School-based family support: Evidence from an exploratory field study

Abstract

The use of family support interventions to address the needs of at-risk families and children has proliferated over the last 2 decades. In particular, family support has been used to strengthen involvement of parents in their child's academic life as a means of preventing academic failure and other problem behaviors. This article presents empirical data from a field-based study of one leading family support model, Families and Schools Together (FAST), based on over 400 families served over a 4-year period. Child participants showed statistically significant declines in behavior problems, and their families showed improved adaptability and cohesion. Observations about program operations and the differential results for subgroups of children are also presented, along with the study limitations.

Overview

FAMILY SUPPORT INTERVENTIONS have been promoted as a viable means for addressing a broad range of issues facing at-risk children and families. However, although the research literature on family support is large and varied, it includes few examples of field-based research on such interventions. The operation and evaluation of a family support program model, Families and Schools Together (FAST), which has been implemented widely in the United States, primarily through local family service agencies. In the present study, data were collected from 427 families who graduated from the FAST program in the metropolitan Atlanta, Georgia, area during a 4-year period (1996-2000). The data, collected from parents and teachers, show that observable and statistically significant differences occurred for many participants during their participation in FAST, particularly in the area of improved behavior among the FAST children.

The evaluation of the FAST program is based on measures involving child behavior, family adaptability and cohesion, and parental involvement and support. The results show that on the initial outcome measures the program has been effective in leading to positive change in the families who graduate. The detected effects are both statistically significant and programmatically meaningful in regard to positive changes in child behavior, family adaptability and cohesion, and parental support.

Program Background

The use of family support interventions to aid at-risk families and children has proliferated over the last 2 decades. In particular, family support has been used as a way of strengthening parental involvement in their child's academic life as a means of preventing academic failure and problem behaviors, such as drug use (Favorini & Pryor, 1994).

The program currently under study, FAST, is an 8-week multifamily group prevention program targeting families with children in early elementary grades (K-3) who are at risk for school problems, failure, and future drug abuse. Initially, FAST was developed specifically as a substance-abuse prevention program but was later found to be applicable to a broader range of at-risk families (McDonald, Bradish, Billingham, Dibble, & Rice, 1991). FAST was developed in 1987 by the staff at Family Service of Madison, Wisconsin, led by Lynn McDonald, Ph.D. Subsequently, FAST was replicated through the Alliance for Children and Families' (formerly Family Service America; www.alliance1.org) network of child- and family-serving agencies). The FAST program has been replicated widely and is estimated to be available in 600 schools in 38 states in the United States and in five other countries (Chaika, 2000). The FAST program has received various awards, including being selected as a model family skills training program by the U.S. Department of Justice (Kumpfer & Alvarado, 1997).

The FAST program involves families in a structured format that seeks to empower parents to address whatever emotional or learning difficulties their child(ren) may have. Each of the eight weekly FAST meetings includes six basic program elements: (a) a shared family meal, (b) family games involving communication tactics, (c) a group discussion for parents separate from a play session for the children, (d) play-focused time involving parent and target child ("special play"), (e) a family lottery in which each family wins once to serve as the host for the next FAST session, and (f) a closing ceremony with singing and recognition of family and individual accomplishments. In addition, there are specific one-time activities used within the program, such as the creation of a family flag in the first session, a presentation to families on substance use and abuse in the fifth session, and a graduation ceremony in the eighth session. Each element of the FAST curriculum was designed on the basis of relevant prior research findings of effectiveness from the body of published mental health literature (McDonald et al., 1997).

Target children are initially identified for the program by teachers and school counselors, and also by their own parents. School staff members are asked to make referrals to the program of children who are at risk of academic failure or who are exhibiting negative social behaviors, often involving children with behavior problems. Once a referral is made, the child's parent is contacted about the family participating in the 8-week program. After FAST has concluded, families are encouraged to participate in a ongoing program called FASTWORKS, which is a group that meets monthly over a 2-year period in order to reinforce the emphases of the FAST program.

Each FAST program team consists of a FAST facilitator, an alcohol and drug counselor, a parent liaison, a teacher liaison, as well as several program aides or volunteers. Each FAST team must be certified to deliver the program, and during the course of this study the Alliance for Children and Families certified all program teams. The standardization of the delivery of the FAST curriculum is handled through a tailored training and certification program. In order to field a certified FAST team, all team members must complete a 2-day orientation to the program, participate in three on-site visits by a certified FAST trainer during the first 8-week cycle, and complete a 1-day training after the completion of the cycle. Each team member receives a copy of the program manual that describes all aspects of the curriculum, program goals, and team member roles (Alliance for Children and Families, 2000). Periodic visits to subsequent sessions by a FAST trainer provide ongoing feedback to the team about their implementation of the program model.

The FAST program is closely identified with the broader movement to develop family support programs and practice (Family Resource Coalition of America, 1998). The literature on family support is diverse and abundant, and this is due, in part, to the diversity of opinions about how family support is defined. In a recent study Layzer, Goodson, Bernstein, and Price (2001) documented the difficulty of defining family support for the purposes of conducting a systematic review of the relevant literature. After extensive work, the authors elected to use a definition that involved "services intended to improve child outcomes by strengthening the capacity of parents to support their children's development" and a measure of the comprehensiveness of the program related to specific practice principles in the field (p. A2-2).

Given the variation in programs in the family support arena, it would be expected that the specific programmatic goals of individual programs would also vary substantially. The specific outcome goals identified for the FAST program are embedded into the program model itself. Theses goals include (a) preventing school failure by improving children's behavior and concurrently their academic performance, (b) enhancing family functioning through strengthening the parent-child relationship and working to empower parents, (c) reducing the everyday stress by developing an ongoing support group for parents and by building self-esteem, and (d) preventing substance abuse by increasing families' knowledge about substance abuse and providing a referral when appropriate (Penry, 1996).

The research base on the FAST model has expanded considerably over time, as the program itself has seen increased use (McDonald & Frey, 1999). In 1999, there were five separate federally funded experimental studies under way examining aspects of the FAST model (McDonald, 1999). In 2001, a major study of family support programs was completed involving empirical findings from in-depth studies of six family support programs including FAST (Layzer, Goodson, Creps, Werner, & Bernstein, 2001). The authors conducted a study of FAST involving 207 participant families and 200 comparison families at nine elementary schools in New Orleans, Louisiana. In respect to behavior problems, the authors reported that FAST children showed significant improvement in externalizing behaviors (parent report) at a 1-year follow-up, but teachers observed no differences in FAST children compared with the control group. Though the authors found few significant positive impacts on FAST families 1 year after participation, they noted that the program as implemented departed in key ways from the FAST model, and the populations served exhibited particularly high levels of high-risk families (Layzer, Goodson, Creps, et al., 2001).

In a separate study by Moberg, McDonald, Brown, and Burke (2003), the authors used a randomized design to compare the effects of FAST to a very limited family education intervention (i.e., FAME). The study, comparing 272 FAST families and 201 FAME families over 2 years of follow-up, showed very few differences between the group outcomes. The authors, however, did report a significant positive overall effect of FAST on the child's academic performance at 2 years based on teacher's report, as well differences for specific ethnic subgroups.

There are also examples in the literature of family support programs that share some goals with FAST, although still differing in specific approach in notable ways (Devaney & Milstein, 1998; Greif, 1993). Among the existing research on the model is a detailed investigation of the implementation of FAST in Milwaukee, Wisconsin, where the model was initially evaluated in 1987. These studies have shown statistically significant improvements among participant families as well as sustained effects over a 6-month follow-up period (McDonald & Sayger, 1998; Sayger, 1996). Other researchers have examined the application of the FAST model in the middle-school setting (Pinsoneault & Sass, 1999). The present study draws on FAST data compiled and reported annually in the metropolitan Atlanta area (see, e.g., Fischer, 2000).

Program Implementation

During the period of this study, professional social work staff delivered the FAST program at selected public elementary schools in the metropolitan Atlanta area. FAST was initially offered in the fall of 1994 in two elementary schools, and over 6 school years (1994-2000), a total of 552 families graduated from the FAST program offered in 13 schools in the metropolitan Atlanta area. The schools that were recruited to participate in FAST were identified by the provider agency as having large numbers of families who could potentially benefit from the intervention. Schools participated voluntarily and only with the consent of the principal and the governing school board. The participating elementary schools were from four school systems in five urban counties. Though most families were recruited into the FAST program after a teacher had identified one or more of the children in the family as at risk of future academic failure or related problems, some parents self-identified their child for the program. The present study includes data from 4 school years (1996-2000) involving 427 graduating families from 47 separate 8-week cycles of the FAST program in 13 distinct middle schools (see Table 1).

The first 2 years of available data (1996-1997) involving 125 graduating families were excluded from this analysis because of differences in outcome measures that precluded the aggregation of these data in a systematic fashion.

Demographic Data on Participant Families

Demographic data on the 427 FAST families who constituted the sample for this study are presented in Tables 2 and 3. Table 2 shows that the majority of the children identified for the program (target children) were male (59%) and, on average, were approximately 7 years of age. Over half of the children were 7 to 9 years of age (56%), 30% were 4 to 6, and 11% were 10 to 12 years of age. Nearly half of the target children were in the first or second grade (44%), and one third were in the third or fourth grade (31%); 18% were in kindergarten, 1% in prekindergarten, and 6% in fifth grade. On the basis of the race of the child's female guardian (primarily the biological mother), nearly two thirds of the children (64%) were African American, 24% were White, 3% were Hispanic, and 2% were of other races.

Table 3 shows additional detail on the sample of FAST families. The vast majority of children were accompanied to FAST by their mothers (86%). Approximately 5% were accompanied by their fathers and 2% by another guardian (most often a grandmother). On average, the target child's guardian was approximately 34 years of age, with 30% being under 30 years of age and 43% being in their 30s.

Over one third of the children had a single parent (35%), and 22% of parents were separated or divorced; 2% were widowed. Less than one third of the parents (29%) were married, and another 5% were unmarried and living together. Over one half of the families (52%) reported an annual income under $20,000, with more than half of these reporting an income under $10,000 per year. Twenty-one percent reported an annual income between $20,000 and $30,000, and 14% reported more than $30,000 in income.

The families' level of interaction with the FAST program varied somewhat even among those families who graduated from the program. One key element of this interaction is the receipt of a single in-home visit conducted by FAST staff prior to the program; the home visit provides an introduction to the program and serves as an avenue for assessment of the family's situation and preprogram data collection. Over three fourths of the families (82%) received one home visit from FAST staff, and 13% were visited twice or more. Four percent of families did not receive a home visit from a FAST representative but did enroll in and graduate from the program. A second measure of interaction is the number of FAST sessions that families attended. The FAST model involves eight family group sessions, and although attendance is strongly encouraged, many families were unable to attend all eight sessions. On average, families attended seven of the eight FAST sessions, and 69% attended seven or eight sessions. More than one third of the families (35%) attended all of the eight FAST sessions, and another third (34%) attended seven of eight sessions. However, one quarter of families (24%) attended six sessions, and 7% attended five of eight sessions.

Outcome Assessment

The assessment of the effects of the FAST program in this study was based on the program's expressed goals. The evaluation model employed here was initially adopted by the Alliance for Children and Families, the association that sponsored the development and distributed the FAST program. The key outcome measures used in this study involved three dimensions: (a) the target child's behavior in school and at home, (b) general family functioning, and (c) the parent's level of involvement in the child's academic life. No data were collected on the substance abuse dimension because of the short-term nature of the study. The study employed a nonexperimental approach involving a single group pre and post design and relied on data collected through surveys of parents and teachers before and after the 8-week program. Preprogram data were collected by the program facilitator during the 2-week period prior to the start of the FAST cycle. Parent surveys were completed in the family's home during a preprogram home visit. Surveys were distributed to the child's teacher in the middle schools and collected by program staff upon completion. Postprogram data were collected from parents either during the final FAST session or during the following 2-week period. Teacher surveys were distributed and collected at the schools during the same time period.

Child Behavior

Child behavior was measured using a standardized instrument, the Behavior Problems Index (BPI; Zill & Peterson, 1990). The BPI is a 28-item scale developed to measure behaviors common among the general child population. The BPI's total score has six subscales that reflect specific types of behavior: antisocial, anxious/depressed, headstrong, hyperactive, immature dependency, and peer conflict/social withdrawal. This BPI scale has been used with a broad range of youth, and its validity and reliability have been tested (Zill, 1985). Each item (problem behavior) on the BPI is rated as often true, sometimes true, or never true, and each item is given a unit score of 1 if either of the first two categories applies. The BPI includes such items as the child "has sudden changes in mood or feelings," "bullies, or is cruel or mean to others," and "is unhappy, sad or depressed." When used with children, ages 4 to 11, the BPI has been shown to have alpha coefficients ranging from .54 to .73 on the six subscales and a higher reliability of .89 on the total scale (Zill, 1985). Table 4 shows the pre- and post-program ratings by parents and teachers for all children and by the child's gender. The preprogram ratings suggest that less than half of the children would be considered to be in the clinical range. Clinical status was determined using data from Zill (1985) showing a median behavior score for children (ages 6 to 11) who had received or needed psychological help in the prior year of 14.73 for boys and 13.82 for girls. Applying these cutpoints to both parent and teacher ratings showed that 40% to 46% of the children were in the clinical range preprogram, with more boys reaching this level (46% to 49%) than girls (33% to 42%).

A first observation is that parents, on average, observed more problem behaviors among these children than teachers did, both before and after FAST. This is consistent with work by Verhulst, Koot, and Van der Ende (1994) showing that whereas teachers report fewer problem behaviors than parents, teacher ratings are a stronger predictor of poor outcomes for children in the future. The correlation between parent and teacher ratings in the present sample ranged from .46 (preprogram) to .39 (postprogram). In another study, Fischer (1999) reported that when using the BPI, parents and teachers agreed in only 43% of the cases about the presence and direction of the change in the child's behavior. This apparent inconsistency in observer ratings may actually reflect the differences in the child's behavior in the two relevant settings of home and school. In a quantitative synthesis of 119 studies of child behavior, Achenbach, McConaughy, and Howell (1987) reported that although ratings by observers of the same type in the same settings are .60 on average, the overall correlation between parent and teacher ratings is much lower (.27). The authors noted, however, that parent and teacher ratings of children ages 6 to 11 showed significantly higher correlation than those for older children, showing that interrater agreement is generally higher for children in the age range served by the FAST program. The contrast between the ratings provided by parents and by teachers likely reflects both differences in child behavior between the home and school environment as well as other situational differences. Given this, the use of multiple raters of differing perspectives is essential to constructing a more complete picture of an individual child's behavior (Safran & Safran, 1985; Stanger & Lewis, 1993).

In the present sample, the data can be used to examine behavior changes that occurred concurrently with program participation. The data showed that, on average, both parents and teachers observed a statistically significant 10% decline in the children's behavior problems after completion of FAST. In respect to gender, parents reported significant decreases of similar magnitude in behavior problems among boys (10%) and girls (11%). Similarly, teachers reported significant declines in behavior problems among both boys and girls, though the decline for girls (15%) was nearly double that for boys (8%). Both the parents and teachers reported larger average reductions in behavior problems among girls compared with boys. Further examination of the ratings according to the age and ethnicity of the target child showed significantly larger effects preprogram to postprogram among the youngest children (ages 4 to 6) and no differences by major ethnic categories (African American vs. White).

An additional analysis was conducted examining the reported changes in child behavior broken out by ratings of the child's behavior preprogram. A working hypothesis of program staff was that the program proved most effective when target children have moderate behavioral issues. However, because FAST is a community-based program, the identification process did not systematically screen out children with extreme behavior problems or those with few behavioral issues. To examine the role of preprogram behavior of children in the changes observed after FAST, I divided the sample into thirds, first based on the parents' ratings and then based on the teachers' ratings.

Ratings provided by parents were analyzed on the basis of the child's status preprogram (see Table 5). Among all children, parents reported a statistically significant 22% increase in behavior problems for the children who had few behavior problems before FAST, a significant 8% decrease in problems among those with moderate behavior problems preprogram, and a significant 21% decrease among those with the most severe behavior problems preprogram. This same pattern is consistent with the data on the subsamples of female and male children as rated by parents.

A similar analysis was conducted using the ratings provided by teachers (see Table 6). Among all children, teachers reported a nonsignificant increase in behavior problems for the children who had few behavior problems before FAST, a statistically significant 10% decrease in problems among those with moderate behavior problems preprogram, and a significant 16% decrease among those with the most severe behavior problems preprogram. This pattern was also repeated in the subsample of girls and boys, except that teachers observed no change in the behavior of girls with few problems preprogram, whereas they observed the largest change among boys with few problems preprogram (a worsening of behavior pre- to postprogram).

According to the pattern of findings on child behavior, parents and teachers reported the largest improvements among children with the most severe behavior problems preprogram, on the order of 16% to 21% improvement. Children with moderate behavior problems preprogram were reported as having more modest improvements (7% to 10%), yet these results were still statistically significant. Lastly, though some individual results were not statistically significant, children with the fewest behavior problems preprogram appear to have exhibited a worsening of behavior. Clearly, this finding is of concern; however, it should be noted that the higher problem behavior scores at posttest were still well below the standard for a clinical level of behavior problems. On the basis of data from Zill (1985) the FAST children in this cohort had behavior scores at approximately one third to one half of the clinical standards at posttest. It is unclear, however, whether this finding indicates that the program experience led some previously inhibited children to act out more after the program, or if this may be due to rater perception. An examination of the six subscales of the BPI showed that the worsening of behavior in this cohort was not uniform; rather it was concentrated in two subscales: (a) Headstrong and (b) Immature Dependency. Anecdotal reports suggested that the use of the BPI might have caused some raters to more closely observe the target children after the preprogram data collection, such that their postprogram ratings reflected more rigorous documenting of behavior problems.

Family Adaptability and Cohesion

Family functioning was measured pre- and postprogram using the 10-item Family Adaptability and Cohesion Evaluation Scales (FACES III), which were completed by the child's parent (Olson, Portner, & Lavee, 1986). Contained in the items in the FACES III form are two separate subscales-one measuring adaptability and the other cohesion. Olson and others developed the FACES III instrument to measure the underlying constructs of their circumplex model of family functioning, family adaptability (flexibility), and cohesion (bonding). It should be noted that there has been considerable debate in recent years about the validity and reliability of the FACES III instrument, to the extent that Olson et al. produced a newer, 24-item FACES IV instrument to address some of these concerns (Franklin, Streeter, & Springer, 2001). As such, the data presented here based on the earlier FACES III should be interpreted with some caution (see Table 7).

Overall, the parents reported significantly higher adaptability and cohesion after completing FAST. Broken out by the gender of the target child, the Cohesion subscale showed significant increases for both boys and girls. In regard to the Adaptability subscale, however, only the girls showed significant increases.

Parental Involvement and Support

Parent behavior was measured using a self-report instrument, the Parental Involvement and Family Support Survey (Family Service America, 1990). The parent completed the survey before and after the 8-week FAST program. The form included a battery of questions related to specific kinds of interaction between the parent and child and the frequency of their occurrence. These activities included the parent's contact with the child's school and the parent's involvement in the child's academic life (e.g., helping child with homework, attending events at school). According to paired data from parents, no statistically significant differences were found in the categories of parental involvement. Apparent differences between involvement before and after the 8-week program were confounded with the activities associated with the program itself. Involvement measures taken at follow-up, some time after the program had concluded, hold better promise as meaningful measures of parental behavior change.

The survey also asked the parents to answer a seven-item section dealing with family support that attempted to measure the parents' feelings of connectedness to others and their satisfaction as a parent. Overall, the parents reported statistically significant increases on the measure of family support from before FAST to after the program. For all children the result was a statistically significant 4.0% increase in family support (p = .004, n = 124). When the results were broken out by the gender of the child, parents of boys reported a statistically significant 6.6% increase in family support (p = .001, n = 76), whereas for girls the difference was not significant (p = .639, n = 48).

Parental Satisfaction

In addition to the data previously described, parents were also asked to provide feedback on their family's program experience through a 14-item written survey. This instrument was implemented with participants during the last 2 years of the study (1998-1999), and 93% completed the survey (n = 208). The survey data showed that parents reported that their families enjoyed the program and its elements of the program; there was 95% to 98% agreement with the specific items (e.g., "FAST was helpful to myself and my family"). Furthermore, parents reported that the program led to important changes in themselves and their families: better relationship with their child (84%), better relationship with child's school (72%), better relationship with other parents (87%), and better child behavior at home (82%). In addition, the parents reported that the program affected their own behavior at home in a positive way (89%) and that they planned to use things they learned in the program in their parenting in the future (94%). These self-report data suggest that parents believed the program to be both enjoyable and beneficial in tangible ways to their families.

Caveats to the Study

Though the data from this field-based study shed additional light on the potential impact of structured family support interventions such as FAST, there are limitations to the methods that should be noted. A first limitation stems from the evaluation design itself and relates to determining program impact and causal attribution. The single-group design employed severely limits any ability to attribute differences observed to the program itself. However, it should be noted that randomization to groups, including a no-treatment control, was deemed unacceptable by the program deliverers and the host schools on ethical grounds. Similarly, no reasonably comparable group of families receiving either no services or alternate services could be located for the purposes of comparison. Unfortunately, the experimental evaluations available on the FAST program do not provide a consensus on the program's impacts (Layzer, Goodson, Creps, et al., 2001; McDonald et al., 1991). A second concern is the selectivity of the sample of families who graduated from the program relative to the pool of potentially eligible families and the families who actually attended the program. Overall, approximately two thirds of the families (68%) referred for the program agreed to a home visit so that they could be introduced to the program, and of these, 87% actually received a home visit. Of families receiving a home visit, 79% subsequently attended at least one session of the FAST program, and of those attending, 80% completed the program. In sum, the sample of graduating families represents 37% of referred families, 64% of families who received a home visit, and 80% of families who attended at least one session of the program. Given the wholly voluntary nature of the program, no data are available on the characteristics of the families who declined or did not receive a home visit or for families who attended but did not complete the program. Thus, there is no systematic way to ascertain the representativeness of the program graduates relative to the broader populations of families. Anecdotal information from program staff suggests that program graduates tended to overrepresent families who were either the lowest functioning or the highest functioning. Recent work by McCurdy and Daro (2001) highlighted the complexity of engaging parents in family support programs and of keeping them involved over time.

A third limitation of note is the short-term focus and narrow nature of the outcomes measured, which relied chiefly on pre- and postprogram measurement over a 2- to 3-month period, in three primary areas. Although these data illustrate the most proximal changes occurring in these children and their families, they are unable to address the duration of these effects over time. Nor are these data able to address the other potential effects in regard to such measures as the child's academic achievement and social development and the family's stability and civic engagement. Further research, including longitudinal investigations, is needed to address these issues.

Conclusion

This investigation of a family support intervention details the evaluation of one widely implemented program model, FAST. The data from this study contribute to the knowledge base in a distinct way. The data presented come from one of the largest operating sites of the FAST program, reflecting on a program that has been implemented over time in a field-based setting and suggesting what occurs when the program is taken to a larger scale. An additional dimension is that the participants in the program represent an important extension of the program to a majority African American population in an urban setting.

Using field-based data, this study offers an examination of the immediate outcomes of the program among an ethnically diverse group of urban families. The data collected through a systematic pre- and postevaluation of the program demonstrate that the program is showing promising results in the primary near-term goals. Statistically significant changes were found in regard to child behavior problems, family adaptability and cohesion, and family support. The findings on child behavior showed that the targeting of the program to the families of children with moderate to high behavior problems resulted in the greatest positive changes. Children with few behavior problems preprogram may actually exhibit more behavioral problems in the short term, though these problems were far from clinical levels of concern. This targeting issue, along with the sustained effects of the program after the 8-week period and the additional influence of the FASTWOKKS component, are issues that require further exploration.

References

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

Alliance for Children and Families. (2000). FAST orientation manual. Milwaukee, WI: Author.

Chaika, G. (2000). Bringing families and schools together-FAST! Retrieved May 31, 2001, from www.education-world.com/a_admin/admin170.shtml

Devaney, E. S., & Milstein, J. P. (1998). Kids + family + school = success: A kindergarten student and family support program. Social Work in Education, 20, 131-138.

Family Resource Coalition of America. (1998, April). Guidelines far family support practice. Paper presented at the Safe Families Symposium, Stone Mountain, GA.

Family Service America. (1990). Parental Involvement and Family Support Survey. Milwaukee, WI: Author.

Favorini, A., & Pryor, C. (1994). Family-school alliances: A centerpiece strategy for alcohol and drug prevention programs. Social Work in Education, 16, 155-169.

Fischer, R. L. (1999). Children in changing families: A pilot-study of a program for children of separation and divorce. Family and Conciliation Courts Review, 37, 240-256.

Fischer, R. L. (2000). Families and Schools Together (FAST): 1999-2000 school year overview. Atlanta, GA: Families First.

Franklin, C., Streeter, C. L., & Springer, D. W. (2001). Validity of the FACES IV family assessment measure. Research on Social Work Practice, 11, 576-596.

Greif, G. L. (1993). A school-based support group for urban African-American parents. Social Work in Education, 15, 133-139.

Kumpfer, K. L., & Alvarado, R. (1997). Effective family strengthening interventions (Juvenile Justice Bulletin Publication No. NCJ 171121). Washington, DC: Office of Juvenile Justice and Delinquency Prevention, U.S. Department of Justice.

Layzer, J. I., Goodson, B., Bernstein, L., & Price, C. (2001). National evaluation of family support programs: Volume A. The meta-analysis. Cambridge, MA: Abt Associates.

Layzer, J. I., Goodson, B., Creps, C., Werner, A., & Bernstein, L. (2001). National evaluation of family support programs: Volume R. Research studies. Final Report. Cambridge, MA: ABT Associates.

McCurdy, K., & Daro, D. (2001). Parent involvement in family support programs: An integrated theory. Family Relations, 50, 113-121.

McDonald, L. (1999). Families and Schools Together (FAST): 1988-1998 Ten years of evaluation. Unpublished manuscript, University of Wisconsin-Madison.

McDonald, L., Billingham, S., Conrad, T, Morgan, A., Nina, O., & Payton, E. (1997). Families and Schools Together (FAST): Integrating community development with clinical strategies. Families in Society, 78, 140-155.

McDonald, L., Bradish, D. C., Billingham, S., Dibble, N., & Rice, C. (1991). Families and Schools Together: An innovative substance abuse prevention program. Social Work in Education, 13, 118-128.

McDonald, L., & Frey, H. E. (1999). Families and Schools Together. Building relationships (Juvenile Justice Bulletin Publication No. NCJ 173423). Washington, DC: Office of Juvenile Justice and Delinquency Prevention, U.S. Department of Justice.

McDonald, L., & Sayger, T. (1998). Impact of a family and school-based prevention program on protective factors for high-risk youth. Drugs & Society, 12, 61-86.

Moberg, D. P., McDonald, L. W., Brown, R., & Burke, M. (2003, June). Randomized trial of Families and Schools Together (FAST). Presented at the Society for Prevention Research 11th Annual Meeting, (Washington, DC).

Olson, D. H., Portner, J., & Lavee, Y. (1986). Family Adaptability and Cohesion Scale III. Department of Family Social Science. St. Paul: University of Minnesota.

Penry, W. P. (1996, August 19). FAST: Proven support to at-risk children and their families. Family Therapy News, 27(4), 12-13.

Pinsoneault, L., & Sass, J. S. (1999). Evaluating the national replication of a prevention program for youth and their families: Middle school Families and Schools Together. The Evaluation Exchange, 2/3,pp. 10-11.

Safran, S. P., & Safran, J. S. (1985). Classroom context and teachers' perceptions of problem behaviors. Journal of Educational Psychology, 77, 20-28.

Sayger, T. (1996). Creating resilient children and empowering families using a multifamily group process. Journal for Specialists in Group Work, 21, 81-89.

Stanger, C., & Lewis, M. (1993). Agreement among parents, teachers, and children on internalizing and externalizing behavior problems. Journal of Clinical Child Psychology, 22, 107-115.

Verhulst, F. C., Koot, H. M., & Van der Ende, J. (1994). Differential predictive value of parents' and teachers' reports of children's problem behaviors: A longitudinal study. Journal of Abnormal Child Psychology, 22, 531-546.

Zill, N. (1985). Behavior problem scales developed from the 1981 child health supplement to the National Health Interview Survey. Unpublished summary.

Zill, N., & Peterson, J. L. (1990). Behavior Problems Index. Washington, DC: Child Trends, Inc.

Robert L. Fischer, PhD, Is senior research associate, Center on Urban Poverty and Social Change, Mandel School of Applied Social Sciences, Case Western Reserve University, 10900 Euclid Avenue, Cleveland, OH, 44106-7164. E-mail: fischer@cwru.edu.

Author's note. During the course of this study the author was employed as the director of program evaluation at Families First, a nonprofit family service agency in Atlanta, Georgia. The author wishes to thank Ms. Joyce Sloan, LCSW, of Families First for her ongoing input to this research and the FAST families who gave of their time and energy for this work. In addition, thanks go to Kimberly D. Farris, MSW, for contributing to the preparation of this research and to Kathleen Quinn-Leering, PhD, for providing useful comments on drafts of the manuscript.

Manuscript received: October 31, 2001

Revised: August 1, 2002

Accepted: August 14, 2002

Copyright Families in Society Jul-Sep 2003
Provided by ProQuest Information and Learning Company. All rights Reserved