The Structural Impediments to Human Service Collaboration: Examining Welfare Reform at the Front Lines

The Structural Impediments to Human Service Collaboration: Examining Welfare Reform at the Front Lines Jodi Sandfort Syracuse University This article examines the ability of frontline human service agencies to collaborate across organizational boundaries. Data come from an in-depth study of the public welfare and private welfare-to-work contractors in two Michigan counties and document significant problems that arise from the inability of these two sectors to collaborate in the provision of welfare programs. I use ethnographic methods that capture the perspectives of frontline workers themselves in order to understand how collaboration is actually thwarted. In spite of dramatically different organizational settings, frontline staff in both sectors draw on the same sources of evidence, that is, past relations, daily experiences, and client stories, when assessing the organizations with which they are mandated to collaborate. These collective beliefs create parameters within which staff interpret events and react to them. Their interpretations and reactions further reinforce the beliefs shared throughout the organization about the legitimacy, efficiency, and effectiveness of the partner organization. In this way, the social process at the front lines of the welfare system creates systemic barriers to collaboration. I conclude by considering how this analysis can help the local manager improve policy implementation and human service collaboration. Volumes have been written about the importance of increasing human service coordination, collaboration, and service integration to improve the quality and availability of social services for children and families.1 Collaborative efforts run the gamut from executive meetings for joint planning of new community initiatives to ‘‘one-stop’’ service centers Social Service Review (September 1999). q1999 by The University of Chicago. All rights reserved. 0037-7961/99/7303-0002$02.00 Human Service Collaboration 315 where clients use a common application form to access a range of services administered by different entities. Federal and state funding streams increasingly require that communities document ongoing human service collaboration in order to secure funding for new initiatives. Private foundations eagerly fund service integration pilot programs that seek to meet client needs in a more efficient, comprehensive fashion. As collaborative initiatives have mushroomed, scholars have exerted considerable effort to develop typologies that help make sense of this diversity.2 While such efforts are important, they do not help us understand more about the collaborative process itself, that is, how it is accomplished or impeded through the daily actions of human service managers and staff. Such an examination is important because, in spite of a focused policy and administrative efforts at all levels of the human service delivery system, collaborative initiatives are difficult to implement and sustain. These difficulties are often attributed to ‘‘politics,’’ ‘‘turf battles,’’ or ‘‘personality conflicts’’ between managers.3 Increasingly, the research community has looked beyond such individual-level explanations to other macrolevel factors. Some have identified organizational factors, such as structure, technology, and resource use as the sources of possible barriers.4 Others have studied the nature of interorganizational networks to identify the forces that may impede collaboration.5 Still others have claimed that policies promoting human service collaboration are largely symbolic, adopted for political rather than substantive reasons.6 Most social science research has been restricted to examining how these organizational, interorganizational, or societal forces influence human service collaboration. Yet, the indeterminacy of many social policies makes the front lines a critically important level of analysis. It is at the street level that many social policies are actually enacted, through the interpretations of staff and their interactions with policy clients.7 In spite of this, frontline conditions have not been investigated in our quest to understand the structural barriers to human service collaboration. Yet, without empirical investigation, we have little understanding of how frontline conditions actually affect the collaboration and coordination process. In this article, I use data from an in-depth examination of welfare service delivery in Michigan. In this case, collaboration entails coordinating the actions of the public welfare bureaucracy and private welfare-to-work contractors. If the actual social processes that occur at the front lines are examined, we discover that frontline staff in both the public bureaucracy and small contractors draw on the same sources of evidence—past relations, daily experiences, clients’ stories—when assessing the organizations with which they are mandated to collaborate. Through the social process within each local office, staff come to share a collective assessment of their collaborative partner agencies. These collective beliefs, in turn, create parameters within which staff interpret events and mobilize 316 Social Service Review organizational resources in response. Such interpretations and reactions further reinforce the schema shared throughout the organization about the legitimacy, efficiency, and effectiveness of the partner organization. In an environment charged with hostility, there is little reason for staff or management to improve communication, improve the efficiency of referrals, or learn the details of administrative policies governing the partner organization. Thus, the negative perceptions of the other organization allow staff to justify actions and legitimate inaction that further contributes to coordination barriers. This article examines this underlying process and argues that it has structural significance according to the definitions of social theorists such as Anthony Giddens and William Sewell, who assert that structure is more dynamic than conventionally assumed.8 Rather than being externally imposed through organizational form, formal procedures, or written rules, structure is more organic and internal. The social structures that guide people’s actions, that is, that help them develop appropriate routines, create justification for inaction, interpret the unexpected, evolve out of the collective beliefs and the concrete resources available within the organizational context. This analytical lens helps us see that collaboration within human service organizations is neither random nor informal. Rather, at the front lines, it occurs within structures that emerge from the nature of daily, street level practice. This understanding of interorganizational collaboration provides a new explanation why human service collaboration remains such an elusive goal and suggests new avenues for management interventions. Research Methodology This analysis is intended to generate new hypotheses about the role frontline workers play in implementing collaborative human service initiatives. I used a comparative, case study design and ethnographic methods to understand the conditions within five frontline welfare and welfare-towork organizations. These organizations were clustered in two counties, Dutchess and Dunn.9 In each county, I examined the local public welfare organization and the paired organizations that provide welfare-to-work services.10 I used a grounded theory approach that develops plausible relationships between analytical categories through a reiterative process of data collection and analysis.11 As a result, the initial analytical categories reflect the ‘‘native’’ meanings developed by the staff from each organization. Multiple ethnographic research methods were used in this examination: semistructured interviews, focus groups, participant observation, and content analysis of written materials.12 In total, I conducted 70 interviews and 13 focus groups with frontline staff and local managers, encompassing 485 hours of participant observation; I also read hundreds Human Service Collaboration 317 of written documents. Data were collected during a number of intensive periods between March 1995 and May 1997. The semistructured interviews and focus groups involved staff responsible for providing income support or welfare-to-work programs and lasted between 40 and 80 minutes. Each interview and focus group was audio recorded and transcribed, and the transcript was reviewed for errors. These data were greatly enriched by extensive notes from participant observation in each organization.13 Data analysis involved a recursive process of data collection and analysis. Initial analytical themes were identified through reading and rereading volumes of interview transcripts, field notes, and official documents. The data were then entered into a qualitative computer software package, QSR Nudist, which facilitated more rigorous analysis.14 Through systematic coding and continued modification of the coding system, this software assisted in the inductive analysis and exploration of interdependent themes. Additional field-based data were collected to expand and modify the coding system and further explore the emerging hypotheses. Through this process, it became apparent that the appropriate unit of analysis for this study of policy implementation was the organization, rather than the individual workers. Although individuals do exert discretion in the interpretation of social programs, it appeared that the more significant forces shaping this discretion were social and were located within the organizations themselves. As a result, this article explores the collective organizational reality shared by frontline staff within each of the organizations. The triangulation of multiple methods allowed me to place individual-level data within its larger, critical organizational context for this analysis. Policy Context Administrative Arrangements To understand the frontline collaborative process in the welfare system, we must first know a bit about its administrative arrangements. Like many states, Michigan abandoned its education and training welfare-towork strategy in the mid-1990s and, instead, adopted a quick labor-force attachment approach that it calls ‘‘Work First.’’ 15 Whereas the prior initiative was administered by the state’s Family Independence Agency (FIA) through contracts with private agencies, Work First was moved into a new cabinet-level agency, the Michigan Jobs Commission. Created in 1994, the Jobs Commission was formed to improve the business climate in the state through a range of economic development initiatives. As part of this mandate, the commission administers public programs that support workforce development among a range of populations, including welfare recipients.16 318 Social Service Review For Work First, the Jobs Commission develops policy guidelines, loosely monitors service provision, and generates reporting requirements. The actual administration, though, is left to the decentralized Michigan Works! system. The 26 Michigan Works! agencies are an assortment of not-for-profit agencies, local governments, and public consortiums that formerly administered the federal Job Training Partnership Act ( JTPA) programs. Because the Michigan Works! system gives maximum flexibility to local entities to shape their welfare-to-work programs, incredible programmatic diversity has evolved. All Work First agencies provide job search assistance to welfare recipients with the goal of helping them find, at very least, minimum wage employment for 20 hours a week. Some agencies offer brief training for job searching that covers such topics as résumé writing, application completion, and interview techniques, as well as focused job search assistance such as on-site interviews. Others require clients to immediately begin their job search with little training or support.17 Since 1994, this decentralized Michigan Works! system has been paired with the public welfare bureaucracy to deliver Michigan’s welfare-related programs. A traditional, state-wide bureaucracy, FIA is responsible for determining and monitoring eligibility for public assistance programs. All county offices operate under the same administrative policy and paperwork. They use the same service technology— completing an intake interview, requesting client verification, processing this verification to determine eligibility, and maintaining public assistance cases once they are open. Figure 1 illustrates the administrative arrangements in Michigan’s welfare system and the mandated lines of coordination at the state and local levels. There is communication from the state to the local level in both sectors. At the state level, managers meet with each other to coordinate policy, establish reporting systems, and monitor welfare-to-work implementation. At the local level, directors of county FIA and local Michigan Fig. 1—Administrative arrangements in Michigan’s welfare system since October 1994. Human Service Collaboration 319 Works! agencies must, at a minimum, sign working agreements. However, as the dotted line illustrates, the daily coordination between local FIA offices and Work First contractors is left up to local management and staff discretion. In Dutchess County, management matched frontline staff from FIA and Work First to allow for case conferencing about the status of a particular client. As we will see, however, this formal pairing was not sufficient to overcome all of the structural barriers to collaboration. In contrast, there was virtually no communication between frontline staff or supervisors in each sector in Dunn County. Problems Arising from Lack of Local Collaboration In implementing welfare programs, both state and local sectors are dependent on the actions of the other. After clients apply for public assistance benefits in the county-level FIA office, welfare reform policy changes require that they participate in a job search program delivered by the Work First contractors. If they do not attend, that information is conveyed to the local FIA office. FIA workers, then, make home visits to assess clients’ reasons for not attending and, if determined to be illegitimate, impose sanctions (reductions in the amount of cash assistance) for noncompliance. Clients then are re-referred to Work First and the sanction is lifted only if their attendance can be documented. This distribution of tasks between the sectors creates many systemic problems in counties with limited coordination between the sectors. For example, although clients provide extensive information to the public welfare office when applying for assistance, it is not shared with Work First operators. When a client is referred to Work First, she must complete another ream of paperwork. If she is referred again, she must complete the same forms again. A similar information gap occurs when a client is referred back to the public welfare office for assessment because she is not attending Work First. The FIA worker is given no information about the reasons for the client’s noncompliance. Did she encounter barriers, such as limited child care, a physical illness, or a broken car? Did she refuse to attend at all or get discouraged after months of searching for a job? Did she leave after the first week, disgusted because of the poor quality of the job preparation seminars? Without coordination, frontline staff on both sides of the system duplicate the information already gathered by their partner organization. This limited information makes it difficult for staff to efficiently monitor clients’ participation in Work First. In the new policy environment that demands clients’ participation, this has serious consequences. For example, 73 percent of the clients referred to Work First in Dutchess County and 63 percent in Dunn County were sent back to FIA in the 1996 program year because they did not comply with program requirements.18 Because of the limited information shared between the sectors, 320 Social Service Review a client can bounce back and forth between the two sectors. Clients need only to agree to cooperate when visited by a public welfare worker, ignore the solicitations from Work First when they are re-referred, and be sent back to FIA for another assessment. Since frontline staff resist imposing sanctions, this cycle can continue indefinitely. In the end, only a small proportion of noncompliant cases actually receive sanctions.19 As a result, clients learn that, despite political and programmatic rhetoric about mandatory participation, the system does not follow through with its threats. The lack of collaboration also causes inconsistent information to be conveyed to clients about new policy stressing work and self-sufficiency. Rarely can staff accurately answer questions about the daily functioning of their partner agency. When faced with clients’ basic questions— ‘‘When will my day-care application be processed?’’ ‘‘What training and services are offered in Work First?’’ ‘‘Should I wear interview clothing to the first day?’’—frontline staff give a guess, rather than appear ignorant to clients. Although FIA has adopted numerous policies to encourage clients to combine welfare and work, Work First staff are not aware of these initiatives because of the limited communication between the sectors. In one instance, Work First staff used a budgeting exercise in the classroom to occupy clients because the normal job training workshops could not be held. The exercise was, as explained by the staff member, ‘‘designed to assist you in planning your own personal budget once you are employed.’’ For the next 30 minutes, the clients worked on completing a weekly record of their expenditures. They then calculated the hourly wage they would need to earn to meet these expenses. Although the state has extensive adjustments to earned income to provide an economic incentive for welfare recipients to work, staff did not instruct the clients to make these adjustments. Surprised because these rules were emphasized every time FIA staff came to the site for orientation, I asked about the omission. ‘‘Oh, I don’t know how they make those budget calculations,’’ replied the case manager in an off-handed manner as he continued to help clients complete the worksheet. As they worked, various clients exclaimed about the high hourly wages they would need to earn to meet their expenses. Although they, too, had received the brightly colored sheets from FIA with detailed information about grant adjustments, none of the clients questioned the exercise. Instead, this activity incorrectly reinforced their perception that in order to pay their basic expenses they would need to get jobs paying over eight dollars per hour. Because they spent time making these calculations for themselves, it is likely that these clients would remember this misinformation—rather than the abstract calculations in FIA pamphlets—when deciding whether or not to accept a particular job. Thus, the lack of coordination between FIA and Work First contributes Human Service Collaboration 321 to system inefficiencies, sustains ambiguous messages given to clients, and conveys misinformation about program rules. These conditions persist in spite of formal policy mandates requiring that both sectors work together in the delivery of welfare programs. They persist although the success of local offices in both sectors depend on their ability to coordinate and cooperate with their partner organization. By looking at front-line conditions in both FIA and Work First organizations in Dutchess and Dunn Counties, we can develop a more complete understanding of how these conditions are created and how they are sustained through daily actions. Analyzing Street Level Collaboration As mentioned earlier, analyzing interagency coordination from the ground up requires a two-tiered approach. First, we must explore the particular meanings both FIA and Work First staff develop about the external environment and, in particular, the other local organizations with whom they are partnered. Yet, these meanings have limited generalizability because they develop out of particular organizational arrangements and practice experiences. To see their larger significance, we also must look beyond these specifics to uncover the underlying process that shapes frontline collaboration. This second level of analysis will yield new insights about the impediments to frontline collaboration in a variety of human service organizations. Collective Beliefs within the Family Independence Agency As a large public bureaucracy, Michigan’s FIA is fairly insulated from the external environment. Civil service rules and strong unions have protected staff from layoffs or drastic restructuring. Daily work involves mastering detailed policies, procedures, and paperwork, a daunting task that is difficult for outsiders to fully grasp. The network of colleagues rarely includes those outside the organization and is limited to others within the FIA ‘‘family,’’ that is, to people they may have worked with at other local offices, or see at training sessions and conferences, and those who share their war stories of life on the front lines of this difficult system. The boundary between the organization and its environment is reinforced by the stigma and frustration staff experience when interacting with outsiders. As state employees, frontline staff believe that the general public sees them as lazy, ineffective, and stagnant. As employees of the ‘‘welfare office,’’ they feel additional stigma. They recount painful stories of being at parties and having acquaintances chastise them when it is revealed where they work.20 In addition, most contact with other human service organizations tends to be adversarial. Social workers from private agencies call on behalf of clients to inquire why a certain action on a case has not been taken, why authorization is so slow, or why additional verification is required. These experiences confirm that others do not un- 322 Social Service Review derstand the technical work of eligibility determination and increase the isolation of FIA staff. From a frontline vantage point within FIA, local Work First operators are merely another human service organization in the external environment. Yet, as we have seen, the relations with these organizations is particularly important because of their role in the delivery of welfare services. In both Dutchess and Dunn Counties, the local FIA offices had prior relationships with the Michigan Works! agencies (MWAs), the fiduciary agencies that administer Work First, before they were given responsibility for the new program. Each had previously contracted with the organizations for specific job-related training through the federal JTPA programs, and both were displeased with the quality of services; they believed the agencies worked only with highly functional clients, provided incomplete documentation of client progress, and struggled with poor management. Neither organization had a good track record working with welfare recipients. When state administrators removed responsibility for Work First from FIA and gave it to these seemingly incompetent agencies, frontline staff and management were skeptical that these local organizations could provide the service effectively. These impressions are slow to change. In fact, they provide a framework from which staff understand interactions with the partner agency. The personal experiences of staff often confirm the negative impression of the Work First organizations. Although certain frontline staff are paired with Work First case managers to facilitate communication about individual clients in Dutchess County, their experience with Work First staff creates boundaries that are difficult to overcome. It is widely known that the program managers from each sector have ‘‘personality conflicts.’’ For example, in the monthly meeting with frontline staff from both organizations, these managers get into disagreements about processing rules, staff capabilities, or client circumstances, which can escalate into shouting matches. Staff members recount details about these meetings to their colleagues who were not present. Others share stories about Work First staff who encourage clients to demand services when the FIA is a bit late processing an application or authorizing payment for child care. Through these mechanisms, staff draw on their collective experience and create the impression, palpable throughout the organization, that the local Work First program is ineffective. In Dunn County, frontline workers are not matched with each other and there is little contact between frontline staff in the two sectors. Certain staff members do attend the Work First orientation to explain formal policies and child-care reimbursement procedures to clients. As in Dutchess County, these ambassadors return to the office and share their impressions with colleagues. At the orientations, staff from Work First contractors get up and make brief, vague presentations, leaving plenty of questions about the nature of the program unanswered for both FIA Human Service Collaboration 323 staff and welfare clients. Some contractors, though, have staff who seem to work well together and treat clients respectfully. FIA workers believe these programs are better than the norm. Such general impressions are shared with colleagues in the office and become what FIA staff in Dunn County ‘‘know’’ about their Work First partners. In both counties, staff members also rely on client testimonies to provide evidence about the effectiveness of their partner organization. Stories of client mistreatment add fuel to the belief that Work First is inefficient and misguided. One frontline supervisor recounted such a story in an interview: We had a client who called three different times to [the Work First agency], got three different people and [was] told three different things. And all [the client was] trying to find out was, ‘‘Should I dress up because am I going straight to a job site after . . . my orientation?’’ One told her, ‘‘I don’t know, call back.’’ One told her, ‘‘Oh, you don’t even have to show up for your orientation.’’ And another told her, ‘‘Call your worker, I don’t know.’’ And so that’s how we got the call. This person didn’t know what to do. . . . I advised [the worker] to tell her . . . it never hurts to be ready. We don’t know [what happens], but it would be better if you’d be ready. Stories like this are repeated throughout the organization because client experiences provide additional data about the effectiveness of local Work First agencies. In fact, clients’ stories are seen as a litmus test for the system. If clients experience poor treatment, inadequate service, or misinformation, frontline staff believe clients will mention it when they come into the welfare office for periodic interviews. After a number of months with few complaints, they wonder whether or not the details of program operation are getting slowly worked out over time. That is, until they hear a report from another client. In a system with little open communication or understanding shared between frontline staff, clients are important sources of information that often confirm the worst impressions of local collaborative partners. Impressions about Work First also evolve when the organizational norms conflict. Michigan’s governor has publicly stressed his goal to increase the privatization of social services and has instituted a number of initiatives that have significantly restructured or reduced the government workforce. In this environment, the private Work First contractors are yet another example of the constant, incremental spread of privatization. Because the Michigan Works! system is oriented toward output measures and financial reward, staff perceive that local programs exist to make a dollar rather than to provide quality programs to clients. As one frontline supervisor recounted, ‘‘at the Work First conference . . . they didn’t talk about the clients, they talked about the ‘client commodity.’ [They forget that] these are people . . . that we need to help!’’ The language and private sector tone of the conference was foreign to her. In 324 Social Service Review contrast to the public-sector environment of FIA, where the provision of social services is normative, this new business-oriented service provision is threatening. In this context, there is little incentive for FIA staff to collaborate with the local Work First operators in order to improve their performance. At a deeper level, FIA workers are frustrated by Work First staff not understanding the details or significance of FIA’s administrative rules. Work First staff call demanding immediate action on this or that client’s case, not realizing the volume of forms that need to be completed, the minute rules that dictate clients’ eligibility, and the complex computer screens that must be navigated. Within FIA, these rules and formal processes are paramount. They are the raw materials of daily tasks. FIA staff members believe that Work First staff not only have little understanding of these rules and processes but also have little regard for their significance. In spite of the mandated collaboration for the delivery of welfare-towork programs, then, these frontline workers in FIA feel negatively about their Work First partners. In part, this can be attributed to FIA’s wariness about the organizations’ abilities given prior experience with their performance. However, this resistance is sustained through staff experience, tales they share about management conflicts, and system inefficiencies. Moreover, it is bolstered by client reports and the organizations’ contrasting priorities. Rather than being viewed as true partners in the delivery of welfare-related programs, Work First operators are seen as an ineffective and slightly threatening force. As one Dutchess County supervisor stated, ‘‘I cannot think of one case that I’ve come across where Work First put anyone to work. Our clientele have had nothing good to say. There has been a terrible lack of communication.’’ Although this man had little direct contact with Work First because of his supervisory position, his comments reflect the viewpoint shared by staff within this local office. However, this perspective is quite different from that found among frontline Work First staff. Collective Beliefs within Work First Contractors In contrast to the insulation of FIA, the boundary between Work First organizations and other agencies is quite permeable. The Michigan Works! system is based on performance monitoring; each organization is assessed by how many of its clients find employment and how many clients retain jobs for 90 days. In fact, the nature of frontline work— engaging clients in the job search program, assessing the needs of local employers, receiving referrals from the local FIA—necessitates that staff interact with a host of players outside of their immediate organization. Since Work First is operated by diverse contractors—schools, private nonprofits, private for-profits—Work First staff are not isolated with other ‘‘welfare’’ professionals but rather spend their days surrounded by Human Service Collaboration 325 colleagues engaged in diverse social programs. They are not stigmatized, as are FIA staff, because they work for the ‘‘welfare office.’’ Soon after Work First was instituted in 1994, frontline staff began to feel that FIA was a resistant, uncooperative partner. This viewpoint emerged from a number of experiences. FIA often referred cases with incomplete or inaccurate information. Sometimes they referred clients who were already employed, and rarely was child care provided in a timely manner so that clients could attend job search preparation sessions. Work First staff often received incorrect referrals for which staff had to dedicate considerable time to doing ‘‘FIA’s work,’’ for example, verifying clients’ employment status, documenting disabilities, assisting with child-care arrangements. FIA also rarely enforced sanctions for clients not complying with program requirements, which, in turn, limited the ability of Work First programs to have high participation rates. As long as clients heard on the street that program rules were not really enforced, that Work First was not really mandatory, resistant clients would continue to avoid participation. These problems with FIA have persisted over time and set the backdrop against which daily interactions occur. It would be reasonable to assume that regular communication between frontline workers might help to dispel this negative view of the public welfare office. However, Work First staff in Dutchess County, where frontline staff from both sectors are paired to facilitate communication at the street level, are the most negative of all providers in their assessment of FIA. For one, staff members resent FIA’s constant monitoring of their program. ‘‘For every form we have, there was a controversy with FIA behind it,’’ stated Jeanine, a case manager, reflecting the frustration felt throughout the organization. Rather than interpreting the increased standardization as a natural element of program establishment, staff members perceive it as being caused by the public welfare bureaucracy. In addition, Work First staff become frustrated by their collaborators’ resistance to exert reasonable judgment. They see FIA workers as reluctant to bend the rules for supportive services or program attendance, even if doing so would give a needy client the assistance she needs to secure that first job. When Work First staff try to intervene on a client’s behalf, many FIA workers often are uncooperative and hostile. In this county, where frontline staff are paired with each other, these actions become charged, personal affronts. In addition to their professional experiences with FIA, Work First staff also draw on their personal experiences with the public bureaucracy. In two of the three organizations examined in this study, nearly one-third of the Work First staff had been on public assistance themselves. One program even hired a few motivated clients to assist in classrooms because management believed such women would be better able to ‘‘relate’’ to the clientele. While this personal experience may bolster the credibility of the program to clients, the personal experiences of these 326 Social Service Review staff also inform the collective knowledge staff develop about their FIA counterparts. Together, staff relate stories about probing interviews, demeaning interactions with workers, and unjust application of policy. Michelle, who was able to leave the welfare rolls because of her job in Dutchess County, shared her impressions of FIA staff: ‘‘I think they pick people who are cold hearted and uncaring. Are you this type of individual? Do you have this type of personality? Then you can work for us . . . Because, I can’t understand how the whole Department’s got that same attitude. . . . But some of them are nice. I’ve run across a few nice ones.’’ The impression, that is, that most FIA staff are ‘‘cold hearted’’ and ‘‘uncaring,’’ is developed from Michelle’s personal experience in the welfare system. When talking with colleagues, she draws on this experience that shapes how they understand interactions with individual FIA workers. The stories clients tell to Work First staff about mistreatment by the welfare system—how workers treat them disrespectfully, how their phone calls are rarely returned, how they wait in the FIA lobby for hours on end—further reinforces this negative viewpoint. The welfare office seemingly functions to purposively frustrate clients efforts to care for their families. It creates unnecessary barriers. While some clients are lucky enough to get ‘‘good workers,’’ that is, people who are pleasant, who treat them with respect, and who may bend the rules slightly, many more seem to get those workers who are incapable. One client in Dunn County, for example, took a Work First staff member into the bathroom to reveal a healing incision on her chest from open heart surgery. Given FIA policy, she technically was required to participate in a job search because her allotted recovery period had expired. Yet to Work First staff, neither versed in FIA policy nor committed to the blanket application of rules, this client was clearly not employable. Her referral was seen as yet another example of an FIA worker’s incompetence. Frontline staff opinions about FIA also emerge out of conflicts between organizational norms. As staff of small, contracted organizations, it is imperative for staff that they respond to clients’ needs, particularly if a client demonstrates a good faith effort to participate with the program.21 Staff in some programs will provide individual transportation for clients who secure an interview, help others find professional clothes at a local thrift store, give suggestions about appropriate hair styles and makeup. As employees of organizations dedicated to moving clients from welfare to work, staff see their role to be assisting clients to leave welfare far, far behind and will go often beyond their official job requirements for, at least, a small proportion of their caseload. This norm is upheld by the actions and values of Work First staff and is experienced in sharp contrast to norms of the rule-bound public welfare bureaucracy. From this perspective, it is difficult to understand the motivation of FIA staff who stringently enforce the rules, who are perceived to unthinkingly Human Service Collaboration 327 push paper, and refer clients to Work First even when they are clearly unemployable. Rather than understanding that such bureaucratic processes help to guarantee more equitable treatment for all clients, Work First staff interpret these actions from the norms of their own organizational context. As is true for their FIA counterparts, frontline staff in these Work First organizations have negative impressions about their mandated partner. Early in the implementation of the new welfare-to-work initiative, staff learned through the daily experiences that FIA was a resistant partner in welfare reform. This impression reinforced the notions some staff had developed from their own experiences as clients in the system and was further solidified by the stories of clients. Over time, this impression culminates in exasperation with the local public welfare office that leaves little incentive for improving communications or interorganizational relations. Using Collective Beliefs to Interpret Events and Justify Actions When organizations are mandated to collaborate with each other, staff members’ collective beliefs are significant to service delivery. Each time staff interact with their counterparts in the partner organization, they must navigate these differing viewpoints and points of contention. In cases such as this, in which staff from both sectors hold negative views of their collaborative partner, this navigation is challenging. It is unlikely that staff will interpret actions of the partner judiciously in an environment charged with hostility. It is more probable that these viewpoints allow staff in each sector to justify actions that further attenuate relations between the organizations. In fact, this analysis reveals that the collective frontline workers’ beliefs provide a framework from which they interpret other events, justify daily actions, and legitimate their inaction. Take, for example, a video developed in Dunn County that explains welfare reform policies to clients. Because of the county’s size, there are many private organizations providing welfare-to-work services. Initially, client orientations for Work First were held at a centralized location, allowing FIA staff to attend the session, explain current welfare rules to clients, and answer any questions. Later, it was decided that the orientation would no longer be centralized but that the Work First providers would conduct such a session at their own site. Since there were 12 Work First agencies at the time, FIA management felt it was inefficient to send a staff member to each site. As a compromise, a group of workers showed initiative and made a simple, amateur video. In 8 minutes, they explained how Work First allowed clients to fulfill their responsibility to the state and illustrated how grants are altered when clients begin working. Inside FIA, management and staff were very proud of this innovation. It was one instance when staff acted proactively to respond to a perceived 328 Social Service Review need in the community. Yet, when talking with Work First staff, some FIA workers got the feeling that not all contractors were using the video because they didn’t like the message it conveyed. This attitude frustrated FIA staff, particularly because they felt the video was the only feasible way to convey accurate information in all of the orientation sessions. The source of resistance came from Work First staffs’ radically different perception of the video. They were appalled by it. How could the FIA put out such unprofessional, sloppy materials? Clearly, they were too lazy to send staff out to the sites to present the information in person. Furthermore, they were suspicious as to why material about other FIA programs and policies was included in a video that was supposed to be about Work First. From their vantage point, the video only fueled client resistance to the program and sustained confusion about the new requirements. When clients heard about the other options available, they wanted to take them. Work First staff resented the confusion it caused. ‘‘I tell them they’re here so they have lost their choice. It is Work First for them,’’ explained Marjorie, one orientation facilitator. Once again, Work First staff felt that they needed to fix a problem that FIA had created. This video, implemented by FIA to simplify working relationships and provide basic information to clients, became a charged symbol within both sectors. Staff emotionally reported to their colleagues when the video was used and when when it was not used. FIA tried to require that it be shown. Work First staff resisted. In the struggle over program authority in Dunn County, this symbol helped to confirm for frontline staff the unreasonableness of their partner organization. In the end, it fueled more resentment among the local organizations. This incident illustrates the challenge of interagency collaboration in a charged environment where frontline staff have only negative beliefs to help them understand the actions of other agencies. In developing the video, FIA staff attempted to convey complex details about eligibility to clients in a clear, concise manner. Given the logistical change and limited resources, the video seemed a good compromise. Since, at that time, Work First was just one of a range of options that clients could use to fulfill their responsibility to the state, from the FIA vantage point, it was important to provide that context to clients. However, Work First staff members did not understand the programmatic details and, instead, interpreted it according to their own framework: because Work First posed a threat to the public welfare bureaucracy, workers would do anything to undermine the program’s success. On the front lines of the welfare system, the impressions that staff develop about organizations in the environment create parameters within which they interpret events that unfold. Such interpretations further reinforce the beliefs shared throughout the organization about the legitimacy, efficiency, and effectiveness of the partner organization. In such a Human Service Collaboration 329 context, there is little reason for staff or management to make efforts to improve communication, improve the efficiency of referrals, or learn the details of administrative policies governing the partner organization. Thus, the negative perceptions of the other organization allow staff to justify actions and legitimate inaction that further contributes to coordination barriers. In this case, the significance of this social process is heightened because of the interdependence between the welfare and welfare-to-work organizations in the delivery of services. FIA is responsible for providing child-care subsidies while a client participates in the Work First program. As is common within the large social service bureaucracy, it takes time to process applications and, often, additional documentation is required before a subsidy can be paid. Work First staff do not understand the caseload or paperwork tasks associated with opening a day-care case. Faced with constant questions from clients about day care, about the source of the delay, they often call FIA workers. Their calls are not returned and frustration rises. Again they try. No response. The clients continue to question. When, after days of trying, someone actually gets an FIA worker on the phone, he is not able to contain his exasperation and explodes at the worker. In his mind and in the minds of his colleagues, FIA staff are purposively sabotaging the success of Work First through the slow provision of child-care subsidy. If clients cannot find child care, it is difficult for them to attend the workshops and difficult for them to find and retain employment. To the FIA worker receiving the explosive phone call, the interaction further confirms the unreasonableness of the Work First agency. While she might respond to this particular request to pacify the aggressive worker, she is unlikely to change her practices to be more responsive to day-care applications. In fact, when any FIA workers receive any phone call from their local Work First partners, they have little incentive to respond quickly. Given their beliefs, which are reinforced by their daily experiences and those of their colleagues, that the local Work First organization is incompetent, inefficient, and adversarial, it is easy to put off returning the call given the other, more pressing demands for their time. It is not unusual for an FIA member staff to withhold information about a client when the Work First worker happens to catch him or her on the phone; since they are outside of FIA, in an organization that seems to threaten FIA’s existence, there is little reason to provide assistance. There are instances, however, where successful communication does occur between frontline staff in FIA and Work First. The rationale behind these interactions, however, reflects staff beliefs about their partner organization. For example, in one Dunn County Work First office, a few staff members assume responsibility for working with FIA. These liaisons, informed by the personal experience of some colleagues that there are a few ‘‘good workers’’ within FIA, look to find some of these same work- 330 Social Service Review ers to help them navigate the massive public bureaucracy and act as a conduit for information sharing. Once these individuals are found, they are cherished like close personal friends. As Karen, a Work First case manager, explained, these relationships evolve slowly over time and must be cultivated very carefully: There’s three different [FIA workers] I talk with. I have their phone numbers [in a special place]. . . . It started out with Jack Riley and he would ask me questions and I would say, ‘‘You know, while I have you on the phone, would you mind answering one for me that I’m kind of curious about?’’ It was for the benefit of the person, for a client, and I could tell he was a helpful person and he was willing. So then I started to call on him, then I started to call on his colleague, Sandy, and it just started. They were asking me questions . . . and we’ve just worked back and forth. At one point, Sandy did say to me, ‘‘You know, this whole system working could make me lose my job.’’ But if you really care about the clients, you want the system to work because you don’t want them to get screwed. And that’s exactly what Sandy is all about. In this case, being able to locate those few helpful FIA staff helped to humanize the collaborative process. It also allows staff members in both sectors to see their common purpose: to assist clients in becoming selfsufficient. Yet, the FIA workers with whom Work First staff develop these special relationships are perceived as exceptions. That is why their phone numbers are kept in a special place for easy access. But, at least within this Work First office, frontline staff do believe that individuals who ‘‘really care about the clients’’ can be found, even within the cumbersome and inefficient public bureaucracy, even when the success of the program may make FIA staff ‘‘lose [their] jobs.’’ Unfortunately, it is more common for frontline staff in both sectors to use the other as a scapegoat. I often observed staff evoking the responsibility of the partner organization when confronted by a hostile client. At Work First orientation, staff in one program regularly stressed to clients how their organization was different from the public welfare bureaucracy. ‘‘We are not here to hurt you, but to help you. You must comply with our requirements or we will refer you back [to FIA]. But we can’t cut your grant. We aren’t FIA.’’ On the other side of the city, an FIA worker talked about her strategy for diffusing clients’ hostility during investigations for noncompliance with the Work First program. She claimed that since Work First programs inform the state about clients’ noncompliance, she often will put ‘‘blame on them’’ when talking with clients. In this environment where frontline staff perceive the partner organization as adversaries, staff think it perfectly legitimate to use this organization to diffuse client hostility and avoid accountability. Thus, the beliefs that frontline staff collectively generate about organizations in their environment are significant in that they impede the collaborative process. For one, they create the parameters within which Human Service Collaboration 331 staff are likely to interpret events. As the FIA video illustrates, an initiative undertaken with the best of intentions can backfire in an environment in which staff mistrust the motivations and actions of another organization. Moreover, frontline beliefs create the norms that staff rely on to prompt their actions and justify their inactions. With the notion that FIA is inefficient and obstructionist foremost in their minds, Work First staff find it perfectly appropriate to make repeated phone calls and demand the attention of FIA staff. With the reality that Work First is incompetent and adversarial foremost in their minds, FIA staff feel it legitimate to avoid these same phone calls. Both feel it acceptable to use the other as scapegoats when faced with a difficult situation. Finally, the actions and events that unfold help to reinforce the very beliefs that justified them. The video further attenuated the collaborative process because it provided additional evidence to frontline staff of the hostile intentions of the partner organization. Each Work First staff who is demanding or aggressive on the phone further reinforces the FIA worker’s negative impression of organization. Each FIA worker who does not return calls further convinces the Work First staff of the unprofessional behavior within the public welfare bureaucracy. The Structural Dimensions of Frontline Collaboration These data documenting barriers to interagency collaboration among the frontline organizations in Michigan’s welfare system reflect the particular organizational arrangements, histories, and practice experiences of FIA and Work First agencies. To understand the larger significance of these data—and the lessons they hold for other attempts at human service coordination and collaboration—we must examine the underlying process they suggest. In spite of dramatically different organizational conditions and service technologies, frontline staff members in both the public bureaucracy and private contractors develop impressions about their environment from their collective experiences. Although the FIA bureaucracy is a traditional bureaucracy with rigid boundaries separating it from the external environmental, and Work First agencies are small organizations that interact frequently with other organizations, frontline staff rely on the same sources of evidence—past relations, personal experience, client stories, normative organizational processes— when making their assessment of other organizations. The similarity of this process, in spite of significant organizational differences, suggests that it may be generalizable to the frontline collaborative process in other human service organizations. P. K. Manning articulates a proposition that is consistent with the findings of this study: ‘‘To some degree, organizations define, structure and shape the environment in which they are operating. . . . The ‘social construction’ of the environment results as the interpretive work of organizational members 332 Social Service Review is accomplished: they are socialized into organizational motives, contingencies, and team work; . . . they absorb tacitly shared assumptions and emergent definitions of contingent situations; and they learn the principles, working rules and practices thought to be the commonsense basis of the occupation.’’ 22 As Manning suggests, the beliefs that frontline staff have about other organizations in the environment are shared. In the course of a day, frontline staff relate stories over the lunch table, talk with each other about a volatile meeting with another agency, and call colleagues over to listen to a hostile phone message left by a staff member from another organization. Through these experiences, individual staff members are ‘‘socialized’’ into common opinions about the external environment. This mechanism creates the ‘‘tacitly shared assumptions,’’ and ‘‘practices thought to be the commonsense’’ within that organization. Because these beliefs are shared, they affect how service provision is accomplished. As we have witnessed, this shared practice knowledge provides staff members with a way to interpret events and provide a rationale for their actions. They cause a video to become a charged symbol of agency resistance, create justifications for not returning phone calls, and normalize the use of the partner agency as the scapegoat with clients. With each incident, staff become even more convinced of the adversarial intent of the partner organization. Such actions reinforce the very beliefs that have normalized them within the organizational context. A body of theoretical and empirical research developed over the last 15 years can help us understand how this frontline process relates to the persistent barriers to interorganizational coordination and collaboration. From diverse disciplines, many scholars offer evidence that such a process—whereby people in organizations draw on collective beliefs and assumptions to interpret events, to evaluate and justify their actions— has structural significance.23 From this empirical and theoretical perspective, people exist within social contexts—societies, communities, organizations—that do not possess inherent structures. While organizational charts, written rules, or other things typically regarded as elements of organizational structure may be imposed, the enacted or deep, social structures that guide people’s actions actually emerge out of a group’s daily experience within a particular context, as they collectively try to make sense of the circumstances in which they find themselves. This process, whereby people create from their collective experience the parameters that shape their understanding and direct their actions, is called ‘‘structuration’’ by some scholars.24 This perspective highlights that people are not passive dupes controlled by larger systemic or structural forces. Rather, the experiences they have each day, the beliefs that develop from these experiences, and the resources present in the organization to respond to these experiences, actually give rise to the structure of their social context. Yet, such structures both emerge from daily experience and constrain Human Service Collaboration 333 daily experience. Because structural parameters define what is rational and justifiable within a particular context, certain actions are curtailed or constrained. Although individual public welfare workers might realize that working collaboratively with Work First agencies might improve the efficiency within which they serve clients, the negative impression of the other agency makes it difficult to imagine how to overcome the structural barriers. Although one Work First agency developed relationships with a few treasured individuals in FIA in order to make welfare service delivery more efficient, staff justified this action in terms of the structure of the agency—that there are a few ‘‘good’’ workers within the public welfare system. Structuration theory stresses that these actions actually reinforce the legitimacy of social structures themselves. As individuals adjust their beliefs and actions in relation to the collective norms, they reinforce the legitimacy of the structure. This is the recursive, mutually dependent nature of action and structure. Structure is not imposed. Rather, it is created from day-to-day experience, solidified by day-to-day experience, and maintained by day-to-day experience. Understanding this social process as a structural one is consistent with how street-level staff in Michigan’s welfare system experience attempts at collaboration. From a frontline vantage point, they observe daily the problems caused by a lack of coordination between the two sectors. They complain about not knowing why a client was referred for assessment by the partner organization. They are frustrated when clients ask them questions about policies and procedures they know little about. They bemoan their inability to monitor client compliance with rules both because they see many clients who ‘‘work’’ the gaps and learn of others who fall into the chasm separating the sectors. However, in spite of experiencing the consequences of a poorly coordinated system, frontline workers feel powerless to change the circumstances. There are, from their experience, problems with ‘‘the system’’ that transcend themselves as individuals. Yet, this analysis reveals how the beliefs and actions of frontline staff, themselves, create and sustain these structural barriers. As individuals learn, in Manning’s terms, the ‘‘tacitly shared assumptions’’ of how their colleagues understand other organizations, they begin to assume these viewpoints as their own.25 They use these viewpoints to interpret an abrupt interaction at an orientation session with a staff member from the partner organization. They share the experience with their colleagues and, thus, provide additional evidence to the collective about the unreasonableness of the organization. Later in the lunchroom, someone else shares a story from a new client about her mistreatment by the organization, further reinforcing its illegitimacy. From this analysis, the collective beliefs shared among street-level human service staff seem to be important factors influencing how interorganizational collaboration is implemented. The following proposi- 334 Social Service Review tions merit further empirical investigation as scholars strive to better understand how service collaboration and coordination is actually implemented at the front lines of human service organizations: frontline staff develop collective beliefs and understandings about effective practices and appropriate organizational-partners through past relations, collective experience, and clients’ stories. Moreover, frontline staff rely on these collective schema to shape their interpretation of events, mobilize resources, and provide a standard for rational action. Finally, these collective beliefs become structural through a reiterative process whereby staff act within these parameters and, thus, reinforce their legitimacy. Such structures support or impede interorganizational coordination and collaboration. These propositions emerge from my analysis of ethnographic data from five frontline organizations in Michigan’s welfare system. The scope of this study, however, prohibited the thorough exploration of these ideas and a detailed investigation of these hypotheses. Future research should explore how such structures are created and sustained at the front lines of human service delivery, and it should explore whether or not existing frontline structures can be used to improve implementation of interagency collaboration efforts. Future research should also examine the process whereby frontline social structures change. My research suggests that the frontline structures that exist in human service organizations may be quite rigid. For individuals working at the front lines of human service organizations, the knowledge they have developed from their daily experience—interacting with other professionals, listening to client complaints, making phone calls that are never returned—is more legitimate than new management initiatives or grandiose plans for a system’s reform. Rather than being rooted in the daily reality of the front lines, these initiatives emerge from abstract ideals and political motivations. The social structures that shape frontline actions cannot be altered by merely vague management initiatives that promote communication and collaboration. Additional research is needed to document how frontline structures respond to management interventions to improve collaboration in various contexts. Implications and Conclusions This analysis of the welfare system reveals that the social process occurring within frontline offices has structural significance that impedes interorganizational coordination. The inability to coordinate frontline actions between the local public welfare and private welfare-to-work organizations creates system inefficiencies, conveys ambiguous messages to clients about policy goals, and causes inaccurate information to be relayed about existing programs. In the face of such consequences, it Human Service Collaboration 335 is important that interventions be developed to overcome such structural impediments. However, initial analysis suggests that frontline social structures may resist overt management overtures, especially when they are couched in the vague promise of service integration and systems change. So how can this analysis help practitioners improve interorganizational coordination and collaboration in human services? For one thing, it shifts our analysis of the cause of collaboration barriers. When we persist in understanding the cause of the impediments to human service collaborations as individual—as management ‘‘turf’’ issues or personality conflicts—we only consider interventions to overcome the problem at the individual level. Executive meetings and planning sessions are sponsored. Formal agreements are reached to institute staff collocation, service referrals, or the sharing of client information. Yet, too often, these initiatives are not fully implemented and incorporated into daily frontline practices. This analysis directs managers to attend to the social process that underpins collaboration efforts. It directs them to not discount the collective experiences that staff share with each other, such as the war stories they discuss at the coffee machine, as ‘‘informal’’ communication but, rather, to recognize that they may be part of a larger structural process. This analysis suggests that the collective understandings of frontline staff in human service organizations are structural; they help to determine how staff respond to problems and address clients’ concerns. This conclusion is consistent with the analysis of Peter Senge, the MIT professor and author of the national best-seller, The Fifth Discipline: The Art and Practice of the Learning Organization, that has dramatically influenced private sector management. Senge writes: ‘‘[Oftentimes] the structure causes the behavior [of individuals]. This distinction is important because seeing only individual actions and missing the structure underlying the actions . . . lies at the root of our powerlessness in complex situations.’’ 26 Senge’s work emphasizes the significance of managers’ and workers’ beliefs, what he calls ‘‘mental models,’’ in constructing the structure within which action occurs. In human service organizations, in which service technology is more uncertainly related to agency outputs, such a process is even more significant than in private for-profit industries.27 In human service organizations, social structures provide a rationale that enables staff to carry out their daily tasks and respond to requests from colleagues in the environment in a particular way. Further, this analysis suggests a new reason behind staff resistance to the imposition of new ‘‘collaborative’’ service efforts by local managers. If these initiatives are to be undertaken with organizations that street level workers perceive as hostile, ineffective, or inefficient, the social structures will help to resist the collaborative effort. The social structures, collectively developed and sustained by frontline workers, pro- 336 Social Service Review vide sufficient justification for the resistance to ‘‘idealistic’’ management entreaties. The final implication of this analysis for improving the implementation of human service collaborative initiatives, then, is to direct managers to alternative types of interventions. Drawing on the concepts of systems thinking and learning organizations, managers must employ new skills to provide a more complete analysis of the causes of interorganizational barriers and identify the leverages for change within the system.28 Managers must train themselves to look beyond daily events, external forces, or illusions of individual impact to understand problems and, instead, examine their larger structural context. In Senge’s model of a learning organization, managers can learn to do structural analysis of organizational problems. By seeing the interrelations of events—how relating stories of charged management meetings with a collaborative partner fuels resistance—managers can begin to identify ways to use the existing frontline structures to achieve policy ends. Rather than imposing collaborative mandates onto frontline staff, this framework pushes managers to find the points of leverage that exist for change within the system. Systems thinking stresses that leverage is found at small, but well focused points in the process. Since managers have little control over staff experience or client stories, they must look for other potential points of intervention in improving human service collaboratives. Staff meetings that elicit the prevailing beliefs about a potential collaborative organization would introduce a process whereby managers could question how past relations could limit future collaboration. Sponsoring staff forums that allowed workers to formally share experiences with each other would both allow managers to learn about prevailing beliefs and to insert alternative interpretations of events. Small, regular unit meetings could allow middle managers to exert their influence over how workers understood a charged interaction with a frontline worker from a partner organization. These, and other practices that flow out of a learning organization approach, would likely mediate the role of preexisting frontline social structures in impeding the collaborative process. Clearly, changing the social structures within which frontline staff operate requires considerable time and resources. Managers, themselves, must be committed to the collaborative goal. Drawing on the concept of a learning organization, they must constantly reflect on how their own actions detract from or reinforce the importance and legitimacy of that goal. Such an awareness begins with the recognition that within an organizational system, many seemingly unrelated events are interconnected. Over time, they create the structures within which people act and implement social programs. Although interagency collaboration is a human endeavor, it is neither caused nor inhibited by individuals. This analysis examines the social structures that develop from the nature of Human Service Collaboration 337 day-to-day experience on the front lines. It is only by attending to these structures that the managers in human service organizations have a chance at improving service delivery for needy populations. Notes An earlier version of this article was delivered at the 1998 annual meeting of the American Political Science Association, Boston, September 3 – 6, 1998. The author thanks Martha Feldman and Ann Lin for their feedback on the ideas presented here. 1. Charles Bruner and Larry Parachini, Building Community: Exploring New Relationships among Service Systems Reform, Community Organizing, and Community Economic Development (Washington, D.C.: Institute for Educational Leadership, n.d.); Robert L. Crowson and William Lowe Boyd, ‘‘Coordinated Services for Children: Designing Arks for Storms and Seas Unknown,’’ American Journal of Education 101 (February 1993): 140 –79; U.S. General Accounting Office, Integrating Human Services: Linking At-Risk Families with Services More Successful than System Reform Efforts (Washington, D.C.: Government Printing Office, 1992); Sharon Kagan, Integrating Services for Children and Families: Understanding the Past to Shape the Future (New Haven, Conn.: Yale University Press, 1994); Alfred J. Kahn and Shelia Kamerman, ‘‘Integrating Services Integration: An Overview of Initiatives, Issues, and Possibilities,’’ discussion paper (National Center for Children in Poverty, Columbia University, New York, 1992, photocopied); Paul Mattessich and Barbara Monsey, Collaboration—What Makes It Work? A Review of Research Literature on Factors Influencing Successful Collaboration (St. Paul, Minn.: Amherst H. Wilder Foundation, 1992). 2. See Kagan; Ellen Konrad, ‘‘A Multidimensional Framework for Conceptualizing Human Services Integration Initiatives,’’ New Directions for Evaluation 69 (Spring 1996): 5–19; John O’Looney, ‘‘Modeling Collaboration and Social Services Integration: A Single State’s Experience with Developmental and Non-developmental Models,’’ Administration in Social Work 18, no. 1 (1994): 61– 86; Jane Waldfogel, ‘‘The New Wave of Service Integration,’’Social Service Review 71, no. 3 (1997): 463 – 84. 3. Eugene Bardach, ‘‘Turf Barriers to Interagency Collaboration,’’ in The State of Public Management, ed. Donald Kettl and H. Brinton Milward (Baltimore: Johns Hopkins University Press, 1996); Pamela Holcomb, Kristin Seefeldt, and John Trutko, One Stop Shopping Service Integration: Major Dimensions, Key Characteristics and Impediments to Implementation. Final Report to U.S. Department of Labor, Employment, and Training Administration (Washington, D.C.: Urban Institute, September, 1993). 4. Anita Farel and Kathleen Rounds, ‘‘Perceptions about the Implementation of a Statewide Service Coordination Program for Young Children: Importance of Organized Context,’’Families in Society: The Journal of Contemporary Human Services (November-December 1998), pp. 606 –13; Marcia Meyers, ‘‘Organizational Factors in the Integration of Services for Children,’’ Social Service Review 67, no. 7 (1993): 547–75. 5. Catherine Alter and Jerald Hage, Organizations Working Together (Newbury Park, Calif.: Sage, 1993); Keith G. Provan and H. Brinton Milward, ‘‘A Preliminary Theory of Interorganizational Effectiveness: A Comparative Study of Four Community Mental Health Systems,’’ Administrative Science Quarterly 40, no. 1 (1995): 1–33, and ‘‘Institutional-Level Norms and Organizational Involvement in a Service-Implementation Network,’’ Journal of Public Administration Research and Theory 1, no. 1 (1991): 391– 417. For detailed reviews of the organizational network literature, see Christine Oliver, ‘‘Determinants of Interorganizational Relationships: Integration and Future Directions,’’ Academy of Management Review 15, no. 2 (1990): 241– 65; and David Whetten, ‘‘Interorganizational Relations: A Review of the Field,’’ Journal of Higher Education 52, no. 1 (1981): 1–28. 6. Janet Weiss, ‘‘Substance vs. Symbol in Administrative Reform: The Case of Human Service Coordination,’’ Policy Analysis (1981), pp. 21– 45. 7. Evelyn Brodkin, ‘‘Inside the Welfare Contract: Discretion and Accountability in State Welfare Administration,’’ Social Service Review 71, no. 1 (March 1997), 1–33; Yeheskel Hasenfeld, ‘‘The Nature of Human Service Organizations,’’ in Human Services as Complex Organizations, ed. Yeheskel Hasenfeld (Newbury Park, Calif.: Sage, 1992): pp. 3 –23; Michael Lipsky, ‘‘Bureaucratic Disentitlement in Social Welfare Programs,’’ Social Service Review 58, 338 Social Service Review no. 1 (March 1984), 3 –27; Michael Lipsky, Street-Level Bureaucracy: Dilemmas of the Individual in Public Services (New York: Russell Sage Foundation, 1980); Marcia Meyers, Bonnie Glaser, and Karin MacDonald, ‘‘On the Front Lines of Welfare Delivery: Are Workers Implementing Policy Reforms?’’ Journal of Policy Analysis and Management 17, no. 1 (Winter 1998): 1–22. 8. Anthony Giddens, The Constitution of Society: Outline of a Theory of Structuration (Berkeley and Los Angeles: University of California Press, 1984); William Sewell, ‘‘A Theory of Structure: Duality, Agency, and Transformation,’’ American Journal of Sociology 98, no. 1 ( July 1992): 1–29. 9. The names of the counties, organizations, and individuals have been changed to preserve their anonymity. 10. One Work First provider was examined in Dutchess County, two in Dunn County. In each county, these organizations constitute the ‘‘implementation structures’’ of welfare service delivery. For more explanation of this construct, see Benny Hjern and David Porter, ‘‘Implementation Structures: A New Unit of Administrative Analysis,’’ Organization Studies 2, no. 3 (1981): 211–27; Provan and Milward, ‘‘Institutional-Level Norms,’’ pp. 391-417. 11. Barney G. Glaser and Anselm L. Strauss, The Discovery of Grounded Theory: Strategies for Qualitative Research (New York: Aldine De Gruyter, 1967); Anselm Strauss and Julie Corbin, ‘‘Grounded Theory Methodology: An Overview,’’ Handbook of Qualitative Research, ed. Norman Denzin and Yvonna Lincoln (Thousand Oaks, Calif.: Sage, 1994). 12. As other researchers have noted, ‘‘triangulating’’ qualitative methods helps to tap different dimensions of human experience. In-depth interviews and focus groups provide a concise self-report of how people within the organization selfconsciously understand their world at one moment in time. Participant observation provides a method for understanding how people exist within a context and for documenting the interactions, practices, and attitudes that are unlikely to be recounted in formal interviews. For more details of this methodology, see Debra Meyerson, ‘‘Uncovering Socially Undesirable Emotions,’’ American Behavioral Scientist 33, no. 3 (1990): 296 –307; and Michael Quinn Patton, Qualitative Evaluation and Research Methods, 2d ed. (Newbury Park, Calif.: Sage, 1990). 13. These methods are elaborated in Robert Emerson, Rachel Fretz, and Linda Shaw, Writing Ethnographic Fieldnotes (Chicago: University of Chicago Press, 1995). 14. Lyn Richards and Tom Richards, ‘‘From Filing Cabinet to Computer,’’ in Analyzing Qualitative Data, ed. A. Bryman and R. Burgess (London: Routledge, 1994); Mike Hannibal and Celia Gahan, Doing Qualitative Research Using QSR Nudist (Thousand Oaks, Calif.: Sage, 1998). 15. For a general discussion of this approach to welfare-to-work and how it is being implemented, see Amy Browne, Work First: How to Implement an Employment-Focused Approach to Welfare Reform (New York: Manpower Demonstration Research Corporation, 1997). For a more detailed examination of the implementation of this welfare-to-work program in Michigan, see Kristin Seefeldt, Jodi Sandfort, and Sandra Danziger, Moving towards a Vision of Family Independence: Local Manager’s Views of Michigan’s Welfare Reforms (Ann Arbor: Michigan Program on Poverty and Social Welfare Policy, University of Michigan, 1998). 16. In addition to welfare recipients, the Commission administers training programs for parolees, displaced homemakers, troubled youth, dislocated workers, the blind, and the physically disabled, as well as other initiatives to improve economic conditions within the state. 17. For a description of a range of the services provided under Work First in Michigan, see Seefeldt, Sandfort, and Danziger, pp. 20 – 40. 18. This compares with 50 percent throughout the state. 19. Although the state publishes extensive data reflecting program operation, information about the rate of sanctions is not published or made available to the public. 20. This experience is so widely recognized as part of frontline experience that Dr. Robert Wertkin included such a story, empathizing with the challenges of frontline work, in a 1995 keynote address to the National Eligibility Workers Association (photocopied). 21. Steven Rathgeb Smith and Michael Lipsky, Nonprofits for Hire: The Welfare State in the Age of Contracting (Cambridge, Mass.: Harvard University Press, 1993). 22. Peter K. Manning. ‘‘Organizational Work: Structuration of Environments,’’ The British Journal of Sociology 33, no. 1 (1982): 122. 23. Pierre Bourdieu, The Logic of Practice (Stanford, Calif.: Stanford University Press, 1990); Richard Harvey Brown, ‘‘Bureaucracy as Praxis: Phenomenology of Formal Organi- Human Service Collaboration 339 zations,’’ Administrative Science Quarterly 23 (1978): 365– 82; Giddens, Constitution of Society (n. 8 above), and in The Giddens Reader, ed. Philip Cassell (Stanford, Calif.: Stanford University Press, 1993); Martha S. Feldman, ‘‘Managing Change in Organizational Routines,’’ working paper (University of Michigan, School of Public Policy, June, 1998); Wanda Orlikowski, ‘‘The Duality of Technology: Rethinking the Concept of Technology in Organizations,’’ Organizational Science 3, no. 3 (August 1992), pp. 399 – 427; Sewell (n. 8 above). 24. This term comes from the work of Giddens, The Constitution of Society. 25. Manning, p. 122. 26. Peter Senge, The Fifth Discipline: The Art and Practice of the Learning Organization (New York: Doubleday, 1990), p. 77. 27. Hasenfeld (n. 7 above). 28. In addition to Senge, many other organizational development scholars and practitioners have been wrestling with these concepts. See Edgar Schein, ‘‘Three Cultures of Management: The Key to Organizational Learning in the Twenty-first Century’’ (MIT: http: //learning.mit.edu/res/wp/10011.html, 1996); Chris Argyris and Donald Schon, Organizational Learning II: Theory, Method and Practice (Reading, Mass.: Addison-Wesley, 1996).