PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
 
Prev Sci. Author manuscript; available in PMC 2010 April 1.
Published in final edited form as:
PMCID: PMC2848498
NIHMSID: NIHMS58218

Community and Team Member Factors that Influence the Operations Phase of Local Prevention Teams

The PROSPER Project
Mark E. Feinberg, Sarah M. Chilenski, and Mark. T. Greenberg
Prevention Research Center, Pennsylvania State University, Henderson S-109, University Park, PA 16802, USA

Abstract

This study examined the longitudinal predictors of quality of functioning of community prevention teams during the “operations” phase of team development. The 14 community teams were involved in a randomized-trial of a university-community partnership project, PROSPER, that implements evidence-based interventions intended to support positive youth development and reduce early substance use, as well as other problem behaviors. The study included a multi-informant approach to measurement of constructs, and included data from 137 team members, 59 human service agency directors and school administrators, 16 school principals, and 8 Prevention Coordinators (i.e. technical assistance providers). We examined how community demographics and social capital, team level characteristics, and team member attributes and attitudes are related to local team functioning across an 18-month period. Findings indicate that community demographics (poverty), social capital, team member attitudes towards prevention, and team members’ views of the acceptability of teen alcohol use played a substantial role in predicting various indicators of the quality of team functioning 18 months later.

Keywords: Community, Coalition, Lifecycle

Introduction

Several authors have noted the popularity and intuitive appeal of community coalition and partnership approaches to addressing public health problems (Backer 2003; Wandersman and Florin 2003). Coalitions can facilitate flexible, multi-sectorial planning and coordination to address problems that arise through complex interactions between multiple levels of communities (e.g. individual behaviors, family relations, neighborhood culture, school quality, economic stress; Bronfenbrenner 1986). In addition, collaborative partnerships1 may facilitate local, democratic processes and empowerment—thereby enhancing local social capital.

Despite these apparent benefits, research on the outcomes of coalition or partnership approaches have generally produced null results (COMMIT 1995; Hallfors et al. 2002; Roussos and Fawcett 2000a; Saxe et al. 1997), which have led some to question the value of the partnership approach in promoting public health activities (Klerman et al. 2005). There are three main reasons for the failure of coalition models to demonstrate positive effects. The first has to do with the difficulties in designing and conducting evaluations that demonstrate positive effects in a definitive way, especially considering that in most models coalitions select different interventions and target different outcomes (Green and Kreuter 2002; Kreuter et al. 2000; Roussos and Fawcett 2000b). Second, coalition models have typically not included the following “best practices” recommended by Hallfors (Hallfors et al. 2002): limited and clearly focused goals, outcomes, and benchmarks; use of evidence-based intervention approaches; and careful monitoring of dosage and quality of program implementation. Finally, in order to implement these best practices effectively, coalitions require an adequate dosage of high-quality and pro-active training and technical assistance support (Chavis 1995; Feinberg et al. 2002; O’Donnell et al. 2000; Spoth et al. 2004b).

When coalition models incorporate sufficient technical assistance and a commitment to best practices, appropriately designed evaluations have demonstrated positive outcomes (e.g. Feinberg et al. 2005; Redmond et al. 2007; Spoth et al. 2007). However, maximal effectiveness and sustainability of coalitions requires that both the designers of coalition models and technical assistance providers have a rich understanding of the factors—both internal and external to coalitions—that influence effective coalition functioning (Butterfoss et al. 1996; Feinberg et al. 2004b; Florin et al. 1993; Wandersman et al. 1996). In contrast to intervention delivery through organizations such as human service agencies or schools, the lack of an established organizational boundary and reliance on the resources of member organizations can leave partnership functioning highly susceptible to the influence of leadership quality, membership participation and cohesion, turf battles, community sentiment, changing funding opportunities, and the like.

Moreover, it is likely that the relative importance of internal and external factors varies across the overlapping, but distinct phases of partnership development (e.g. organization and formation, operations, and institutionalization phases). Untangling causal chains of influence in coalition development is best guided by longitudinal studies that examine communities from the initial formation through program implementation. This report is based on longitudinal data collected from a randomized trial of a coalition model, PROSPER, that aims to disseminate evidence-based preventive interventions targeting youth substance use and other problem behaviors (Spoth et al. 2007). This paper examines the predictors of high-quality functioning of community-based prevention teams.

The PROSPER model is comprised of three tiers: local prevention teams, university prevention scientists, and Prevention Coordinators, who serve as liaisons between the teams and scientists by providing technical assistance, guidance, and consultation to the teams and feedback to the scientists. The local prevention teams are catalyzed and led by university Cooperative Extension Service (CES) educators, who are charged with bringing scientific advances to local communities in a wide array of areas (e.g. agriculture, positive youth development, nutrition, and community development). The local teams are comprised of representatives from the school system, key substance use and mental health agencies, parents and youth, and other local leaders.

In the first 5 years of the project, the teams are charged with selecting, implementing, and obtaining sustainable funding for one evidence-based family-focused and one school-based intervention. Interventions are selected from short menus of similar family-focused interventions appropriate for 6th graders and their families, and of classroom programs appropriate for seventh graders. Initial findings have demonstrated positive effects of PROSPER in randomly-assigned intervention communities compared to controls on child-reported proximal family relations (e.g. parental management of child behavior, parent-child affective quality and substance use; see Redmond et al. 2007; Spoth et al. 2007). In future work we plan to investigate the relation of coalition functioning to these family and youth outcomes. In this paper, we investigate the determinants of coalition functioning among the intervention communities at a particular phase in the coalition lifecycle, 18 months after they were formed.

The Lifecycle of Collaboration

As with other models of partnership processes (Chinman et al. 2004; Florin et al. 2000; Hawkins et al. 2002), we posit a partnership model in which local community teams proceed through a series of broad developmental phases (Livit and Wandersman 2004; Stevenson and Mitchell 2003). The first, organizational phase usually lasts for 6-8 months and involves partnership formation activities including recruiting key members, receiving training in the model, deciding on programming goals based on local needs and resources, and coalescing as a team. The second phase, the operations phase, consists of implementing chosen programs and/or policies; its length may depend on the model, length of initial funding and other considerations; in PROSPER this phase lasts 2-3 years. The third broad phase, which overlaps with phase 2, is described as the sustainability phase. This phase focuses on sustaining the effective activities of the local partnership and often involves engaging other community entities to create a more permanent structure for the team’s operations and sponsored activities.

This report extends work reported in an earlier article that examined individual differences in effective functioning among 14 PROSPER local community partnership teams during the team organizational phase (Greenberg et al. 2007). Results indicated that both team member and community characteristics, assessed as the teams were forming (at the pretest, or Wave 1 of the data collection), predicted team functioning during the organizational phase, assessed 6 months later (Wave 2). The current report utilizes data from both Waves 1 and 2 to predict effective coalition functioning during the operations phase—in this case, 18-months after team initiation (Wave 3).

By our own observation and through reports of the Prevention Coordinators, we understand the period assessed at Wave 3 is considered to be relatively stressful for the teams. The teams have completed the implementation of a family-focused intervention offered to families of sixth graders and are planning implementation of the same program with a second cohort of families the next year. The teams also are coordinating with their respective school systems to support the implementation of a seventh grade, classroom-based prevention program. Issues that arise at this operations phase include team member burn-out after the initial surge of activity, the need for team leaders to maintain interest and enthusiasm of local team members, recruitment and orientation of new members to formal and informal aspects of the coalition, ongoing attention to internal team dynamics (e.g. conflict, strain), and planning for sustainability.

Multiple Levels of Influence

We conceptualize factors at three levels that may impact local team functioning: the community level, the team level, and the individual (team member) level.

Community Level

There has been little examination of the how the structural, financial, and social capital characteristics of a community may influence team functioning or sustainability (Herrenkohl et al. 2000; Osgood and Chambers 2000; Reisig and Cancino 2004). For example, communities that have few well-functioning institutions, have a history of mistrust or failure, or have a relatively hierarchical (vs. democratic and collaborative) local political culture may have more difficulties with developing and sustaining partnerships. Measures of social capital such as the readiness of a community to support collaborative efforts, as well as a history of collaboration, have been shown to positively predict early team functioning (Feinberg et al. 2004a; Foster-Fishman et al. 2001; Jausa et al. 2005). In addition, in our prior report on the factors influencing PROSPER team functioning at the organizational phase, the community level of poverty was relatively strongly associated with the quality of team functioning (Greenberg et al. 2007). We hypothesized that higher rates of community poverty may stress agencies and institutions, with greater territoriality between agencies as a consequence.

While low levels of resources and social capital may present barriers to positive local team functioning, a greater perception of need for an initiative should be a motivating factor for team members that would facilitate cooperation and collaboration. Thus, we also examine whether perceptions of the community’s acceptance of substance use, implying a need for an intervention, is associated with more positive partnership qualities.

Team Level

Given the number of tasks to be accomplished during the period under study in this report (see above), we hypothesize that support from technical assistants will be an important factor in predicting the quality of team leadership, task-focus, cohesiveness, and conflict resolution. The PROSPER model includes an “enabling system framework” (Chavis et al. 1992) to support local capacity building: PCs with substantial experience either in prevention science or Cooperative Extension programming serve as liaisons between university prevention scientists and the local teams. Each Prevention Coordinator (PC) provides technical assistance and regular proactive guidance to one or two local teams (on a part-time basis), including attendance at meetings and regular communication with the team leaders. Recognizing the potential role of this support in effective local team functioning, we include measures of the teams’ openness to PC support as potential predictors of team functioning.

Individual Team Members

Collaborative entities are comprised of individuals, and a large social psychological literature investigates the characteristics of individuals that contribute to cooperative task performance (Kerr and Tindale 2004). In this context, we are most concerned with member characteristics related to knowledge of prevention science, attitudes toward substance use, and skills members bring to the collaboration. In our prior report, we found that attitudes towards prevention, expectations of the project, and perceived partnership skills predicted various indices of early team functioning (Greenberg et al. 2007). We hypothesize that initial attitudes towards prevention, history of past collaborations, and skills may decline as predictors of team functioning over time, because the experience of engaging in intervention implementation, the evolving team culture, and local community, agency, and political events and dynamics would contribute to team functioning over time. Moreover, we examine whether team members’ own norms regarding adolescent substance use predict team functioning; we hypothesize that when members personally do not accept or tolerate adolescent substance use, they will be more committed to and engaged with the local prevention team.

Finally, we investigate the ratio of perceived personal benefits to costs derived from serving on the PROSPER teams. Individuals report a range of benefits (e.g. sense of helping others or the community, networking, learning, developing skills) and costs (e.g. time, frustration) associated with coalition involvement (Wandersman et al. 1987). We hypothesize that a higher ratio of benefits to costs will predict more positive coalition functioning. We utilize longitudinal data in which benefits/costs are assessed at the organizational phase (i.e. Wave 2. All other predictors in this study are assessed at Wave 1, however, members were not in a position to report on benefits and costs of involvement at that point).

Team-level Outcomes

While a number of studies have examined coalition-level processes, the bulk of this research has been cross sectional, and thus unable to examine which predictors are linked to team functioning in different phases of operation (Feinberg et al. 2004a; Greenberg et al. 2007). Nonetheless, this literature points to potentially important factors that can be considered either predictors of later team functioning or indicators of team functioning (i.e. dependent variables). For example, within-team factors of participation, leadership, task-focus, cohesion, and identity are related to indicators of early team success (Florin et al. 2000; Gottlieb et al. 1993; Kegler et al. 1998; Greenberg et al. 2007). Teams that lack effective leadership, are riddled with conflict, or lack motivation make ineffective decisions or show poor implementation (Emshoff et al. 2003). Based on these and other studies, we focus on four team-level constructs as indicators of effective functioning in this paper: leadership, work focus, team culture of cohesion, and internal team tension/conflict.

Multi-informant Research

This project examines both predictors and outcomes of team processes from the perspective of multiple informants. Community and coalition processes are complex, and participants’ may have idiosyncratic or biased views. For example, a meeting may proceed smoothly from the point of view of one participant, but another who disagreed with decisions made may have an opposite view. Thus, we utilize both team members and PC reports of team functioning, in separate analyses, to investigate whether the predictive associations hold across different reporters’ ratings of team functioning. In addition, the predictors of team functioning are variables representing a range of sources. The sources for these variables include census data and archival datasets, as well as interviews and question-naires administered to team members, directors of local human service agencies, and school principals.

Materials and Methods

Procedure

The participating universities’ Institutional Review Boards authorized the study before participant recruitment began. Detailed information about the community recruitment is provided elsewhere, see (Greenberg et al. 2007). Briefly, 28 communities (14 in Iowa and 14 in Pennsylvania) were selected that met four project eligibility criteria: (a) a community school district enrollment of 1,300-5,200 students; (b) 15% or more of the school district families eligible for free or reduced cost lunches, (c) less than 50% of the community population employed by or attending a college or university, and (d) no current involvement in any other university-affiliated prevention research projects with youth. After the 28 communities were selected, communities were blocked by size and geographic location and then randomized into partnership intervention and control conditions. Communities in the overall sample are mostly rural and small towns, with an average of 7% of the families living below the poverty threshold, a median household income of $37,070, and 29% of the students in the district receiving free or reduced-cost lunches.

Data were collected in three waves (see Table 1). Wave 1 occurred within 2 months of team initiation (Spring 2002). Wave 2 occurred approximately 6 months later (Fall 2002), shortly before the implementation of the family program for Cohort 1. Wave 3 occurred 1 year later (Fall 2003); at that point, Cohort 1 was in seventh grade and teams were coordinating with schools regarding implementation of a classroom based program. Teams also were preparing for the implementation of the family program for Cohort 2. By Wave 3, the community teams had been meeting for approximately 18 months.

Table 1
PROSPER intervention coalition activities and assessment schedule, 2002-2003

Participants

The respondent sample includes four types of individuals from the 14 PROSPER intervention communities in Iowa and Pennsylvania: 1) prevention team members, 2) directors of human service agencies, 3) middle school principals, and 4) PCs. In a limited number of cases a single individual served in multiple roles and there is duplication across categories. For example, team members may also be agency directors or school principals. For this reason, the total number of respondents does not equal the sum of each respondent profile.

Prevention Team Members

The prevention team member sample includes 137 individuals that participated over one or more of the three waves. Team members include local Cooperative Extension and school representatives, local substance abuse and mental health agency representatives, and parents (youth members were not interviewed). For data collection, we selected from each team at least one substance abuse and one mental health agency representative (if available), the team leader (i.e., the Extension Educator), at least one school representative, two parents, and representatives from other organizations or businesses represented on the team. An average of 9.8 individuals per community responded about team functioning and community characteristics over the three waves (range: 7-14). Respondents ranged in age from 24 to 59 (M=42.6, SD=8.47) and 31% are male. Ninety-nine percent of the sample self-identified as White, which is consistent with community demographics (96% White). All respondents had a minimum of a high school education, with 88% of the sample having obtained a college degree. Eighty-seven percent of the sample lived in or near the school district that organized the PROSPER team.

Directors and Administrators

Directors and administrators of local organizations linked to PROSPER (including human service agencies and schools) were interviewed about their organizations’ resources, support of prevention, and support of PROSPER. The director and administrator sample includes 59 individuals who participated at the first wave. Directors and administrators were selected to be interviewed if they were the supervisor of a PROSPER team member (N=38), or if they themselves were a team member but did not have an immediate supervisor willing to participate who could provide information at an organizational level (N=21) An average of 4.2 Agency Directors participated per community (range: 0-6)2. Respondents ranged in age from 31 to 62 (M=49.1, SD=8.06), 61% are male, and 100% self-identified as White. All respondents completed a minimum of a high school education or GED, with 95% of the sample having obtained a college degree. Fifty-seven percent of the sample lived in or near the school district that organized the PROSPER team.

Middle School Principals

In order to collect information about school programs and policies, we interviewed principals from all schools in intervention communities that contained a seventh grade, and 16 principals participated over two waves. One community had two school buildings with seventh grades. In that one community, we utilized the mean score across the two principal interviews to arrive at a community-level score. Across all principals, tenure at their particular school averaged 7.5 years (range: 0-23; SD=7.12).

Prevention Coordinators (PCs)

The nine PROSPER PCs are the community teams’ primary technical assistance contacts, and they are the linking mechanism between university researchers and local community teams (Spoth et al. 2004a) as described in detail earlier. Thirty-three percent of the PCs are male and all are White. Prior to their work with PROSPER, PCs had an average of 19.4 years of experience in Cooperative Extension and/or prevention programming.

Assessment

At each wave, participants in the Team Member and Directors samples participated in 1-h face-to-face computer-assisted interviews that assessed individual, community, and workplace characteristics, and local team dynamics. Directors who were also were members of a team received both interviews in an integrated fashion. Participants were compensated $20 for their participation. At Waves 2 and 3, the principals (including those principals who participated in the Team Member or Director interview) participated in a 1-h face-to-face computer-assisted interview that assessed school climate and resources. School principals were not individually compensated for their participation in this interview (although modest incentives for all data collection, including student surveys, were paid to schools). PCs responded to several questions assessing team dynamics and contact through a web-based quarterly survey from the Spring to the Fall of 2002 (three quarterly assessments).

Measures

Several constructs describing the community, team, and school were created. Unless otherwise noted, response options for the scales range on a four-point Likert-type scale from Strongly Disagree to Strongly Agree. All scales are formed by taking the mean of the scale items and then aggregated to the team level. Descriptive statistics of all scales are reported in Table 2.

Table 2
Descriptive Statistics (measured at Wave 1 unless otherwise specified)

Community Demographics

Six measures describe the community context; two community structural characteristics and four social capital measures. The two community structural measures are community poverty and district academic rank. Community poverty, or the percent of families living below the poverty threshold (US Census 2000) was aggregated to the school district level by the National Center for Education Statistics (2003). The district academic rank was based on the percent of 8th grade students that tested as “proficient” or above on standardized tests as defined by each state’s academic standards in the 2001-2002 academic school-year 3. Given that the two states use different standardized tests, the ranking of the districts in the sample were computed separately by state. The final variable was the mean of each district’s math and reading (within-state) rank within the PROSPER sample. Low scores on this variable indicated a low rank, that is, relatively low average math and reading scores for the district (Chilenski-Meyer et al. 2007).

The four community social capital measures are community readiness, substance use norms, community collaboration, and school collaboration. Team Members and Agency Directors reported on four conceptually based subscales assessing community readiness, the pre-existing capacity of a community to implement a successful change-effort (α=0.75;). Team Members responded to six items measuring their community’s substance use norms (α=0.83); an example item is, “Adults in (this community) think the use of alcohol is a normal part of growing up.” Agency Directors responded to six items assessing community collaboration; the degree to which community organizations collaborate around prevention activities (α=0.90)4. School Principals reported on school collaboration (α=0.85; adapted from Arthur et al. 1998), the total number of joint planning activities, sharing of funds, and cosponsoring relationships that the middle school has with community organizations.

Team Characteristic at Wave 1

At Wave 1, six scales were used to assess team member attitudes, knowledge, and experience before PROSPER. Collaboration experience assesses members’ prior involvement in collaborations and their leadership within the collaboration. Value of prevention (two-items, r=0.42) assesses the degree to which members see prevention programs as valuable; an example item is, “Violence prevention programs are a good investment.” PROSPER expectations (11-items, α=0.82) assesses the degree to which team members expected that PROSPER could make positive changes in the community. Team member acceptance of adolescent alcohol use was assessed with one yes/no question (1 = yes/0 = no), “Is it ever ok for teenagers to drink alcohol?” Team member skills and resources (13-items, α=0.85) assesses the amount of skills and resources team members bring to the team process.

Team Perceived Net Benefits

At Wave 2, team members reported on perceived net benefits, or the balance of benefits to costs associated with participation in PROSPER (Feinberg et al. 2002). The net benefits scale was derived using three steps. First, a benefit scale (five-items, α=0.73) was created from items such as, “How much benefit have you gained from your involvement with PROSPER in these areas: learning new skills?” Second, a cost scale (three-items, α=0.78) was created from items such as, “How much has PROSPER interfered with: your work schedule and responsibilities?” Third, the standardized costs scores were subtracted from the standardized benefits scores, yielding perceived net benefits.

Contact with Technical Assistance

At Wave 2, two measures described the teams’ interactions with technical assistance. PCs responded to seven items to assess team’s effective TA collaboration (seven-items, α=0.85), the degree to which the team communicates and works effectively with the Prevention Coordinator. PCs also assessed the level of team contact with PC (two-items, r=0.52).

Team Functioning Outcomes

At Wave 3, four scales assessed team functioning. Three of the four constructs were assessed by both team members and the PCs (team focus on work, team culture, and team tension), and one construct was assessed by team report only (team leadership). Team focus on work (five-items, Team α=0.72; PC α=0.89; adapted from Moos 1981) assesses the work-orientation of the team; an example item is, “People pay a lot of attention to getting work done.” Team leadership (eight-items, α=0.85, adapted from Kegler et al. 1998) assesses the degree to which team leadership encourages input and consensus, along with promotes a friendly work-environment; an example item is, “the team leadership...gives praise and recognition at meetings.” Team culture (eight-items, Team α=0.91; PC α=0.89, adapted from Kegler et al. 1998) assesses the team atmosphere; an example item is, “there is a strong feeling of belonging in this team.” Team tension (one-item, four-point scale ranging from 1 = no tension to 4 = a lot of tension) assesses the degree of conflict and tension in the PROSPER team. PC ratings were aggregated over three quarters, from Spring 2002 to Fall 2002, corresponding to the period reported on by team members at Wave 35.

Statistical Analysis

Two sets of statistical analyses were conducted. Both sets followed the same steps and used the same predictor variables. The first set utilized team member reports of team functioning, whereas the second set of analyses utilized PC reports.

Each set of analyses followed three steps: First, zero-order correlations were calculated between the set of predictors and the measures of team functioning. Second, given the importance of community poverty in earlier analyses (Greenberg et al. 2007), partial correlations which controlled for level of community poverty were calculated. As in our previous report (Greenberg et al. 2007), we cautiously interpret correlations that are equal to 0.38 (p=0.10) or higher, as a community-level sample size of 14 affords limited statistical power. Third, in order to assess whether predictor variables account for overlapping variance in team functioning, or whether suppressor effects may be present, we conducted multiple regression analyses. Because the sample size limits the degrees of freedom, we restricted these models to a limited number of predictors by conducting these regression analyses separately for each of four conceptual domains of predictors: community demographics, community social capital, early team characteristics, and contact with technical assistance. Thus, we conducted four regression models for each dependent variable, and each regression model included all predictors in a domain.

Results

Preliminary Analyses

Table 3 presents the zero-order correlations among the Wave 1 and Wave 2 independent variables. Moderate to strong associations were found among the community structural and social capital measures; higher poverty communities had a lower district academic rank (r=-0.78), lower ratings of readiness (r=-0.39), and were more accepting of adolescent substance use (r=0.36). Communities that were more accepting of adolescent substance use also had lower ratings of readiness (r=-0.62) and their middle school(s) had more collaborative relationships with community organizations (r=0.48).

Table 3
Zero-order correlations among Predictor Variables

A few cross-domain associations are also worth noting. For example, communities that had higher ratings of collaboration had teams that perceived more net benefits (r=0.62). Teams with members that were more accepting of adolescent alcohol use perceived fewer net benefits (r=-0.67) and were less cooperative with their PCs (r=-0.38). Additionally, more contact with the PC was positively associated with community poverty (r=0.61) and negatively associated with readiness (r=-0.39). PC contact was also positively associated with team member acceptance of adolescent substance use (r=0.42), and community collaboration (r=0.40).

Team Member Report of Team Functioning

Zero-order Correlations Between Predictors and Team Functioning Variables

Zero-order and partial correlations (controlling for community poverty) between predictor and dependent variables are presented in Table 4. Of the community structural and social capital characteristics, Wave 1 measures of community poverty, community readiness, community substance use norms, and school collaboration had moderate associations with ratings of Wave 3 team functioning. For example, higher poverty communities had team leaders that were rated as less skilled (r=-0.59), whereas communities that had higher ratings of readiness had a more positive team culture (r=0.40). Both community substance use norms and school collaboration were associated with Wave 3 ratings of team tension: communities that were rated as more accepting of adolescent substance use (r=-0.39) and communities that had middle schools with more collaborative relationships (r=-0.50) had prevention teams with less tension and conflict.

Table 4
Zero-order and partial correlations (controlling for poverty) between predictor and team-member report of dependent variables*

Early characteristics of team members, such as team member value of prevention, attitudes regarding adolescent substance use, and net benefits, had moderate relations with Wave 3 team functioning. For example, teams with members that placed a high value on prevention activities had less tension and conflict (r=-0.45), and teams with members that were more accepting of adolescent alcohol use were less focused on work (r=-0.62).

Both of the technical assistance measures also predicted Wave 3 team functioning. Effective collaboration with the PC had a moderate to strong positive association with team leadership (r=0.53) and team culture (r=0.48), indicating that greater team cooperation with university-related technical assistance corresponded with more skilled team leadership and a more positive interpersonal atmosphere at Wave 3. Contact with the PC had a moderate positive association with team tension (r=0.38), indicating that higher levels of PC contact were associated with more tension and conflict at Wave 3.

Partial Correlations Controlling for Poverty

Controlling for community poverty affected some, but not all, associations between predictors and team functioning. For example, the value of prevention remained significantly associated with Wave 3 team tension; team net benefits remained moderately to strongly associated with team focus on work, team culture, and team tension; and team collaboration with the PC remained moderately to strongly associated with team leadership and team culture. On the other hand, there were increases in the magnitude of associations of both community substance use norms and team member acceptance of alcohol with Wave 3 team functioning variables.

Multivariate Regression Models

Multivariate regressions were conducted to investigate the presence of collinearity or suppressor effects among predictors within a domain. With only one exception, the results of the multivariate regressions were highly consistent with the partial correlations. This consistency implies that associations of predictors within a domain with team functioning were fairly independent of each other. The one exception involved community poverty, readiness, and substance use norms (for brevity, only these regression models are reported here; see Table 5). Community readiness and community substance use norms have a suppressor effect on each other in predicting team functioning. Reciprocal suppression occurs when the two independent variables are negatively related to each other, but they are positively related to the dependent variables (Tabachnik and Fidell 2001). Thus, the effects of community readiness and substance use norms appear stronger when they are included in the same model—as in the regression models—rather than separately, as in the partial correlations.

Table 5
Domain regressions predicting team-member report of team focus on work, team culture, team leadership, and team tension with community readiness and community substance use norms

The PC as the “Rater” of Team Functioning

Three of the team functioning variables were independently rated by PCs: team focus on work, team culture, and team tension. PCs and team members demonstrated high agreement on their perceptions of team focus on work (r=0.74) and team culture (r=0.84); they evidenced little agreement in perceptions of team tension (r=0.10).

Zero-order and Partial Correlations between Predictors and Dependent Variables

Overall, the associations between the independent variables and the PC-reported dependent variables were quite similar to those found for team member reports of dependent variables. The results are presented in Table 6. With two exceptions (the associations between team member collaboration experience and team focus on work; and between team member acceptance of alcohol use and team tension), all bivariate and partial correlations were in the expected direction. A few similarities to team member findings are noteworthy. The collaboration constructs were consistent predictors of team functioning across both sets of analyses: community collaboration evidenced a moderate to strong association with team culture, and school collaboration evidenced strong negative associations with team tension. Team member acceptance of alcohol use, team net benefits, and collaboration with PC were also fairly consistent predictors across the two sets of analyses. Team member acceptance of alcohol use evidenced a consistently strong (negative) association with team focus on work. Collaboration with PC was consistently associated with focus on work and team culture, and contact with PC was consistently associated with team tension.

Table 6
Zero-order and partial correlations (controlling for poverty) between predictor variables and PC report of dependent variables

Two team member reported predictor variables were more strongly associated with PC report of team functioning than with team member report of team functioning. Community readiness more strongly predicted PC report of team tension, and community collaboration more strongly predicted PC report of team focus on work.

Multivariate Regression Analyses

As above, the domainlimited regression results were generally consistent with the partial correlation results, and results were similar to those for team members. For the community variables of community poverty, readiness, and substance use norms, the same pattern of suppressor effects as obtained above was found. However, in predicting the PC report of team functioning, the betas for these community-level predictors were smaller than above and not always significant6.

Discussion

The primary goal of this investigation was to understand the predictors of effective community team functioning over an 18-month period, across two phases of team development. In projects where a partnership model is implemented in multiple communities, it is fair to assume the presence of “individual” differences—that is, the teams in some communities fare better than others. For example, in studies showing no overall effects, some community teams may have produced a substantial effect while most did not. And in studies demonstrating significant effects, it is reasonable to expect that teams in some communities fell short. Understanding predictors of “individual” differences among community teams may be an important source of knowledge for refining and enhancing models of team success.The results from the longitudinal analyses reported here indicate that previously-measured aspects of communities, team members, and the teams themselves influence the functioning of community teams during the operations phase (assessed when teams were 18 months old). As we found in previous research when the same teams were in the organization stage (at 6 months into the project; Greenberg et al. 2007), the level of poverty in the community was associated with team cohesion and leadership quality, but not with team work focus or conflict. Specifically, communities with higher levels of poverty evidenced lower levels of team cohesion and leadership quality. These findings were consistent across both team member and Prevention Coordinator (PC) report.

Analyses that controlled for level of poverty in the community yielded results indicating that other measures of community context and social capital—such as community readiness, substance use norms, community and school collaboration—were moderately associated with indices of team functioning. In these analyses, higher levels of community protection and lower risk (i.e., greater community readiness and greater collaboration) were associated with better team functioning. These results suggest that even after controlling for the fairly strong predictor of community-level poverty, the social context and social capital of a community explains variance in the quality of partnership team functioning. The implication of this finding is that communities that have low level levels of resources, both economic and social, may require greater levels of support in successfully adopting collaborative prevention approaches.

Characteristics of individual team members (aggregated at the team level) were also associated with team functioning at this stage in the team’s lifecycle. At 18 months, teams whose members reported at pretest that they more highly valued prevention and were less accepting of adolescent alcohol use, tended to function better. However, team members’ prior collaboration experience and related skill-sets were not highly related to team functioning. These results indicate that team member attitudes regarding prevention and substance use, rather than prior experiences and skills, are more closely linked to quality of team functioning at the 18-month time point. Thus, a shared level of commitment to the partnership team mission seems to influence overall team functioning. We should note that these results do not imply that prior experiences and skills are unimportant; such team member characteristics may influence the quality of teams’ work products (e.g. planning documents, development of logic models, marketing materials, grant proposals), which were not examined in this report. Rather, prior team member experiences and skill-sets appear to be unrelated to the measures of team process examined in this report.

In addition, we found moderate to strong associations from team members’ perceived net benefit of participating in the project 1 year before to indices of team functioning in the operations phase. Teams whose individuals were more satisfied with the ratio of personal benefits to costs demonstrated better functioning over time. This may in part be a circular, reinforcing process, where team member satisfaction leads to increased motivation and engagement, leading to enhanced team functioning and success, leading to decreased costs of involvement (e.g. personal tensions) and increased benefits (e.g. sense of helping the community, networking).

One implication of these results concerning aspects of individual team members relates to the importance of training and technical assistance. Given that individual attitudes and, by inference, commitment to the initiative are important predictors of team functioning, it is important to recognize the value of providing training and ongoing support to team members (Feinberg et al. 2002). Such training should devote considerable effort to clearly describing the benefits and potential of both prevention, in general, as well as the specific program’s logic model in ways that participants can readily understand. Specific techniques, such as practice articulating the value of prevention and the program’s logic model through role-playing presentations to outside groups, could help team members more thoroughly understand and adopt these perspectives.

Finally, at the team level, collaboration and contact with the PC was moderately related to team functioning. Teams that cooperated with PCs through returning phone calls, being open to consultation, and encouraging contact, tended to function better over time. This result echoes our finding that technical assistance is an important predictor of the success of these community teams in recruiting families to the family-focused prevention program (Spoth et al. 2007), as well as prior research indicating the important role that technical assistance plays in supporting prevention programs and community collaborations (Feinberg et al. 2004b;O’Donnell et al. 2000).

In terms of understanding the differential impact of factors on team functioning over time, the results in this report should be considered in combination with the results from our prior report examining influences of team functioning during the organizational phase (Greenberg et al. 2007). During the organizational phase, team functioning was primarily influenced by the level of community poverty and team member prevention attitudes and skills. In this report, examining influences on team functioning 1 year later during the operations phase, poverty and team member prevention attitudes remained strong influences on team functioning. However, while the influence of team member skills waned, the influence of community social capital and team member attitudes regarding adolescent alcohol use appeared to become stronger. There are several possible reasons for these changes over time, and we offer two speculations that suggest hypotheses to be assessed further in future research.

First, as the responsibilities and task demands grew over time, the teams’ functioning during the operations phase, a period of relatively higher stress corresponding to increased demand, may demonstrate aspects of the teams that were not observable during an earlier period. Such functioning under stress reveals vulnerabilities and strengths that may not be apparent at other times. Thus, some teams may appear to be functioning cohesively during lower-demand periods but relatively more divided when confronted by stress. In contrast, other coalitions may pull together during times of stress in order to meet a particular challenge. As a result, differences in predictors of functioning may be related to actual differences in functioning either across time or across levels of task-demand.

Second, there may be an initial level of enthusiasm that, among some teams, declines over time as the burden of the tasks increases and interpersonal divisions emerge. The developmental processes that teams undergo may be influenced by a number of factors, such as the ones that we found to be influential in the operations phase but not at an earlier phase. The social capital of a community—the level of trust, collaboration, and social cohesion among residents and local organizations—may influence the way that a coalition handles challenges and remains cohesive over time. In addition, members’ commitment to the team’s work would likely influence the way that the team members jointly focus on the tasks at hand, relate to each other cordially and supportively, minimize individual turf-related goals, and so on. Thus, to the degree that members’ attitudes of the acceptability of youth alcohol use index members’ perception of and commitment to changing a community problem, such attitudes will be reflected over time in the ways that teams function—particularly under stress.

We have noted above the limitation that the sample size places on the analyses. In addition, the communities selected for this study were, by design, limited to rural and small-town communities. Thus, generalization to coalitions in more urban environments should be cautious.

In sum, this study has documented longitudinal influences on community prevention coalition functioning during the operations phase of the community team lifecycle. In addition, results here were generally consistent across different reporters of team functioning. We found both continuity and change in the set of individual, team, and community level influences compared to our prior research which examined team functioning during the team organizational phase. As described above, these results have implications for how community-based prevention projects are designed and supported. These types of findings could guide the design of partnership models that could facilitate better functioning of community teams. In future research, we will examine the degree to which the quality of community team functioning determines the long-term sustainability of such teams. We also plan to examine whether the quality of team functioning is linked to the community outcomes (e.g. youth substance use) that the coalitions aim to achieve through the implementation of evidence-based interventions.

Acknowledgments

Work on this paper was supported by research grant DA 013709 from the National Institute on Drug Abuse

Footnotes

1Although one could draw distinctions, we utilize the terms coalitions, partnerships, and local teams interchangeably in this manuscript. These terms in our usage refer to a range of informal through formal collaborative groups formed to enact positive change in a community.

2One community did not have any Agency Directors participate in the interview at wave 1.

3The Pennsylvania School District achievement data was collected from the Pennsylvania System of School Assessment District Report Cards (2003). The Iowa School District achievement data was collected directly from the participating school district offices.

4We estimated the AD community collaboration value with a multivariate OLS regression for the one community that did not have any Agency Directors participate in the interview.

5Technical reports for each measure by each respondent are available from the first author.

6Results for all regressions can be obtained from the first author.

References

  • Arthur MW, Hawkins DJ, Catalano RF, Olson JJ, Social Development Research Group . Diffusion project: Fall 1998 community key informant interview. University of Washington; Seattle: 1998.
  • Backer TE. Evaluating community collaborations. Vol. 158. Springer; New York, NY: 2003.
  • Bronfenbrenner U. Ecology of the family as a context for human development: Research perspectives. Developmental Psychology. 1986;22(6):723–742.
  • Butterfoss FD, Goodman RM, Wandersman A. Community coalitions for prevention and health promotion: Factors predicting satisfaction, participation, and planning. Health Education Quarterly. 1996;23(1):65–79. [PubMed]
  • Chavis DM. Building community capacity to prevent violence through coalitions and partnerships. Journal of Health Care for the Poor and Underserved. 1995;6(2):234–245. [PubMed]
  • Chavis D, Florin P, Felix MRJ. Nurturing grassroots initiatives for community development: The role of enabling systems. In: Mizrahi T, Morrison J, editors. Community organization and social administration: Advances, trends, and emerging principles. Haworth; Binghamton, NY: 1992.
  • Chilenski-Meyer SE, Greenberg MT, Feinberg M. Community readiness as a multi-dimensional construct. Journal of Community Psychology. 2007;35:351–369.
  • Chinman MJ, Imm P, Wandersman A. Getting to Outcomes 2004: Promoting accountability through methods and tools for planning, implementation, and evaluation. RAND Corporation Technical Report; Santa Monica, CA: 2004.
  • COMMIT Community intervention trial for smoking cessation (COMMIT): I. Cohort results from a four-year community intervention. American Journal of Public Health. 1995;85(2):183–192. [PubMed]
  • Emshoff J, Blakely C, Gray D, Jakes S, Brounstein P, Coulter J, et al. An ESID case study at the federal level. American Journal of Community Psychology. 2003;32:345–357. [PubMed]
  • Feinberg ME, Greenberg MT, Osgood DW. Readiness, functioning, and perceived effectiveness of community prevention coalitions: A study of Communities That Care. American Journal of Community Psychology. 2004a;33:163–176. [PubMed]
  • Feinberg ME, Greenberg MT, Osgood DW. Technical assistance in prevention programs: Correlates of perceived need in Communities That Care. Evaluation and Program Planning. 2004b;27:263–274.
  • Feinberg ME, Greenberg MT, Osgood DW, Anderson A, Babinski L. The effects of training community leaders in prevention science: Communities That Care in Pennsylvania. Evaluation and Program Planning. 2002;25:245–259.
  • Feinberg ME, Greenberg MT, Osgood WO, Sartorius JS. Can community coalitions have a population level impact on adolescent behavior problems? CTC in Pennsylvania; 2005.
  • Florin P, Mitchell R, Stevenson J. Identifying training and technical assistance needs in community coalitions: A developmental approach. Health Education Research. 1993;8(3):417–432. [PubMed]
  • Florin P, Mitchell R, Stevenson J, Klein I. Predicting intermediate outcomes for prevention coalitions: A developmental perspective. Evaluation and Program Planning. 2000;23:341–346.
  • Foster-Fishman PG, Berkowitz SL, Lounsbury DW, Jacobson S, Allen NA. Building collaborative capacity in community coalitions: A review and integrative framework. American Journal of Community Psychology. 2001;29(2):241–261. [PubMed]
  • Gottlieb NH, Brink SG, Gingiss PL. Correlates of coalition effectiveness: The Smoke Free Class of 2000 Program. Health Education Research. 1993;8(3):375–384. [PubMed]
  • Green LW, Kreuter MW. Fighting back or fighting themselves? Community coalitions against substance abuse and their use of best practices. American Journal of Preventive Medicine. 2002;23(4):303–306. [PubMed]
  • Greenberg MT, Chilenski-Meyer S, Feinberg ME. Community and team member factors that influence the early phase functioning of community prevention teams: The PROSPER project. Journal of Primary Prevention. 2007 in press. [PMC free article] [PubMed]
  • Hallfors D, Hyunsan C, Livert D, Kadushin C. Fighting back against substance abuse: Are community coalitions winning? American Journal of Preventive Medicine. 2002;23:237–245. [PubMed]
  • Hawkins JD, Catalano RF, Arthur MW. Promoting science-based prevention in communities. Addictive Behaviors. Special Issue: Integration substance abuse treatment and prevention in the community. 2002;27(6):951–976. [PubMed]
  • Herrenkohl TI, Maguin E, Hill KG, Hawkins JD, Abbott RD, Catalano RF. Developmental risk factors for youth violence. Journal of Adolescent Health. 2000;26(3):176–186. [PubMed]
  • Jausa GK, Chou C, Bernstein K, Wang E, McClure M, Pentz MA. Using structural characteristics of community coalitions to predict progress in adopting evidence based programs. Evaluation and Program Planning. 2005;28:173–184.
  • Kegler MC, Steckler A, McLeroy K, Malek SH. Factors that contribute to effective community health promotion coalitions: A study of 10 Project ASSIST Coalitions in North Carolina. Health Education and Behavior. 1998;25(3):338–353. [PubMed]
  • Kerr NL, Tindale RS. Group performance and decision making. Annual Review of Psychology. 2004;55:623–655. [PubMed]
  • Klerman LV, Santelli JS, Klein JD. So what have we learned? The editors’ comments on the coalition approach to teen pregnancy. Journal of Adolescent Health. 2005;37:S115–S118. [PubMed]
  • Kreuter MW, Lezin NA, Young LA. Evaluating community-based collaborative mechanisms: Implications for practitioners. Health Promotion Practice. 2000;1(1):49–63.
  • Livit M, Wandersman A. Organizational functioning: Facilitating effective interventions and increasing the odds of program success. In: Wandersman D. M. F. a. A., editor. Empowerment evaluation principals in practice. Guilford; New York: 2004. pp. 123–142.
  • Moos R. Work environment scale. Consulting Psychologists Press; Palo Alto, CA: 1981.
  • O’Donnell L, Scattergood P, Alder M, San Doval A. The role of technical assistance in the replication of effective HIV interventions. AIDS Education and Prevention. 2000;12(Suppl A):99–111. [PubMed]
  • Osgood DW, Chambers JM. Social disorganization outside the metropolis: An analysis of rural youth violence. Criminology. 2000;38(1):81–115.
  • Redmond C, Spoth RL, Shin C, Schainker L, Greenberg M, Feinberg M. Proximal outcomes of an evidence-based universal family-focused intervention implemented by community teams. 2007 Under review.
  • Reisig MD, Cancino JM. Incivilities in nonmetropolitan communities: The effects of structural constraints, social conditions, and crime. Journal of Criminal Justice. 2004;32(1):15–29.
  • Roussos ST, Fawcett SB. A review of collaborative partnerships as a strategy for improving community health. Annual Review of Public Health. 2000a;21:369–402. [PubMed]
  • Roussos ST, Fawcett SB. A review of collaborative partnerships as a strategy for improving community health. American Review of Public Health. 2000b;21:369–402. [PubMed]
  • Saxe L, Reber E, Hallfors D, Kadushin C, Jones D, Rindskopf D, et al. Think globally, act locally: Assessing the impact of community-based substance abuse prevention. Evaluation and Program Planning. 1997;20(3):357–366.
  • Spoth R, Greenberg M, Bierman K, Redmond C. PROSPER community-university partnership model for public education systems: Capacity-building for evidence-based, competence-building prevention. Prevention Science. 2004a;5(1):31–39. [PubMed]
  • Spoth RL, Greenberg M, Bierman K, Redmond C. PROSPER community-university partnership model for public education systems: Capacity-building for evidence-based, competence-building prevention. Prevention Science. 2004b;5(1):31–39. [PubMed]
  • Spoth RL, Redmond C, Shin C, Greenberg MT, Clair S, Feinberg ME. Toward public health benefits from community-university partnerships: PROSPER effectiveness trial results for substance use at 1 1/2 years past baseline. American Journal of Preventive Medicine. 2007 in press.
  • Stevenson J, Mitchell R. Community level collaboration for substance abuse prevention. American Journal of Community Psychology. 2003;23:371–404.
  • Tabachnik BG, Fidell LS. Using multivariate statistics. Allyn and Bacon; Boston: 2001.
  • Wandersman A, Florin P. Community interventions and effective prevention. American Psychologist. Special Issue: Prevention that works for children and youth. 2003;58(67):441–448.
  • Wandersman A, Florin P, Friedmann R, Meier R. Who participates, who does not, and why? An analysis of voluntary neighborhood organizations in the United States and Israel. Sociological Forum. 1987;2(3):534–555.
  • Wandersman A, Valois RF, Ochs L, de la Cruz DS, Adkins E, Goodman RM. Toward a social ecology of community coalitions. American Journal of Health Promotion. 1996;10(4):299–307. [PubMed]