A multiple case study of implementation ijn 10 loca


Published on

  • Be the first to comment

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide

A multiple case study of implementation ijn 10 loca

  1. 1. HEALTH EDUCATION RESEARCH Vol.13 no.2 1998Theory & Practice Pages 225–238 A multiple case study of implementation in 10 local Project ASSIST coalitions in North Carolina Michelle Crozier Kegler, Allan Steckler1, Sally Herndon Malek2 and Kenneth McLeroy Abstract during the planning phase and lack of member input into the action plan. ConflictCommunity health promotion relies heavily on contributed to staff turnover, reluctance tocoalitions to address a multitude of public conduct certain activities and difficulty inhealth issues. In spite of their widespread use, recruiting members, all of which had implica-there have been very few studies of coalitions tions for implementation.at various stages of coalition development. Thepurpose of this study was to identify factors that Introductionfacilitated or impeded coalition effectiveness inthe implementation stage of coalition develop- In recent years, communities have formed coali-ment. The research design was a multiple case tions to address a multitude of public health andstudy with cross-case comparisons. Each of social problems (Herman et al., 1993; Kumpferthe 10 local North Carolina Project ASSIST et al., 1993; Rogers et al., 1993; CDC, 1995;coalitions constituted a case. Data collection Butterfoss et al., 1996; Fawcett et al., 1996).included: semi-structured interviews, observa- Indeed, community health promotion as it istion, document review, and surveys of members currently practiced relies very heavily on coalitions.and staff. Some of the major factors that In some communities, coalitions—defined asfacilitated implementation included: the ability organizations and individuals working togetherof the coalition to provide its own vision, staff to achieve a common goal (Feighery and Rogers,with the skills and time to work with the 1990, Brown, 1984)—exist for almost everycoalition, frequent and productive communica- disease, risk factor and social problem. This growthtion, cohesion or a sense of belonging on the in the popularity of coalitions is due to mandatescoalition, and complexity of the coalition struc- by funding agencies, in combination with theture during the intervention phase. Barriers to belief that by bringing multiple community sectorseffective implementation included: staff turn- together to share resources and combine energy,over and staff lacking community organization coalitions can demonstrate widespread support forskills, dependence on the state-level staff action and provide a vehicle for solving problems that are too complex for single agency solutions (Black 1983; Brown, 1984; Feighery and Rogers,Department of Health Promotion Sciences, College of 1990; Butterfoss et al., 1993). Further, coalitionsPublic Health, University of Oklahoma Health Sciences are believed to foster community ownership which,Center, Oklahoma City, OK 73190, 1Department of in turn, is thought to increase the likelihood ofHealth Behavior and Health Education, School of Public long-term changes in physical and social environ-Health, University of North Carolina, Chapel Hill,NC 27599, USA and 2Division of Adult Health Promotion, ments that support health-related and health-dir-North Carolina Department of Environment, Health, and ected behavior (Winett et al., 1989; Bracht, 1990;Natural Resources, Raleigh, NC 27611, USA Thompson and Kinne, 1990).© Oxford University Press 225
  2. 2. M. C. Kegler et al. In spite of their popularity and widespread of membership outweighing costs are related touse, until recently, there have been relatively few satisfaction and participation among coalitionsystematic studies of coalitions (Altman et al., members.1991; Mizrahi and Rosenthal, 1992; Butterfoss Given that coalitions evolve over time, it is likelyet al., 1993). A better understanding of coalitions that different factors are important to coalitionwould help us assess under what conditions a functioning at various stages of development. Verycoalition approach is best suited and how to little research has been conducted on the factorsincrease the likelihood that a coalition’s efforts that facilitate or inhibit coalition effectiveness inwill have a positive impact, both on the public’s the middle or later stages of coalition development.health, and on the capacity of a community to Gottlieb et al. (1993) reported on correlates ofrespond to future social and health concerns. implementation among state-level coalitions, butDeveloping an understanding of coalitions is com- these were not community-based coalitions andplicated, however, by the many different types membership averaged only five organizations.of coalitions and the various stages of coalition Fawcett et al. (1996) have developed a monitoringdevelopment. system for community coalitions in which the The literature discusses two ways to conceptu- number of volunteers recruited, planning productsalize stages of coalition development. The first produced, resource generated, community actionsfocuses on relationships among partners. Alter and and community changes are tracked. Although theHage (1993) describe networks as beginning as system provides valuable feedback to the parti-loosely linked organizations held together by cipating communities, its primary use has notindividuals who serve as boundary spanners by been to identify factors that influence coalitionexchanging resources and coordinating tasks effectiveness in various stages of coalitionacross organizational boundaries. As the relation- development.ships between partners evolve, more long-term, The implementation stage of coalition develop-formal relationships are formed to jointly produce ment, defined as the carrying out of planneda product, program or service. A second approach activities, is a critical link between organizingto conceptualizing stages of coalition development a community coalition, engaging in a planningis to ground the stages in the community-wide process and changing behavior, social andhealth promotion process (Butterfoss et al., 1993; physical environments and health status. WithoutFlorin et al., 1993). Florin et al. (1993), for implementation of activities that logically lead toexample, describe seven stages of coalition the outcome of interest, the impact of coalitionsdevelopment: initial mobilization, establishing on the public’s health will be minimal. In spite oforganizational structure, building capacity for its critical nature, the implementation stage ofaction, planning for action, implementation, coalition development has not received muchrefinement and institutionalization. attention in the literature. Recent studies in community health promotion The purpose of the present study was to usehave focused on the formation and early mainten- a multiple case study design to identify factorsance of coalition development, using member that influenced coalition effectiveness in thesatisfaction, participation, commitment and quality implementation stage of coalition developmentof the action plan produced by the coalition as in 10 community-based tobacco control coalitions.intermediate measures of coalition effectiveness The study was guided by a conceptual model that(Kumpfer et al., 1993; Rogers et al., 1993; posited relationships between various coalitionButterfoss et al., 1996). These studies have shown factors and implementation. The factors werethat competent leadership, shared decision making, categorized into operational processes and struc-linkages with other organizations, a supportive tural characteristics of coalitions (Table I). Opera-group environment, communication and benefits tional processes, defined as factors related to the226
  3. 3. Project ASSIST coalitions in North CarolinaTable I. Study factors and case study questionsStudy factor Case study questionsLeadership How are leaders selected? Which community sectors are leaders from? What tasks do chairs perform? When in the process are chairs selected? How skilled is the leadership in working with the coalition? How does leadership influence implementation?Decision-making How much influence do various players have in developing the action plan? in selecting chairs? in budget decisions? in setting coalition policy? How is member influence in decision making related to implementation?Communication How does the coalition keep its members and leaders informed? What are the mechanisms for communication? What is the quality of communication among members? Between staff and members? How is the quality of communication related to implementation?Conflict What types of conflict arise on the coalition? How is the conflict resolved? How does conflict influence implementation?Benefits and costs What are the organizational costs and benefits of participation? How are the organizational costs and benefits of participation related to implementation?Organizational climate What are the members’ perceptions of cohesiveness and task focus of the coalition? What are the members’ perceptions of cohesiveness and task focus of the coalition?Staff roles How is work distributed between the coalition members, leaders and staff? How skilled are staff in supporting the coalition? How does staff role influence implementation?Capacity building What capacity-building activities are conducted? Which activities are most helpful? What linkages exist between the community sectors? Which are formed through the coalition? How does capacity building affect implementation?Member profile Describe the composition of the coalition membership, including expertise, experience, demographic characteristics and community sectors. How does the composition of the coalition membership influence implementation?Recruitment pattern Describe the recruitment pattern. Was there a core group? Which sectors were involved first? Were there recruitment cycles? How does the recruitment pattern influence implementation?Organizational structure What is the organizational structure of the coalition? How did it evolve? How formal is the coalition? How does level of formalization affect implementation?Community capacity Had the community addressed tobacco control prior to this project? Had the lead agency addressed tobacco before? Did the coalition build on a pre-existing network? How does the capacity of the community to address tobacco influence implementation?ongoing operational functioning of the coalition, Table I presents the factors studied and theincluded leadership, decision making, communica- accompanying case study questions.tion, conflict, costs and benefits of participation,organizational climate, staff role, and capacity Methodsbuilding. Structural characteristics were definedas the relatively stable, descriptive characteristics Description of the coalitionsof the coalition, and included member profile, All 10 of the community-based tobacco controlrecruitment pattern, complexity of the structure formed as part of the American Stop Smokingand level of formalization. Community capacity Intervention Study (Project ASSIST) in Northto engage in tobacco control was also examined. Carolina (Shopland, 1993; Malek and Enright,The overall research questions were: (1) what 1995) participated in this study. Project ASSIST isfactors facilitate coalition effectiveness in imple- a 7 year, two phase, collaborative effort between thementation and (2) what factors inhibit coalition National Cancer Institute (NCI) and the Americaneffectiveness in implementation? A series of case Cancer Society (ACS) designed to reduce thestudy questions were developed for each factor prevalence of smoking in participating states byand its influence on implementation of activities. 1998 (NCI, 1991). At the state level, the state 227
  4. 4. M. C. Kegler et al.Table II. Site-ordered matrix of key descriptive characteristics of coalitionsCoalitionsa by No. of sectors No. of sectors No. of Urban or Single or No. ofimplementation involved during involved during members during rural multi-county activitieslevel (high to low) planning phase intervention phase intervention phase coalition implementedA 7 7 76 urban single 12J 4 7 47 rural single 10B 3 4 32 rural multi 8D 4 5 63 urban single 7G 5 7 44 urban single 6I 4 5 53 rural multi 5F 3 3 13 rural single 5H 2 3 26 rural multi 4E 4 5 36 rural multi 3C 4 4 40 rural single 2aPrimary case coalitions are in bold.health department and the state division of the (51% of members). The most common coalitionACS served as lead agencies in forming tobacco structure was an advisory board, a coordinatingcontrol coalitions and designing and implementing committee, and task forces for worksites, schools,statewide tobacco control initiatives. North health care settings, community groups and publicCarolina formed a statewide coalition, and recruited education.and funded 10 local ASSIST sites through an Implementation, operationalized as the numberapplication process. These local sites were required of activities completed by the coalition, is alsoto form local coalitions. County health departments listed in Table II. Table III contains brief descrip-served as lead agencies, in partnership with the tions of selected activities implemented by thelocal ACS chapters. This study focused on the local Project ASSIST coalitions in the first year of2 year planning phase and the first year of interven- intervention. The state level NC ASSIST stafftion (October 1991–September 1994). The plan- and coalition members provided strong technicalning phase (October 1991–October 1993) included support for several of these activities, includingcoalition development, site analysis, development the youth tobacco buying operations, the non-of a 5 year strategic plan and development of an smoking worksite policy manual and the trainingaction plan for the first year of the intervention events.phase. Phase II, the 5 year intervention phase, Although the study involved all 10 of the localbegan in October of 1993. NC ASSIST coalitions, five primary cases were Table II lists each of the coalitions and some selected for more in-depth data collection andaccompanying descriptive information, including analysis. The primary cases were selected tomembership size, number of community sectors capture variation in the role of the staff andinvolved (out of a total of 11 sectors such as structure of the coalitions. Other considerationsmedia, business, health, recreation, etc.), single were region of the state, whether the coalitionversus multiple county coverage and population covered a single or multiple counties and whethersize. Four of the coalitions operated in more than it was primarily rural or urban (see Table II).one county and six served a single county. Three Individual case studies were written for the primaryof the coalitions were considered urban, with case coalitions (Kegler, 1995). This article reportspopulations greater than 300 000. The health sector key findings from the cross-case comparisonswas the best represented community sector by far across all 10 coalitions.228
  5. 5. Project ASSIST coalitions in North CarolinaTable III. Description of NC Project ASSIST local coalition activitiesActivity DescriptionYouth tobacco buying operation Sent youth to area stores to purchase cigarettes in an effort to demonstrate youth access to cigarettes; followed up with a press conference.Merchant education campaign Distributed ‘Do not sell to minors’ materials to area merchants in an effort to increase their awareness of and commitment to complying with youth access regulations.Commit to quit Invited smokers to quit smoking for 1 month in return for eligibilty for a major prize such as a $1 000 shopping spree. Gave away prize at a big party planned for one month after the start date. Usually heavily promoted through the media.Health professional training Trained health professionals to counsel smokers to quit using NCI’s 4 A’s approach.Youth and Elders Teleconference on smoking and youth Organized satellite hookups for a national press conference on the release of the Surgeon General’s Report on tobacco and youth.Non-smoking worksite policy manual Assessed smoking policies of area worksites and developed a manual with model worksite non-smoking policies using worksites from the ASSIST communities as the models.Community smoking cessation resource guide Compiled information on area smoking cessation services, published and distributed a resource guide.Public service announcements Distributed public service announcements on environmental tobacco smoke to local media outlets.Smoking control ordinances Provided information and support to local governments attempting to pass smoking control ordinances.Data collection ended questions were developed for each majorThe study used multiple methods of data collection: category of key informant.semi-structured personal interviews with key Observationinformants, observation of coalition meetings,document review, and surveys of the coalition All 10 of the coalitions were observed once in themembership and staff. The research design and all planning phase (Spring 1993) and the primarydata collection instruments were reviewed and case coalitions were observed again during theapproved by the University of North Carolina’s intervention year (Spring 1994). An observationInstitutional Review Board. guide was developed for both rounds of observa- tions. In addition to the site visits, over 15 stateInterviews coalition meetings, retreats and training eventsFifty-two semi-structured interviews were con- were observed.ducted with key informants during the planningand intervention phases of ASSIST. During the Documentsplanning phase, interviews were conducted with The documents collected for each case included10 local coordinators and three NC ASSIST staff. the following: original grant application to theDuring the intervention phase, interviews were state health department, membership rosters duringconducted with 12 local coordinators (including the planning phase and again for the interventionfour co-coordinators), five health directors, 15 phase, bylaws, meeting agendas, minutes, sitecoalition and/or task force chairs, and seven NC analysis reports and the year 1 action plans. TheseASSIST staff. Interview guides containing open- documents were used in writing case descriptions 229
  6. 6. M. C. Kegler et al.and developing some of the measures. In addition, All of the case materials were coded using openNC ASSIST documents were collected. These and axial coding as discussed by Strauss andincluded quarterly reports to NCI, state coalition Corbin (1990). Open coding refers to the initialmeeting minutes, agendas and handouts, and train- labeling and categorizing of the data. Axial codinging materials. refers to making connections between categories identified in the open coding process. The factorsMember and staff surveys specified in the conceptual model served as initialA survey of all of the local coalition members sensitizing concepts for the open coding. More(N 430) was conducted during the first year of specific coding categories emerged during theintervention. The survey instrument was a self- actual coding of the data. The interview transcriptsadministered questionnaire adapted from those and field notes were coded and organized usingused by Butterfoss et al. (1996) and Rogers et al. Ethnograph (Seidel et al., 1988), a software pack-(1993). The Member Survey addressed each of the age designed for analyzing qualitative data. Thiscoalition factors specified in the conceptual model helped to facilitate cross-case analysis by providingand is described in detail elsewhere (Kegler et al., a storage and retrieval system for the large amount1998). A survey of the local Project ASSIST of data.staff was also conducted during the first year The case study data were analyzed one case atof intervention. It contained many of the same a time for the primary case coalitions. Data werequestions with slight wording changes to reflect a displayed in matrices as described by Miles andstaff rather than member perspective Huberman (1994). Initial displays were developed for each study factor with dimensions of the factorImplementation measure identified through the coding on one axis and dataThe main outcome of interest for this study was source on the other axis. The cells were completedthe implementation of activities in the first year of with the actual data. These matrices were used tointervention. The number of activities that were answer the case study questions for each of thecompleted during the first year of intervention, primary case coalitions. Once the primary casesregardless of the number of activities the coalition were analyzed and tentative statements thathoped to accomplish, was the primary outcome accounted for the patterns in the primary case datameasure used in the cross-case comparisons. This were formulated, the site-ordered matrices wererelatively crude measure of implementation was expanded to include data from the secondary cases.used over a more complex measure of quality of The statements were then revised to explain theimplementation due to ease of data collection, patterns in all 10 of the cases. These patternssimilarity of activities across coalitions and simpli- and relationships are reported in the followingfication of the cross-case comparisons. results section.Case study analysis ResultsA case study protocol was followed for two roundsof site visits (Yin, 1989). In addition, a case study This section presents the case study findings on thedatabase was maintained for each coalition. This factors that influence health promotion coalitions ininvolved maintaining the data in a manner such implementation. A variety of factors previouslythat other investigators could review the evidence; shown to be important in earlier stages of coalitionthus increasing the reliability of the case study development were examined.(Yin, 1993; Patton, 1990). The coalition databasesincluded relevant documents, field notes, interview Factors found to facilitate implementationtranscripts and summaries of the survey data for Leadership was defined as the skills of the coali-each coalition. tion leaders in guiding the coalition toward the230
  7. 7. Project ASSIST coalitions in North Carolinaaccomplishment of its goals, including running completing the site analysis required by the NCImeetings, articulating a vision for the coalition, and state-level NC ASSIST, and continuing withand nurturing commitment to the coalition from their other job responsibilities. Lack of adequatemembers, their organizations and the community. staff time during the planning phase, combinedOne of the dimensions of leadership that emerged with fairly inflexible deadlines for variousas important to implementation was the source of deliverables such as the site analysis report andvision for the coalition. The 10 Project ASSIST the annual action plan, contributed to a tendencycoalitions varied in terms of who actually provided among some of the local coordinators to completedirection and vision. Vision was provided by either these activities without much input from the coali-the local coordinator, the state-level staff assigned tion members.to consult with the local coalition or it was shared Beginning in the first intervention year, fundsbetween the coalition chairs and the local coordina- were used to hire paid coalition staff. Over half oftors. The data strongly suggested a relationship the coalitions hired coordinators who had beenbetween source of vision and implementation. involved with ASSIST in some capacity duringThose coalitions for which the vision was held at the planning phase. The others hired new staff,the local level had higher levels of implementation with the planning phase coordinators becomingthan coalitions that were lacking a local vision. less involved and playing a supervisory role. StaffThe staff-dominated coalitions, in which the chairs time committed to the coalitions during the(if they existed) and coalition members did not implementation phase ranged from 20 h in theseem to share a vision for the coalition, generally coalition with the lowest level of implementationhad medium levels of implementation. Coalitions to 75 h in one of the coalitions with a high levelthat were dependent on state-level NC ASSIST of implementation.staff for direction and leadership had the lowest The way staff perceived their role with thelevels of implementation. coalition also influenced the level of implementa- Staffing was defined as the ability of the paid tion. As illustrated in Table IV, the coordinatorsstaff to guide and support the coalition, including spoke about their roles with the coalitions differ-the ability to shift responsibility from themselves ently and five types of roles emerged. Table Vto the coalition as the coalition evolved. The case examines the relationships between staff charac-study data suggested several dimensions of staffing teristics—including staff role—and implementa-that may facilitate coalition effectiveness in imple- tion. With one exception, the coalitions where thementation, including the amount of time staff can coordinator saw herself as primarily responsibledevote to a coalition and the role the staff play for carrying out the coalition’s activities (the Doers)with a coalition. The case study data, supported had the lowest levels of implementation.by the survey results reported elsewhere (Kegler Communication was defined as the frequencyet al., 1998), suggested that coalitions with more and productivity of communication between thestaff time devoted to them had higher levels of coalition staff and coalition members, and theimplementation. During the planning phase of members themselves. Coalitions with high levelsProject ASSIST, for example, the coalitions were of implementation had frequent and productivejust one responsibility among many for the local communication among staff and members. Bothcoordinators, most of whom were already the quantitative and qualitative data supported theemployed by the county health department. The importance of communication in implementation.local coordinators’ time was donated and the actual The case study data showed that communicationtime spent working with the coalitions during the between staff and coalition chairpersons on someplanning phase ranged from 10 to 30 h per week. of the coalitions was relatively infrequent and thisMany of the local coordinators felt stretched thin appeared detrimental to implementation. In one ofbetween the responsibility of forming a coalition, the coalitions, the chair and the coordinator had 231
  8. 8. M. C. Kegler et al.Table IV. Five staff roles with coalitionsThe Coach For me, I think the role of the health department is coaching the community along...How do you make good leaders out of people who don’t take on public health projects everyday? And, how can we help them do that?The Director We don’t want to recruit people to just sit on task forces. We want to recruit people to do specific things, which is again, very non-profit organization....‘Here is what you are going to do, you are going to be a part of this project and when this project is over [waves goodbye].’Linking Agent I have been seeing my job as the liaison between the state and local people in each county, you know, to get information to them and [to get] them to act on it within their county.The Doer Well, basically...all the different reports we’ve had to do, I’ve done them sort of singlehandedly.The Coordinator Doing the preparation, organizing meetings, attending meetings, getting minutes out for the meetings...being staff to each of those groups [the task forces].Table V. Site-ordered matrix of state and local staff characteristics by level of implementationCoalitionsa by Local staff time spent Local staff skill Were local intervention Local staff role in State staff role inimplementation on ASSIST per week in in intervention staff involved planning phase planning phaselevel (high to low) intervention phase (h) phase in planning phase?A 52 high yes coach resourceJ 75 medium yes doer resourceB 50 medium yes linking agent resourceD 60 medium yes director resourceG 46 insufficient data yes coordinator hands-onI 30 insufficient data yes linking agent resourceF 35 low no linking agent hands-onH 43 insufficient data no doer hands-onE 51 low no doer hands-onC 20 low no doer hands-onaPrimary case coalitions are in bold.not spoken to one another in several months. In members in spite of meeting for over a year. Inanother coalition, at least one monthly meeting another of the coalitions with low implementationwas canceled because the chair and the coordinator levels, a core group of action team chairs waswere not able to reach each other to set a coalition quite cohesive, but the larger coalition was frag-meeting date and time. These difficulties in com- mented and lacking in cohesion. The coalitionsmunication were in direct contrast to communica- with higher levels of implementation appeared totion in the coalitions with the highest levels of have a ‘Project ASSIST identity’ and high cohe-implementation where the chairs and the co- sion, particularly within task forces. One of theordinators spoke frequently. strong themes that emerged across the cases was Two dimensions of organizational climate were the necessity of having a tangible activity for taskexamined: cohesiveness and task focus. The forces to plan and conduct. This was expressedqualitative data supported the quantitative finding frequently by the coordinators as a reason whythat cohesiveness was related to effectiveness in some of the coalition’s task forces were doingimplementation (Kegler et al., 1998). In one of the better than others.coalitions with low levels of implementation, some Complexity of coalition structure, defined asmembers reported not knowing many of the other the number of functioning committees and task232
  9. 9. Project ASSIST coalitions in North Carolinaforces, was correlated with implementation in the tion. Staff skill was assessed both qualitatively andintervention phase (Kegler et al., 1998), but the quantitatively. Some illustrations of low staff skillcase study data suggested that complexity in the included lack of follow-up on a major recruitmentplanning phase may not be as important. During event, not having a list of the coalition membership,the planning phase, the majority of the coalitions not following through on coalition suggestionsappointed ‘channel’ or task force chairs to aid in and not sharing leadership roles with coalitionplanning, with the task forces remaining small or members. Those coalitions with more skilled staffnon-existent until the intervention phase. The urban were also those coalitions with higher levels ofcoalitions were exceptions, with two forming func- implementation.tioning task forces relatively early in the planning Staff turnover also slowed the implementationphase. Forming a complex organizational structure of planned activities. The coalitions that hiredin the planning phase did not appear to be essential intervention phase staff who were new to Projectto implementation, as demonstrated by one of the ASSIST and not involved in the planning processcoalitions with a high level of implementation. had lower levels of implementation. This is due,This coalition designated ‘channel chairs’ during in part, to the time required for staff to get orientedthe planning phase, but kept meeting in one large to the project and coalition. Periods where theregroup until a few months into the intervention year. were no staff also contributed to a slowing of One of the components of capacity building implementation.explored in the case studies was technical assist- The role of the state-level NC ASSIST staffance and training. Capacity building was defined appeared to have influenced implementation. Theas the knowledge and skills transferred to coalition state-level staff assigned to consult with the localmembers and their organizations through technical coalitions viewed their roles differently during theassistance and training, and the interorganizational planning phase. One of the roles was verylinkages created through the coalition. The state ‘hands-on’ with the state-level staff bringinghealth department played a major role in translating agendas, sitting at the head of the table andnational research findings into practice at the local playing a visible leadership role. Another rolelevel. Correspondingly, activities supported by was that of being a resource person to the localstate-level NC ASSIST staff were more likely to be coalitions with fairly low visibility at coalitionimplemented. These activities included the youth meetings.tobacco buying operation and merchant education The coalitions where the state-level NC ASSISTcampaign, health care training, Youth and Elders staff were ‘hands-on’ during the planning phaseTeleconference, Commit to Quit training, media had the lowest levels of implementation (Table V).spokesperson training, and an assessment of There are two possible explanations for this. Oneworksite policies. may be that these local coalitions were weaker and To summarize, factors that facilitated imple- needed a great deal of help. Another explanationmentation included a local source of vision, staff may be that the ‘hands-on’ help fostered depend-with the skills and time to work with the coalition, ency and when the help was withdrawn, the coali-frequent and productive communication, cohesion tions floundered. One of the coalitions with a lowor a sense of belonging to the coalition or one of level of implementation, for example, was veryits task forces, complexity of the coalition structure dependent on its state-level staff during the plan-during the intervention phase, and strong technical ning phase, but as the state-level staff becamesupport from state-level NC ASSIST staff. busier with state-level functions, this coalition was forced to rely on local leadership and staff. ThisFactors found to inhibit implementation pattern occurred in several of the coalitions—theCoalitions with staff lacking community state-level staff was very involved with the localorganization skills had lower levels of implementa- coalitions during the planning phase, providing a 233
  10. 10. M. C. Kegler et al.fair amount of ‘hands-on’assistance. As the state- newly hired staffperson to quit after just a couplelevel staff role changed in the intervention year of months. The coalition essentially stood stilland less time was spent with the coalitions and for several months while a new coordinator wasnewly hired coordinators, these coalitions appeared hired.to flounder. Another source of conflict stemmed from per- Coalitions with minimal member input into sonal ties to tobacco farming. This source wasplanning had lower levels of implementation. All only evident in one of the coalitions and severalof the coalitions had at least minimal input into coalitions had clauses in their bylaws forbiddingthe plan. For example, in two of the coalitions the membership by someone with a tobacco interest.coordinators wrote the plan and presented it to the One of the coalitions, however, had members withcoalition which then had an opportunity to review personal ties to tobacco, and this contributed toand revise the plan. Another spent several meetings lack of coalition member recruitment and resistancediscussing potential activities in a general way, but to carrying out some of the coalition’s plannedthe actual plan was written by the staff. One of activities.the local coordinators from this coalition explained Conflict from economic ties to the tobaccothat the plan was based on what members had industry influenced implementation as well. Duringdiscussed and what ‘needed to be in it’. Two of Project ASSIST’s second year, the North Carolinathe urban coalitions had functioning task forces at General Assembly passed a ‘smokers rights’ billthe time of plan development, and these task forces that gave local governments a deadline after whichgenerated ideas and developed their own action they could not pass any smoking control rules.plans, although the coalition with the highest level Several coalitions then became involved in educa-of implementation went the furthest with member tional efforts to increase community support toinvolvement by having the task forces actually pass local smoking control ordinances. Althoughcomplete the forms provided by state-level NC for the most part the emotion and hostilityASSIST. The data suggested that the coalitions generated by this debate was not directed atwith minimal member input into the action plan Project ASSIST, the political environment inwere the coalitions with lower levels of imple- which the coalitions operated became quite turbu-mentation. Surprisingly, coalitions that recruited lent in some of the counties. This atmospherenew members in the intervention phase did nothave lower levels of implementation in spite of seemed to energize and unite one of the coalitions,the fact that these new members were not involved although in the long run the board of health wasin developing the action plan. The case studies sued, board members were replaced with tobaccosuggested that it is important to involve current interests, the health director resigned and theremembers in planning in meaningful ways, but it is was talk of a bill that would abolish many ofpossible to recruit new members after the planning the boards of health across the state. Thisphase and still develop their ownership through atmosphere contributed to the caution andthe planning of specific activities. tentativeness of more than one health director Conflict was defined as friction on the coalition concerning tobacco control activities. This clearlycaused by differences of opinion, personality affected the types of activities the ASSISTclashes, hidden agendas, power struggles or other coalitions engaged in.sources of tension. Several types of conflict influ- Thus conflict, or a desire to avoid conflict,enced implementation. One source of conflict influenced implementation through staff turnover,stemmed from poor interpersonal relationships. In reluctance to recruit members to the coalition, andone of the multi-county coalitions, the local co- a general apprehension of taking a stand andcoordinators developed an antagonistic working conducting activities that could be perceived asrelationship which was serious enough to cause a controversial.234
  11. 11. Project ASSIST coalitions in North Carolina Discussion tors who had been involved during the planning phase turned over responsibility for the coalitionPrevious research on community health promotion to newly hired intervention staff. These new staffcoalitions has tended to focus on the early stages of came on board at a critical time without havingcoalition development, used member satisfaction, benefited from the planning process and, in someparticipation and quality of the action plan as cases, with minimal understanding of communityoutcomes, and relied on quantitative data health promotion or tobacco as a public health(Kumpfer et al., 1993; Rogers et al., 1993; problem. In coalitions where the new staff hadButterfoss et al., 1996). Several of the factors not been involved with the coalition during thefound to be related to implementation in the present planning phase, implementation was lower.study were related to member satisfaction and This study also suggests that staff skill inmember participation in earlier stages of coalition working with a coalition and how staff see theirdevelopment in other studies. role with the coalition are both related to coalition This research raises numerous issues for how effectiveness in implementation. The case studywe work with coalitions in community health data suggested that skilled staff are visionary butpromotion. It suggests that the amount of staff capable of attending to details, comfortable withtime spent working with a coalition is related ambiguity and sharing control, have strong inter-to coalition effectiveness in implementation. The personal skills, and understand complex, multi-research also suggests that coalitions that channel, multi-level interventions. If the local paybenefited from local vision rather than depending structure prohibits hiring well-trained, experiencedon state-level NC ASSIST staff had higher levels staff with the appropriate skills, then extensive andof implementation. Developing a local vision for a ongoing training and technical assistance shouldcomprehensive community-based health promotion be provided to the coalition staff. State-level NCproject designed at the federal level is challenging, ASSIST successfully provided ongoing technicaland some of the local Project ASSIST coordinators assistance through its field staff during the planningstruggled with understanding Project ASSIST and phase. However, the field staff became involveda community health promotion model themselves, in state-level activities in the intervention year, andmuch less communicating it to groups they were were not able to provide consistent and intensivetrying to recruit to the coalition. Most of the technical assistance to the newly hired coordina-coordinators worked with the Project ASSIST tors. Additional hiring of state-level field staffcoalitions part-time during the planning phase. One ultimately remedied this situation, but not beforeof the state-level staff commented that the local it affected implementation in the first year ofcoordinators often mentioned that if they could intervention.work on ASSIST full-time, they would be able to Another issue raised by this research is thethink one or two steps ahead of the coalition (rather possible need to add flexibility to funding timelinesthan depending on state-level NC ASSIST staff). to accommodate the natural evolution of coalitions.Thus, it seems likely that full-time coordinators During the first two planning years of Projectduring the planning phase may have helped instill ASSIST, the coalitions seemed to be driven, ina vision for Project ASSIST at the local level. large part, by deadlines set by state-level NC Another advantage of funding at least one full- ASSIST and the NCI, and related deliverables. Intime staff during the planning phase would have an attempt to meet the deadline for the first yearbeen a reduction in staff transition during the move action plan, many of the local coordinators draftedinto the intervention phase. In NC ASSIST, funds the action plan themselves with relatively littlewere not available to cover local staff salary member input. The case study data showed thatuntil the third year when intervention funds were these coalitions had lower levels of implementa-available. In many of the coalitions, the coordina- tion. In addition, after the action plans were written, 235
  12. 12. M. C. Kegler et al.the coalitions had to wait several months before members themselves. Coalitions in whichintervention funds were available. The long plan- members were not communicated with frequentlyning period contributed to the reluctance of some had lower levels of implementation. For thelocal coordinators to recruit additional members majority of coalitions, the most important methodwhen there was nothing for the members to do at of communication was group discussion at meet-that time. Adding flexibility to the funding timeline ings. Meetings should be structured to facilitatewould allow coalitions to evolve at their own purposeful interaction and communication betweenpace, involve members in planning and move members rather than as simply opportunities forimmediately into interventions after planning. updates. At a minimum, coalition staff and leadersFunding for each stage could be contingent on should send minutes and maintain phone contactsuccessful progression through previous stages. with members if meetings are somewhat infrequent. Another issue raised by this research is the The case study data also demonstrated thatongoing need to fund, develop, evaluate and dis- conflict and politics affected implementation in aseminate innovative interventions. The case study variety of ways. Although conflict may not bedata suggested that task forces that had concrete, avoidable, it should be acknowledged andtangible activities were the task forces that were addressed before it is too damaging. When coali-able to maintain the interest of the members and tions are engaging in ‘controversial’ work suchaccomplish something. Bracht et al. (1994), in an as tobacco control in a tobacco growing andarticle on the institutionalization of the Minnesota manufacturing state, there is a need to openlyHeart Health Program, discuss the idea of a ‘shelf- acknowledge and discuss strategies for how bestlife’ for certain programs. Although not yet an to deal with political opposition. North Carolinaissue with the Project ASSIST coalitions, youth Project ASSIST was fairly skilled in this area,buying operations and Commit to Quit events may successfully framing tobacco control as a healthlose their potency in later years of the intervention issue, and focusing on youth, pregnant women andphase. There needs to be a steady flow of new, tobacco users who want to quit.tested, interesting interventions for the coalitionsto implement. Study strengths and limitations This research also has implications for coalition Qualitative researchers emphasize the importanceprocedures, including the importance of clarifying of triangulation for enhancing the validity ofstaff roles, membership criteria, leader selection qualitative research (Lincoln and Guba, 1985;and decision-making methods. The case study Patton, 1990). One of the major limitations of thisdata suggested that bylaws or written operating research was the lack of triangulation among eitherprocedures served multiple purposes for the coali- observers, interviewers or analysts due to limitedtions. In several of the coalitions, it was the resources. This study was, however, strengthenedbylaws that served as the catalyst for identifying by triangulation of methods and data sources.community leaders for coalition chair positions An important step in qualitative researchrather than staff. Some of the bylaws restricted methods is the ruling out of other plausible inter-coalition membership to persons or organizations pretations to increase the internal validity of thethat did not have a conflict of interest with respect to findings (Patton, 1990; Yin, 1993). This involvestobacco control. Also, the decision-making process organizing the data in different ways, in combina-was unclear in many of the coalitions and these tion with thinking logically about other explana-coalitions tended to be those with lower levels of tions and seeing if the data support theseimplementation. alternatives (Patton, 1990). Due to the large number The findings also indicated the importance of of factors that influence implementation, it wasestablishing mechanisms for regular communica- difficult to tease out with any certainty which oftion between staff and members, and among the the factors were more important to effectiveness in236
  13. 13. Project ASSIST coalitions in North Carolinaimplementation. There were numerous differences Acknowledgementsbetween the coalitions; positive characteristicstended to occur together in some coalitions and Special thanks to Dr Christine Jackson, Dr Nancymore negative characteristics occurred together in Milio and Dr Pat Fischer for their guidance, andother coalitions, thereby making it difficult to to state and local NC Project ASSIST coalitiondetermine which factors were most important in members and staff for their participation andimplementation. numerous demonstrations of support for this Another limitation of this study was the use research. This study was part of a doctoral disserta-of the number of activities as the measure of tion funded through multiple sources including:implementation rather than a more comprehensive the NCI, the University of North Carolina’s Centermeasure of implementation quality. It is possible for Health Promotion and Disease Prevention, thethat the use of a more complex measure of University of North Carolina’s Cecil G. Shepsimplementation would have resulted in different Center for Health Services Research, and thefindings. In addition, the economic and political University of North Carolina Graduate School.climate associated with tobacco control in NorthCarolina may limit the generalizability of these Referencesfindings to other types of coalitions in other Alter, C. and Hage, J. (1993) Organizations Working Together:settings. Coordination in Interorganizational Networks. Sage, Newbury Park, CA.Directions for future research Altman, D., Endres, J., Linzer, J., Lorig, K., Howard-Pitney, B. and Rogers, T. (1991) Obstacles to and future goals ofThis research provided systematic documentation ten comprehensive health promotion projects. Journal ofof one type of coalition: tobacco control, commun- Community Health, 16, 299–314. Black, T. (1983) Coalition building—some suggestions. Childity based, mandated by a federal funding source, Welfare, 62, 263–268.long-term and relatively formal. This research Bracht, N. (ed.). (1990) Health Promotion at the Communityprovided an in-depth understanding of these coali- Level. Sage, Newbury Park, CA. Bracht, N., Finnegan, J., Rissel, C., Weisbrod, R., Gleason, J.,tions during implementation. Additional research Corbett, J. and Veblen-Mortenson, S. (1994) Communityis needed on other types of community health ownership and program continuation following a health demonstration project. Health Education Research, 9, 243–coalitions and on other stages of coalition develop- 255.ment to provide a foundation on which to base our Brown, C. (1984) The Art of Coalition Building: A Guideunderstanding of coalition behavior and coalition for Community Leaders. The American Jewish Committee, New York.life cycles. Butterfoss, F., Goodman, R. and Wandersman, A. (1993) Future research should also focus on specific Community coalitions for prevention and health promotion.dimensions of coalitions more intensively than was Health Education Research, 8, 315–330. Butterfoss, F., Goodman, R. and Wandersman, A. (1996)possible in this research. Our understanding of Community coalitions for prevention and health promotion:coalitions as a vehicle for community involvement Factors predicting satisfaction, participation and planning. Health Education Quarterly, 23, 65–79.would be further enhanced by focused investigation Centers for Disease Control and Prevention. (1995) CDC’son selected factors such as staffing or participation. IMPACT Program bolsters states’ role in tobacco control.Finally, two critical questions remain unanswered. Chronic Disease Notes and Reports, 8, 1–5. Fawcett, S., Paine-Andrews, A., Francisco, V., Schultz, J.,First, are coalitions any better than other strategies Richter, K., Lewis, R., Harris, K., Williams, E., Berkley, J.,for promoting health at the community-level? Lopez, C. and Fisher, J. (1996) Empowering community health initiatives through evaluation. In Fetterman, D.,Secondly, in what circumstances should coalitions Kaftarian, S. and Wandersman, A. (eds). Empowermentbe formed to address public health issues? Answers Evaluation. Sage, Thousand Oaks, CA.to these questions will help move the field of Feighery, E. and Rogers, T. (1990) Building and Maintaining Effective Coalitions. Guide No. 12 in the How To Guidescommunity health promotion to a more sophistic- on Community Health Promotion. Stanford Health Promotionated level. Resource Center, Palo Alto, CA. 237
  14. 14. M. C. Kegler et al.Florin, P., Mitchell, R. and Stevenson, J. (1993) Identifying Morrison, J. (eds). Community Organization and Social technical needs in community coalitions: a developmental Administration. Haworth Press, New York. approach. Health Education Research, 8, 417–432. National Cancer Institute (1991) ASSIST Orientation-MeetingGottlieb, N., Brink, S. and Gingiss, P. (1993) Correlates of Summary, December 3. coalition effectiveness: the Smoke Free Class of 2000 Patton, M. (1990) Qualitative Evaluation and Research Program. Health Education Research, 8, 375–384. Methods. Sage, Newbury Park, CA.Herman, K., Wolfson, M. and Forster, J. (1993) The evolution, Rogers, T., Howard-Pitney, B., Feighery, E., Altman, D., Endres, operation, and future of Minnesota SAFEPLAN: a coalition J. and Roeseler, A. (1993) Characteristics and participant for family planning. Health Education Research, 8, 331–344. perceptions of tobacco control coalitions in California. HealthKegler, M. (1995) Community coalitions for tobacco control: Education Research, 8, 345–357. factors influencing implementation. Doctoral Dissertation, Seidel, J., Kjolseth, R. and Seymour, E. (1988) The Ethnograph: University of North Carolina at Chapel Hill. A User’s Guide. Qualis Research Associates, Amherst, MA.Kegler, M., Steckler, A., McLeroy, K. and Malek, S. (1998) Shopland, D. (1993) Smoking control in the 1990’s: a National Factors that contribute to effective community health Cancer Institute model for change. American Journal of promotion coalitions: A study of 10 Project ASSIST Public Health, 83, 1208–1210. coalitions in North Carolina, Health Education and Behavior, Strauss, A. and Corbin J. (1990) Basics of Qualitative Research: 12, in press. Grounded Theory Procedures and Techniques. Sage,Kumpfer, K., Turner, C., Hopkins, R. and Librett, J. (1993) Newbury Park, CA. Leadership and team effectiveness in community coalitions Thompson, B. and Kinne, S. (1990) Social change theory: for the prevention of alcohol and other drug abuse. Health applications to community Health. In Bracht, N. (ed.), Health Education Research, 8, 359–374. Promotion at the Community Level. Sage, Newbury Park, CA.Lincoln, Y. and Guba E. (1985) Naturalistic Inquiry. Sage, Winett, R., King, A. and Altman, D. (1989) Health Psychology Newbury Park, CA. and Public Health: An Integrative Approach. PergamonMalek, S. and Enright, T. (1995) North Carolina Project Press, New York. ASSIST: a call to action. North Carolina Medical Journal, Yin, R. (1989) Case Study Research: Design and Methods. 56, 56–58. Sage, Newbury Park, CA.Miles, M. and Huberman, A. (1994) An Expanded Sourcebook: Yin, R. (1993) Applications of Case Study Research. Sage, Qualitative Data Analysis. Sage, Thousand Oaks, CA. Newbury Park, CA.Mizrahi, T. and Rosenthal, B. (1992) Managing dynamic tensions in social change coalitions. In Mizrahi, T. and Received on November 22, 1996; accepted on May 9, 1997238