhibited more interest, initiative, and involvement in the group activities. However, when participation during the first year of intervention was accounted for, this relation lost significance. One possible explanation for this finding lies in rater source bias: Family coordinators rated both quality of participation and therapeutic engagement. Although these ratings were obtained at different time points, a common view of the parent held by the family coordinator may explain this association. However, another possible explanation is that the quality of parents' participation is less affected by the family coordinator's engagement than by other influences not measured in this study. For example, composition of parent groups—demographic and interpersonal—may have affected how actively involved and interested parents were during sessions (e.g., Shen, Sanchez, & Huang, 1984). As Yalom (1995) has observed in therapy groups, different interpersonal styles of interaction among parents may have influenced group functioning, and thus the quality of participation. Another possible influence lies in family coordinators' skills in group management and facilitation: The degree to which family coordinators were able to motivate and maintain parent interest in the curriculum being implemented may have influenced quality of participation. Because skills to promote dyadic therapeutic engagement and effective group functioning may differ (e.g., Webster-Stratton & Herbert, 1993), the level of engagement between family coordinator and parent may have had little unique impact on quality of participation.

These findings extend the literature on the role of the therapist's contribution to the working alliance in populations served by interventions like Fast Track. Although the literature on the working alliance in individual therapy suggests that the therapist's role in the working alliance is related to client-initiated termination (e.g., Hartley & Strupp, 1983; Kokotovic & Tracey, 1990; Tryon & Kane, 1993), there has been little research regarding the therapist's contribution to the therapeutic relationship in community-based interventions until now. These findings also are informative regarding mixed model interventions in which parents of children at risk for conduct problems receive both home visits and group-based training (Prinz & Miller, 1994). Moreover, these findings suggest the importance of future research on group composition and dynamics as well as “therapist” group facilitation skills as influences on parent participation in group-based services.

In this study, the family coordinator's therapeutic engagement was operationally defined in terms of the family coordinator's analysis of her own ability to remain engaged in the therapeutic role and to maintain the delivery of services to the parent, despite the high degree of parent resistance (e.g., Chamberlain et al., 1984). This approach appears to represent a valid assessment of the family coordinator's relationship with the parent, particularly given the observation of Patterson and colleagues (Patterson & Chamberlain, 1988,1994; Patterson & Forgatch, 1985) that parent treatment response in this population is highly dependent on the family coordinator's ability to allay parent resistance and to remain engaged with the parent. Moreover, family coordinators assessed the therapeutic engagement at the end of the first year of the intervention, after 22 parent–group sessions, the period of the intervention during which parent resistance would be expected to be at its height. Patterson and Chamberlain note that the family coordinator's level of engagement with the parent is most critical during the time frame when the teaching of new parenting skills is underway. It is during this period that resistance is posited to increase and to peak. Thus, the definition and timing of assessment of the family coordinator's level of engagement appear to have yielded a sensitive measure of the effort family coordinators must extend in the face of parent resistance. However, an assessment of the temporal stability of this measure of therapeutic engagement, which was not included in this study, would provide more information regarding the reliability and validity of this measure. In fact, other measures of therapeutic engagement may be even more helpful in establishing the validity of this measure. For example, parent rating of therapeutic engagement not only would serve as a validity check on the family coordinator's report of therapeutic engagement, but also would provide important information about the relationship between family coordinator and parent (Horvath, 1994). Moreover, an independent rating of family coordinator engagement with parents during the group sessions would yield highly relevant information about contextual stability in therapeutic engagement and its relation to participation. Thus, future research in this area should include the view not only of the family coordinator, but also of parents and independent observers in assessment of the therapeutic relationship. It is also important to learn from parents the extent to which they like the program and find its content relevant to their needs. In addition, use of multiple time points in subsequent research also would aid in clarifying any changes that may occur in therapeutic engagement over time and the way such change relates to rate and quality of participation.

What contributed to the family coordinators' level of therapeutic engagement with parents? Given that a necessary condition of a strong therapeutic engagement is the family coordinator's empathic sensitivity to the parent, it was hypothesized that the therapeutic engagement would be more easily facilitated when family coordinators were of the same race as the parent, and when they, as children, had experienced socioeconomic circumstances similar to those of the parent. It was also hypothesized that the therapeutic engagement would be further reinforced if family coordinators were sensitive to the experiences of this population. Cultural sensitivity was operationalized as the direct experience, either as children or as adults, of life events similar to those experienced by the parent, such as oppression, divorce, family conflict, and single parenthood.

Bivariate correlations indicated that when the family coordinators were similar to parents in their socioeconomic background and life experiences, they were better able to remain engaged in the therapeutic relationship. However, when all three variables assessing family coordinator similarity to parents were considered jointly, as in the regression, racial similarity and socioeconomic similarity emerged only as marginally significant. Follow-up exploratory analyses which examined various combinations of influences on therapeutic engagement suggested that socioeconomic similarity may be a unique contributor when combined with either racial similarity or similarity in life experiences. Exploratory analyses of race differences in the relation between racial similarity and therapeutic engagement suggested that European American family coordinator - African American parent matches may have posed the greatest challenge to achieving therapeutic engagement. Although this finding was not statistically significant (probably due to insufficient power to test this interaction), this possibility commands attention. There is evidence that the establishment of an effective therapeutic relationship depends on the therapist's ability to enter into and to understand his or her client's worldview. It appears to be equally important that the client perceives that there is therapist-client congruence in worldviews (e.g., Trevino, 1996; Worthington & Atkinson, 1996; Duncan & Moynihan, 1994). Should this finding be replicated with larger samples, perhaps therapeutic engagement in European American family coordinator—African American parent matches could be improved by ensuring they share the same socioeconomic background.

The modest percentage of variance associated with therapeutic engagement that was explained by racial and socioeconomic similarity and relevant life experiences (R² = .12) reveals that a family coordinator's level of engagement with a parent is explained only partially by these three influences. The family coordinator's racial and socioeconomic similarity to the parent may serve as a kind of “foot-in-the door,” perhaps facilitating rapport earlier in the relationship. Influences that also might affect rapport in the relationship include other areas of similarity such as age and partnership status. Indeed, the significant positive correlation between parent age and therapeutic engagement suggests that family coordinators had higher levels of engagement with older parents. Although measured in this study, parent age and partnership status were not included as predictors due to the limitations of the sample size. Other influences, not measured in the current study, likely contribute to the therapeutic engagement. In fact, one important influence is likely to be the family coordinator's therapeutic competence. Patterson and Chamberlain (1988) suggest that therapeutic competence becomes more influential than demographic similarity in the maintenance of a strong and positive therapeutic relationship over a long period. One could envision that a family coordinator's competence could transcend a parent's initial misgivings that arise because of demographic dissimilarity. Conversely, demographic similarity likely would not compensate for poor therapeutic skills. The relation of therapist competence to rate and quality of parent participation in this type of intervention is a promising area for further study.

Whereas the therapeutic engagement was a strong predictor of both participation rate and quality, only one parent characteristic was found to be significantly associated only with quality of participation. Consistent with past studies, African American parents participated less actively in sessions than did European American parents. There are several possible explanations for why African American parents did not engage as actively as did European American parents. Indeed, just as the development of conduct problems is multiply determined, so too may be the quality of participation of African American parents in this intervention. For African Americans, perhaps a history of longer, more pernicious social disadvantage posed an additional barrier to their ability to develop the social skills necessary for active participation (Dumas & Wahler, 1983; Hopps, Pinderhughes, & Shankar, 1995). It also may be that Fast Track's parent-training curriculum, which emphasizes praise, ignoring, and time out (CPPRG, 1992), may have resonated more with European American parents' values than with African American parents' values. However, the effectiveness of this type of parenting training was recently shown with a sample of African-American parents (Kumpfer et. al., 1996). Moreover, recent analyses of the effect of the Fast Track intervention indicate that there have been no signs of differential effectiveness due to race of participants (CPPRG, 1996). Thus, this issue awaits further research. Another possible explanation lies in a history of prior negative contact with social service systems. As a result of prior disempowering experiences (e.g., Hopps, Pinderhughes, & Shankar, 1995), African American parents may have viewed the intervention with a healthy skepticism as they assessed the legitimacy of the family coordinator and the project. The plausibility of these different explanations suggests that participation of African-American parents in an intervention serving families-at risk is complex and warrants further study. It may be, however, that the project's emphasis on empowerment through a strong therapeutic engagement may enable these parents to develop into more active participants in subsequent years of the intervention.

It is surprising that parent age, socioeconomic status and partnership status did not predict quality of participation. This may be due to the fact that quality of participation was assessed 2 years after the intervention began, perhaps giving time for more proximal influences such as therapeutic engagement and group dynamics to operate and minimize the effect of these parent characteristics. Indeed, Patterson and Chamberlain (1988) note that parent characteristics likely exert their influence earlier in the intervention. These findings also may be due to restricted ranges of the variables studied. Because families of children already showing relatively high levels of conduct problems were recruited to participate in the study, the associations between parent predictors and participation rate and quality may have been underestimated. The restricted range of parenting and family experiences likely represented in this high-risk sample may have attenuated relations between parent predictors and participation rate and quality. Stronger relations might be found in a community sample representing a broader range of family and parenting experiences.

The finding indicating no demographic differences in attendance also is surprising, given previous research on participation rate in clinic-based services (e.g., Kazdin, 1990; Kazdin, Mazurick, & Bass, 1993; Kazdin & Mazurick, 1994). At first glance, it seems that family coordinators in this school and community based intervention were effective in mobilizing all parents to attend parent groups, again indicating the significance of their therapeutic engagement. However, the location of service delivery in neighborhood schools may have facilitated the rate of participation among African-American, single parents, younger mothers or low-income mothers who would be expected to be less likely to continue in clinic-based services. In addition, other ancillary services such as transportation, child care and payment for attendance may have promoted participation rate among these groups.

Other external constraints may have affected participation rate, as well. Some parents were regular nonattendees and could not be included in the analyses of quality of participation. A missing data analysis revealed that extremely low parent-group attendance was related to the number of hours worked per week, suggesting that work obligation is an important external influence on participation rate. Thus, influences internal and external to the program likely affected participation rate. Future research should examine the role of location and ancillary services in facilitating participation in different demographic groups.

Implications of these findings must be considered in the context of the study's limitations. In addition to the limitation of rater source bias previously discussed, the sample of only 12 family coordinators restricted the type of analyses that could be conducted. Given that the unit of analysis was the family coordinator-parent dyad (and each family coordinator was matched to several families), independence of dyads cannot be assumed. In addition, as a result of the small sample size there was insufficient variation among family coordinator–parent matches to examine different Race-Match and SES-Difference configurations. Thus, the differential impact of race and socioeconomic matches on the therapeutic engagement between family coordinator and parent remains unknown. Third, the variable assessing family coordinators' relevant life experiences was matched to the experiences of typical parents in this sample, rather than to individual parents. Consequently this variable may have lacked sensitivity to the dyadic similarity that may have existed.

Finally, and perhaps most critically, very little is known about the impact of parent characteristics on participation. Although this study contributes to an understanding of the impact of parent demographics on both rate and quality of participation, it contributes nothing about the impact of psychosocial characteristics on participation. Although stress was not found to be related to participation in this study, there is ample evidence pointing to its role in limiting parent participation in various interventions (e.g., Kazdin, Mazurick, & Bass, 1993; Prinz & Miller, 1994). Other characteristics such as parents' expectations, attributions, and beliefs (Prinz & Miller, 1996) must also be examined. For example, parents' causal attributions about their children's behavior problems may differentially predict participation. More specifically, parents who acknowledge a lack of parenting skills in child behavior management may be more likely to attend parent skills training than parents who attribute child behavior problems to innate characteristics of the child.

Just as these findings clarify the need for greater empirical understanding of influences on parent participation, they also have implications for service delivery with families at risk for dysfunctional outcomes. Family coordinators must foster actively and monitor closely their therapeutic relationship with parents. Moreover, the findings on the relation between family coordinator-parent racial and socioeconomic similarity and family coordinators' level of therapeutic engagement suggest that when pairing service provider with parent, attention to demographic similarity is warranted. However, it is clear that similarity in demographic and experiential backgrounds, although important, is not the only critical consideration. Other considerations may include the service provider's therapeutic competence. Indeed, as future research examines the role of therapeutic competence in the level of therapeutic engagement as well as in group facilitation, the training of service providers may need to address strategies to work effectively with culturally diverse parents.