Introduction 

Central Research Questions 

Special Methodological Considerations 

Results 

Conclusions and Future Research 

References 

Increasingly, this research has identified the need to explore theoretical issues that emerge as various forms of collaboration between schools and social service institutions take shape. In earlier work, Crowson and Boyd (1993), for example, drew on field studies of selected case sites throughout the U.S. in order to conceptualize collaboration as a continuum, ranging from limited, technical arrangements for resource sharing, on the one hand, to full-fledged joint enterprises in which shared languages and meanings are created through common activities, understandings of problems and boundary-spanning roles, on the other. The notion of a continuum suggests that the structures, attitudes, and activities that foster or impede collaboration are of particular interest, and that identifying patterns of such factors may help us understand where and why these initiatives succeed. 

Similarly, research has increasingly identified both the need for and the dearth of knowledge about the role of site-level leadership in these initiatives. As integrated services have become commonplace, an opportunity has been created to examine (via data collection from role incumbents) the ways this growing cadre of individuals are defining the state of practice. Understanding and identifying the issues and activities they regularly encounter would be helpful in the planning of such programs as well as in the preparation of collaborative leadership. 

This paper describes the technical aspects and preliminary results of a study that explored the role of the program director and the program director's perceptions of critical aspects of collaborative services initiatives. In the following sections, we describe the central research questions of the study, along with general methodological considerations; special methodological considerations involved in conducting the study, including the development of an appropriate sampling frame and an instrument that reflected our central questions; procedures for data collection, including the statistics on the final sample used for analysis; and the most salient findings in each of our areas of study. 

Several concerns animated this project. Our review of the extant literature, as well as our field research, suggested tentative hypotheses about the characteristics of collaborative services programs, many of which are detailed below (see Crowson and Boyd, 1993). At the same time, we were cognizant that knowledge about the state of practice is still relatively incomplete. Thus, the purpose of this study is twofold-descriptive and empirical. First, we attempt to get a national picture of the growing field of collaborative services and document the variation that exists across basic characteristics of these programs. Second, we test some of the hypotheses about the process of collaboration that have been generated through earlier work.

At the outset of the project, we made two delimiting decisions. First, we decided to focus as much as possible on programs that contained a school component, be they "school-based" (located in a school) or "school-linked" (connected in some way to, but not located within, a school). Second, we decided to examine as many programs as possible, thus making a mail survey the most feasible and logical methodological choice for collecting data. This survey instrument is described in detail below.

Prior to instrument development, however, we met early in the course of the study to identify areas of interest that were most compelling given a national sample of collaborative services programs. Four large areas of study were designated. First, we decided that general information about programs was needed to obtain a national "snapshot" of the field as it currently stands. Second, we formulated detailed questions about the process of collaboration, including the beliefs, attitudes, and values of professional participants, as well as specific structures that aid or hinder the collaborative process. Third, we sought to characterize the relationships between the collaborative programs and the communities they are intended to serve. Finally, we had many questions about the ways directors of school-linked programs construct their worklives, including the range of activities in which they engage regularly and the professional personnel with whom they interact. Each of these four areas is described in greater detail below. A conceptual map of the survey instrument designed to capture these elements may be found in Table 3.1. 

General Information about the Program 

One of our primary goals was to develop characterizations of the nature and scope of the services individual programs deliver. To our knowledge, no general overview/assessment of the scope of national programs or typology of collaborative school programs based on systematic data collection has been developed as yet. Thus, we asked program directors to report the types of services provided by their programs. We were also interested in the average scope and size of programs, including the number of schools involved, sites involved (presuming that some programs may deliver services in nonschool as well as school settings), and full- and part-time program employees. Additionally, we sought to collect basic information about the length of time programs had been in existence and general information about their locations.

The Process of Collaboration

As stated earlier, we view collaboration as occurring on a continuum. One of our primary goals in designing this project was to apply this conceptualization to a national sample of programs, with particular emphasis on three key questions: Do programs differ in the amount of collaborative processes they exhibit? On which dimensions do programs differ most, and on which are they most similar? Finally, are there aspects of program structure or location that help to predict the degree of collaboration that is manifested?

Our conceptualization of collaborative processes has two components. The first concerns the beliefs, attitudes, and values of the professionals in the program; the second details more specifically structures and practices that may characterize or affect collaboration.

Beliefs, Attitudes, and Values about Collaboration

Development of a sense of trust. Without trust, little to no collaboration is possible. Thus, we were interested in determining the extent to which working professionals perceive that such trust exists.

Shared sense of problem. Research indicates that collaboration can lead to a redefinition of client problems and a shared understanding of how they might be addressed. Thus, for a given program, we wanted to know the extent to which professionals agreed on the definition of client problems, on the procedures for dealing with clients, and on the ways these problems should be addressed.

Development of coordinated processes. When collaboration occurs, we expect that ways of working together become more routinized as client issues are increasingly handled successfully. Thus, we would expect coordinating processes to emerge that coalesce professionals around client issues and address recurring concerns. We were interested in both the changes that have been made and the ways professionals believed that the program had changed their habits of professional interaction. Thus, we wanted to know if the ways schools and agencies work together had changed, including the manner in which client problems are identified and acted upon, and the methods through which activities are coordinated. Determining whether professionals perceived that a clear sense of shared responsibility had emerged was also a key concern.

Existence of a shared language. Perhaps the most extreme example of collaboration occurs when professionals who have been socialized differently and taught to work independently from one another develop a shared language that enables them to communicate effectively about their work. Previous research has indicated that, though essential for effective collaboration, this level of communication is often difficult to establish. This being the case, we were interested in the ease with which professionals communicated with one another and their beliefs about the degree to which they had learned to share perspectives.

Development of new attitudes towards collaboration. Professionals who have been taught to work independently may bring to collaborative efforts attitudes that range from enthusiasm to suspicion. We deemed it essential to determine the degree to which participants perceived that their fundamental attitudes and beliefs about collaboration had changed.

Structures that Affect Collaboration

Time. We were interested in several aspects of the dimension of time with regard to collaborative services programs. For example, given the rapid proliferation of programs, we wanted to know if professionals believed that the planning and implementation schedules of their respective programs facilitated success. Specifically, we sought to determine how time was allocated internally, especially with respect to the flexibility with which professionals' days could be scheduled for joint activities and shared discussion.

Incentives. Many programs have been constructed with financial incentives that are designed to promote participation by cooperating professionals. At the same time, some research has suggested that the incentive structures of collaborating agencies may be at odds with professionals' participation in the project. Thus, we were interested in the use of incentives and their role in the promotion of program success.

Staff. Given that programs are created in a decentralized manner, they can vary tremendously in the staff positions they include. We were interested in determining what kinds of positions characterized different programs and if staff resources were deemed adequate. We also wanted to know how issues of staff training and cooperation affected collaboration.

Written agreement. The value of written agreements in promoting collaboration and eliminating confusion about shared responsibilities and resources has been well documented in the literature. We sought to gauge the extent to which programs believed such arrangements are useful.

Space. As was the case with time, we wanted to characterize the physical space that programs reported was required for their specific needs and if their current accommodations were adequate for program activities.

Other resources. Given that many programs are entrepreneurial in nature, we hoped to determine other resources, including the ability to attend conferences, resources for publicity and duplication materials, and other staff and program needs, that were employed or deemed necessary.

Assistance for problem solving. Given that multiple actors are engaged in the development, funding, and implementation of collaborative services programs, we were interested in delineating what resources programs used when they needed assistance and which were most helpful in solving problems.

Technology. Many programs are located on multiple sites that may make face-to-face communication difficult. We were interested in the extent to which programs used technological resources, such as computer mail, to promote communication.

Existing procedures. Much of the literature has indicated that one of the most difficult aspects of collaboration is modifying existing procedures in schools and agencies that militate against joint activities. Among the most problematic of these types of challenges are the procedures designed to protect client confidentiality. Thus, we were interested in the extent to which such confidentiality procedures were perceived as barriers to collaboration.

Relationship with the Community

The aim of many collaborative services programs, especially those linked with schools, is to develop schools' connections with the many communities that so deeply affect students. The rhetoric of these types of programs often speaks to creating more broadly based connections between the school and a range of community resources and personnel. Thus, we were interested in the following dimensions of the relationship between the program and the community it serves.

Quality of relationship. Does the program have a positive relationship with the surrounding community? Has the quality of relationship changed as a result of program activities? Has the program created demands from the community it cannot meet?

 Degree of community involvement. To what extent does the program receive support from the community? To what extent is the community involved in interactions with program staff?

Outreach and awareness. To what extent do the program staff perceive themselves as reaching out to the community? To what extent are they aware of community concerns?

The Role of the Program Director

Much as the educational literature has delineated the "average" day of a school principal, we were interested in documenting the kinds of activities that are regularly undertaken by individuals who head collaborative services programs. Using the experiences of program directors, we created a list of activities that included such items as generating paperwork and reports, planning activities, meeting with staff, securing resources, and dealing with various aspects of the institutionalized environments of the program. Given these options, we wanted to know how program directors allocated their time. In addition, we wanted to know which activities they perceived as most critical to the functioning of their respective programs.

In the course of undertaking this study, we encountered special problems directly related to the emergent nature of the field. It is helpful to delineate briefly these obstacles insofar as they transmit some sense of the natural variation that abounds among collaborative services in the U.S. and, more importantly, because they raise a number of special concerns for researchers.

• There is little agreement in the field about the terminology that should be used for the phenomenon we have termed "integrated services." From a technical standpoint, when identifying and addressing the desired respondents in the instrument and cover documents, the variant terminology required the use of multiple terms with explanatory definitions to ensure that the respondents fit our target sample. Although we decided to target only programs that provided direct services to children and families, the names of programs often told us little about their nature. Many programs appear to have "nonstandard" nomenclature; specifically, there appears to be an unusual penchant in the field for catchy acronyms, such as Project REACH or SAVE. Such acronyms provide little information in the absence of an explanatory statement, and we found that it was not always possible to determine the nature of the program without contacting it for additional information.
• Identifying and tracking programs are complicated by both the decentralized nature of education as well as the highly decentralized character of the oversight and funding for these efforts. Funds for collaborative services have been provided from both public and private sources, including foundations (e.g., Kellogg, Annie E. Casey) states (New Jersey, California, and Kentucky are prominent examples); collaborations among districts, or local education authorities (such as New York's Board of Cooperative Educational Services) as well as within districts, often split among various administrative units.
• Encouragement for such efforts, even at similar levels, can originate from widely differing sources. State departments of education are prominent actors in some states. In others, however, the locale for the encouragement and organization of programs may be elsewhere. In Kentucky, for example, the Cabinet on Human Resources provides oversight, while in New York a major role is played by the legislative committee of the State Senate that is charged with the responsibility for education issues. National organizations designed to document and bring coherence to the field are still in embryonic stages. As yet, no complete national census of such programs exists or is compiled on a regular basis by any federal agency.
• The nature of funding sources may be related to prominence in the news media and the frequency with which these programs are found in the literature. Our sample was created by a national center that compiled large amounts of printed material (both produced by the programs and via media reporting their activities). It is likely that small, "revenue-neutral," rearrangements of funds might have been neglected as these programs were identified. Thus, although our sample was compiled from available literature by a national clearinghouse, large or well-funded programs may be overrepresented.
• The organizational and physical locations of collaborative efforts vary, even within categories such as "school-based" or "school-linked." The range of possibilities for location includes small, site-based programs that are housed informally in schools to larger efforts that have their own space both on-site (in schools), and off-site (in the surrounding community). Within districts, programs may have "homes" in a variety of administrative units, ranging from general education programs, special education departments, pupil personnel services, or units created for the express purposes of housing the program. The "grassroots" nature of many programs virtually ensures that the natural variation with respect to locus and administrative structures will be great.
• Both mortality and mutation are common events in these organizations. Our data show that many organizations identified only two years ago as delivering integrated services have either "gone out of business" or changed beyond recognition. New programs have undoubtedly also developed during this time. Thus, the state of the field changes dramatically as funding sources are developed or eliminated.

Development of an Appropriate Sampling Frame

The study's first step involved identifying a wide range of programs throughout the U.S. that are funded through either public or private agencies and link some set of social services for children to schools and their surrounding communities. We began with the understanding that no such national list of programs exists. Faced with the onerous task of creating such a list, the need for collaboration became readily apparent. Early on in our study we had conversations with staff at the National Center for Service Integration in Washington, D.C. As part of the Center's mission is to document collaborative efforts throughout the country, they were engaged in creating data files on program efforts, a project on which they were working in conjunction with the National Center on Children in Poverty, located at Columbia University. Together, staff from these two agencies had constructed hard-copy files of information that catalogued efforts throughout the nation. These data included stories in the print media, annual reports of funding agencies, research and evaluation reports, papers disseminated at national conferences, bulletins describing pilot programs, and records of personal communications. Data were organized by state as well as by funders and scope of programs. Although there had been some efforts to create a computerized list of these data, virtually all of this information was still in its original form.

Building on the substantial work already accomplished by the two above-mentioned organizations, we created a computerized database into which we entered basic information about each program. Using the Centers' files and adding programs as we became aware of them, a database of more than 400 collaborative service organizations, located in 40 states and two U.S. territories, was created; eventually these data were used to generate mailing labels.

It should be noted that the information provided did not always enable us to determine if given organizations provided direct services to children or if they simply had school components. We were hesitant to eliminate programs from our initial mailing, however, since we suspected that some organizations identified might have subunits that provided direct services that would be of interest. We responded to these issues in the following ways. First, we designed a process through which organizations and programs in our initial mailing were permitted to de-select themselves from the target sample by returning a form on which they indicated that they did not deliver direct services. Second, we used respondents' information about the school components of their programs; only those that reported they included schools were considered as part of the final analytic sample. Finally, our cover letter instructed the addressee to forward the survey to another respondent in the organization if appropriate.

Ideally, we would have liked to begin with a listing of all programs and to create from this comprehensive list a representative sample, assuming of course, that an appropriate working definition of "representative" could be determined in such an emergent and heterogeneous field. Our database is but a first step in the creation of such a universal listing; thus, we do not make the claims for generalization commonly made in such representative samples, nor are we inclined (at this point) to use techniques of statistical inference that rely heavily on such a representative sample. We do know, however, that in the sampling frame provided by our work with the collaborative research centers are represented examples of most of the major foundation-funded collaborative programs, as well as examples of large state-funded efforts and many "grassroots" efforts. Despite out best efforts to augment the information with additional programs where possible, we have doubtless overlooked several important initiatives.

We are realistic about the nature of the sample and acknowledge that the collaborative nature of the process leaves room for omissions. But given both the emergent nature and the heterogeneity of the field, we are confident that our sampling frame represents a good overview of national programs and includes a wide variety of different types of initiatives. Moreover, we know of no more comprehensive listing of programs.

We approach these data, then, as a preliminary "snapshot" of a field that was creating itself as we documented it. In so doing, we attempt to test the findings of case-based literature against a broader sample drawn from a heterogeneous field as it existed in the last months of 1994.

Focus on Program Director as Respondent

Although the creation of the database was time consuming, it provided information that was useful in making our decision regarding the appropriate informants for our study. Although programs differed widely on many dimensions, virtually every entry had a name identified as the "director," "coordinator," or "contact" for the program. It appeared that although titles differed, most programs had an analogous position for the "head" of the program who could be identified relatively simply. The same was far from true, however, for other staff in the program, where it seemed that both the nature of the positions included and the titles by which they were identified varied greatly. Although we initially wanted to contact multiple respondents for each program, we were faced with the substantial task of determining who the appropriate parallel respondents might be at the site level and how they might be contacted. Given our resources, we decided that a more judicious decision was to gear our efforts toward obtaining a reasonable rate of response from program directors.

This choice clearly limited our ability to examine multiple perspectives within programs and constrained our ability to collect information about roles other than that of the program director. Such a decision is not unprecedented, however; the literature on schools and principal leadership contains many examples of studies based on surveys in which single respondents report on complex organizations. Furthermore, our study is hardly the "last word" in the field, and we note that it is both derived from earlier case studies and likely to generate leads for further study at selected sites. In addition, we envision that at some future time additional study of sites that are now known to be established and representative of the field may be possible.

Development of an Appropriate Survey Instrument

Finally, we needed to develop an appropriate instrument that adequately captured our research questions. In this regard, our first task was to determine if any instrument existed that could be used or modified for our purposes. While a search of the literature showed that no likely candidate would suit our purposes entirely, we did identify portions of existing instruments that could be altered to suit our needs. Our census of program services, for example, was adapted from a similar question on a California survey. Similarly, several of the items used to assess community relations were modified versions of items used in a study of Chicago's decentralized schools. We also borrowed formats and partial wording for items assessing attitudes, values, and beliefs from several federal education surveys, including High School and Beyond, the Schools and Staffing Survey, and the National Educational Longitudinal Studies. Where needed, new items were developed.

The items identified were keyed onto the conceptual map of our survey (Table 3.1), and several drafts of the instrument were prepared in consultation with program staff at field sites. Instruments were reviewed for purposes of content validity, face validity, and clarity of wording. A near final draft of the instrument was piloted with program staff and likely respondents and revisions were made as necessary.

One of the concerns that emerged in the pilot stage was the reluctance of program directors to complete items through which they believed they or their programs might be identified, despite assurances of confidentiality. We responded to this issue by minimizing all demographic information about the respondent. Similarly, we employed the standard practice of placing the most sensitive items at the end of the instrument, which encouraged respondents to return nearly completed surveys even if the last items had been left blank.

Formatted using a machine-scannable program, the instrument was professionally printed and disseminated with a cover letter from the National Center on Education in the Inner Cities. A slightly modified version of the survey was developed for follow-up purposes which encouraged respondents to return a form through which they could de-select themselves from the final sample if their program did not deliver direct services. In addition, a separate sheet was included on which respondents were requested to list their official title, both protecting their confidentiality and allowing us to eliminate any respondents whose titles seemed to indicate that they were clearly inappropriate.

The final version of the instrument contained more than 100 questions (in a variety of response formats) that created multiple indicators of the dimensions outlined under each of the key areas of interest.
Data Collection: Creation of the Analytic Sample

Table 3.1 summarizes the three-stage process used to identify respondents in the target population, that is, directors of collaborative programs providing direct services to children. As noted above, we began with a national list in excess of 400 organizations identified as potential providers.

In some cases, however, the data available did not allow us to determine if the organization should be included in our target population. A Program Director's Survey was mailed during the Fall of 1994 to 413 organizations identified in the data potential providers. A duplicate, follow-up survey was sent in late 1994 to all programs that had not responded.

Thirty-four surveys were returned by the U.S. Post Office as un-deliverable, leaving a total of 379 surveys that were assumed to have been delivered to the addressees. Sixty respondents eliminated themselves by returning the enclosed form. Of the 319 surveys remaining, 163 were returned completed and 156 never responded. A total of 223 of the original 413 surveys (or 54%) responded with either completed surveys or a form indicating they were not appropriate information providers for our study.

In the second stage, we examined the completed surveys to determine if respondents should be deleted because they were not appropriate providers. Comments made by two respondents indicated they were confused about their status; these surveys were deleted from the final sample, as was a duplicate survey. A total of 160 usable surveys were returned, representing approximately 50% of the target sample of 319 programs. We believe this is a conservative estimate of the rate of response, given that it assumes that all nonrespondents were appropriately identified as part of the target sample.

Finally, the data were inspected to determine which programs reported that they were either located in a school or linked to a school. It was possible for programs to answer both questions in the affirmative, given that, in theory, a program might both be located physically in one school and linked to another school site (in fact, some 49 programs answered in the affirmative to both questions). Only those programs that responded in the affirmative to one of these questions were included in the final analytic sample of 137 programs.

Below we present some of the key findings in each of the areas of concern outlined in Table 3.1, using descriptive measures that represent the central trends reported in these data. We will address in greater detail some of the policy implications that can be drawn from these findings in a separate report (Driscoll, Boyd, & Crowson, in preparation).

General Information about Programs

Table 3.2 reports on responses to items on which program directors were asked to check all services delivered by their programs. Recall that only programs that reported having either school-based or school-linked components were included in the final sample of 137 respondents.

By far, respondents indicated that the most common service provided was "parenting education," with more than 81% of the programs reporting that they deliver this service. "Family support and advocacy" (68.6%) and "other health education" (67.2%) ranked second and third. At least half of the programs reported offering some basic health services, including screenings and immunizations; 56% reported providing some type of individual therapy.

With regard to patternistic traits of responding programs, a "pedagogical" emphasis emerged. In addition to the nearly 82% of the programs that reported providing parent education, 46% reported offering tutoring, about a third cited adult and literacy education, 39% academic counseling, and 67.2% "other health education" services. Perhaps "service" to parents and families is still heavily imbued with the notion that providing better information to a presumably information-needy clientele is important, and this orientation is fueled by the belief that informed people will help themselves.

While 50% of the programs reported that they engage in school system advocacy, it is noteworthy that only a third of the programs reported providing academic services such as adult education and literacy education. It would appear that these programs concentrate on family and health services, despite their reports of being linked with school settings. Table 3.2 also shows that more than half of the projects did not indicate provision of reproductive counseling, a potentially controversial service. Additionally, less than a third reported offering pre/postnatal care.

It is somewhat surprising that a third of the programs (32.8%) reported engaging in employment-related services, a feature not highlighted in the existing literature on service integration. Moreover, despite what the literature would suggest, direct-rather than referral-services, seem to be the norm. The case management that is much touted in the literature did not emerge as one of the most well-represented components in our sample of programs.

It is clear that many projects provide assistance well beyond what can be characterized as "information-as-service." These vital services include basic needs provisions such as food (51.8%) and child care (53.3%), as well as quality-of-life services such as complete medical physicals (50.4%) and health screenings (59.1%). It is interesting that nearly 60% of the projects extend beyond information provision to "therapy." This may imply that, in terms of a continuum of collaborative activity, these programs are not collaborative in the sense that they offer integrated services, but in that the basic philosophies of various helping professions are represented-from the information-provision of the educator, to the family-service and family-counseling of the social worker, to the wellness concerns of the medical professional.

Although Table 3.2 provides an adequate total picture of the services provided by the responding programs, it tells us little about the extent to which there are differing patterns of service delivery or how the programs are arrayed with respect to the number and intensity of services that are provided. Table 3.3 shows the mean number of services offered in each category. For example, out of a possible six academic services listed, the average program offered slightly more than two, while it offered more than four health services. We created the variable "Sum of Services" to capture the intensity of services provided by each program; this represents a sum of all services offered by each program across all categories. As Table 3.3 indicates, no program offered fewer than two services, and some offered as many as 35. The mean number of services offered per program was 14.23, although the data display substantial variation from this average.

Table 3.4 presents the mean number of services provided by programs further categorized by size (number of schools in the program) and location (urban, suburban, and rural). Note that this table reports the means for 129 programs with complete data on their urbanicity; 109 of these programs also provided data about their size. True to the caveats expressed by our focus groups, program directors with otherwise complete surveys often did not respond to questions the answers to which that they believed might identify their programs, resulting in a relatively higher level of missing data on demographic items as compared with others.

It appears from the data presented in Table 3.4 that in urban settings, the smallest programs (measured by the number of sites served) provide the highest number of services. In suburban and rural settings, however, slightly larger programs (serving 2-4 schools) provide the highest intensity of services. Table 3.4 also shows that the vast majority of the programs in our sample are located in urban settings (87 out of 129 reporting location data), with rural programs outnumbering suburban programs 26 to 16. Thus, differences in means across these subsamples must be weighed against the fact that the sample sizes are quite unequal.

The Process of Collaboration

Our analyses of the process of collaboration took several forms. In the initial analyses, we examined some of the structures and systems that have been reported in the literature as supporting collaboration among services, as well as some of the barriers to collaboration. We formulated two types of questions to elicit information from program directors about these issues. In one set of questions, program directors were asked about the degree to which eight factors contributed to the success of the program. In another set of questions, they were asked to identify the extent to which other factors were barriers to collaboration (focus groups indicated that directors were substantially more comfortable answering "success" questions than they were identifying problems; we carefully balanced these sets of questions to account for this tendency). Thus, while the literature reports many more issues that are likely to be barriers to collaboration than supports for success, our questions reflect an even-handed approach with exactly the same number of items in both our "problem" and "success" scales. In addition, we obtained other information about these issues via Likert-scale items, in which the "Strongly Disagree" to "Strongly Agree" format appeared to mitigate the "negativity" program directors associated with our lists of problematic factors.

Table 3.5 reports the means and standard deviations of the items that asked directors about factors contributing to the success of their programs. The response scale ranged from 1 (factor has not contributed to success) to 5 (factor has contributed to success a great deal). Directors reported that having space dedicated to program staff activities and being able to schedule time in ways that permitted joint meetings contributed most to program success (mean scores of 4.14 and 4.12, respectively). Technological links and financial incentives were reported least frequently as contributors to success. It is interesting to note that the existence of a written agreement (mean score 3.67) appeared to be only moderately associated with perceived success, despite indications in earlier literature that such agreements are very helpful. Similarly, program directors do not list incentives that promote participation on the part of service professionals as one of the primary factors in program success, even though recent literature has suggested that such incentive structures may be critical in supporting collaborations. It may be that from the site-level perspective of the program director, pragmatic resources such as adequate space and flexible scheduling appear to be much more crucial for effective functioning.

Table 3.6 reports the mean scores and standard deviations for the eight items in which program directors were asked the extent to which various factors were problematic to their programs. Once again, the response scale ranged from 1 (not a problem) to 5 (is a serious problem). Note that mean scores are all in the low range, that is, the average reported incidence of these problems is nonexistent to infrequent. Even so, the relatively low incidence of some factors is worth noting; for example, few directors report that the procedures designed to protect confidentiality of clients impede their work, although much of the literature has reported that such procedures can be barriers to collaboration. Least problematic is the responsiveness of clients (mean score of 1.81). Finding the time to discuss and plan with one another is cited as the most common problem; taken together with the reports of success in Table 3.5, an emerging set of practical constraints regarding time and space at the site level becomes evident.

In addition to testing hypotheses suggested by recent research on structural issues that foster collaboration, we were also interested in using the data to explore the ways in which the process of collaboration was reported by directors and how the many aspects of the processes described in the literature related to one another. Our preliminary analyses have begun to examine some of the ways these issues are manifest in the data. In addition to examining the data item by item, we found in the course of this investigation that it was helpful to condense information collected from several questions which tapped common dimensions. Thus, we created scale variables from items that were logically related to one another and that displayed relatively strong statistical correlations with one another.

Table 3.7 shows the item composition of a scale that we believe begins to measure some of the critical aspects that have been reported as crucial to collaboration. The items from which this scale is composed all give some indication of the presence of collaborative behavior and attitudes among program personnel. Among these items are a sense of trust, agreement on the identification of client problems and appropriate actions, and the sense that professionals agree about operating procedures. Note that through these items (and therefore in the scale variables that were subsequently created), the directors presented a relatively "rosy" picture of their programs, with the data skewed such that an overall positive perception of their programs becomes apparent. For example, on a 1 to 5 Likert Scale, with 3 being a "neutral" position, the mean across the sample for many items is well above this neutral midpoint, often falling clearly into the "agree" range. As will be discussed below, however, we can also discern instances in which stresses on collaboration and service integration are evident.

The four items shown in Table 3.7 were summed to create a collaboration scale with a Cronbach's alpha of .759. This scale was intended to measure the levels of perceived collaborative activity in the programs. The mean of the scale variable (in its unstandardized form, ranging from 4 at the low end to 25 on the high end) is 15.41. The standard deviations across items are relatively small; thus, overall perceptions were nearly uniformly positive. Still, closer analysis reveals that some programs perceived a much higher degree of collaborative activity than others.

Table 3.8 shows a similar item composition for the scale that was created to assess program directors' perceptions of the degrees to which their respective programs had affected collaboration among professionals and changed the way services were delivered to clients. Thus, the scale variable "Change" measured the program directors' perceptions of the extent to which collaborative activity has increased due to the programs. In this scale of seven items, five questions utilized a format that permitted program directors to report change on a continuum ranging from "not able to judge" and "very little change" to "almost everyone has changed." In addition, two Likert items were included that measured perception of change. As in the collaboration scale, these items were summed to form a change scale, with a Cronbach's alpha in excess of .85.

It should be noted that the mean responses on these "Change" items indicated a moderately positive assessment of change in these programs. The uniformly positive assessments of present collaborative activity cited earlier, however, appear to have exceeded program directors' reports of the degrees to which their programs have contributed to increased collaboration; fewer respondents to the change questions, for example, used the most extreme rating for assessment of change (that is, 5, "Almost everyone has changed") than the number of respondents who "strongly agree" with Likert items used to assess collaboration. In addition, standard deviations on the individual items as well as on the change scale itself indicate greater variance in perception about the extent to which the program has affected collaboration than is the case with the collaboration scale. Further, change was not perceived equally in all program dimensions. It is clear from Table 3.8, for example, that program directors were the least optimistic about changes in fundamental attitudes and the ability to develop a shared language among professionals.

Table 3.9 displays the item composition and mean scores for each item used in another scale variable, "Stress." These items included three Likert-type items reflecting moderately negative perceptions of program effects and communication among professionals; note that these items did not evince the agreement biases of the questions on positive perceptions cited earlier, and the mean perceptions across the sample ranged from disagreement with a statement about hasty implementation, moderate agreement with a statement regarding the need for more time to be "really effective," and mild disagreement with a statement that refers to occasional communication difficulty among profes-sionals. Standard deviations are also relatively high when compared with items discussed earlier. Three other items were included about potential problems that might be affecting collaboration, including personality conflicts, unclear responsibilities, staff training needs, and necessary time to discuss important issues with one another. (These last three items are also presented in Table 3.6.) As we have seen, the data show that while the reported incidence of problems was generally low, the most often reported issue was the lack of time to confer among professionals. Taken together, these items combine to form a rough index of some reported stresses on the collaborative process. The scale variable "Stress" (alpha = .72) has a mean across the sample of 17.63.

In a later section we will discuss some of the correlations among these variables and how other factors, including the number of services offered, are related to these scores.

Relationship with the Community

Table 3.10 presents descriptive data on program directors' responses to items relating to the quality of the relationships between the collaboratives and the communities they serve, degrees of community involvement, and levels of program outreach to and awareness of the communities' needs. All items presented were based on a Likert scale, with 1 indicating "strongly disagree" and 5 "strongly agree."

With respect to quality, program directors seemed positive about the nature of program-community relationships, agreeing both that these relationships were good and that they have gotten better as a result of the programs. On a separate item (not presented in Table 3.10), some 90% of the directors disagreed or strongly disagreed with the statement "Since this program began parents have become too demanding." In fact, program directors were also generally optimistic about the degrees to which communities are involved in their programs, and presented a moderately positive response to a question about the level of community support. Program directors were quite positive about the degrees to which they perceived their staffs reach out to their communities, strongly agreeing in most cases that efforts for outreach and awareness of community problems were high.

These items have been combined to form the scale variable "Community Relations," with a mean of 24.24 and a Cronbach's alpha of .78.

Relationships among the Variables Describing the Process
of Collaboration and Community Relations

Table 3.11 shows the results of some preliminary analyses in which we have begun to explore how these summative scale variables are related to one another. The Pearson correlation coefficients (and significance levels for a two-tailed probability test) are displayed for the scale variables "Collaboration," "Change," "Community Relations," and "Stress."

The perception of collaborative activity (Collaboration) and the attribution of increased collaboration due to program activities (Change) have a strong positive association with one another. Thus, in many cases in which directors reported high degrees of collaboration, reports of high levels of change were also evident. A somewhat unexpected finding was the association of Change and Collaboration with perceptions of good relationships between programs and communities. Positive perceptions of community relations are associated with collaboration, but they are even more strongly associated with perceptions of change.

The index of stresses to collaboration has a predictable negative association with collaboration and change. We note for further investigation the prominent negative association between stresses and the perception of positive community involvement. Given these correlations, we can discern that when community relations are not positive, stresses to collaboration are more likely to be reported. It is difficult to determine at this point in our analyses whether communities responded negatively to programs under stress, or, conversely, whether lack of community support was itself so critical that stresses to further collaboration resulted.

Table 3.12 displays another interesting set of correlations among these variables. Though the number of services offered had little to no effect on perceptions of collaborative activity (Collaboration) or, surprisingly, perceived stresses to collaboration (Stress), it had a strong association with perceptions of change, and nearly as strong an association with positive assessments of community relations. In other words, the more services offered by a given program, the greater the likelihood that said program's director would perceive that (a) the program had contributed to change in service delivery, and (b) community-program relations were good.

The Role of the Program Director

Finally, we turn to some of the reports found in these data that help to describe the role program director. In one set of questions, directors were asked to report the amount of time they spent in a the past year on a range of activities. The response scale was as follows: 6 (very frequently, or two to three times a week or more); 5 (frequently, or at least once a week); 4 (sometimes, or one to two times a month); 3 (occasionally, or 3-4 times a year); 2 (rarely, or once a year); and 1 (never). In short, the higher the score, the more time spent on the activity. Table 3.13 reports the mean scores for amount of time spent on activities, as well as standard deviations. In the fourth column of the table, the mean scores are ranked; in this column, the lower the rank, the greater the amount of time spent on the corresponding activity. As is evident, the most time is spent on administrative paperwork; least time is spent speaking to the press.

Table 3.13 also reports the results of another question that presented directors with the same list of activities and asked them to indicate which three were most important to their program, regardless of the amount of time spent on each. Not surprisingly, the activity in which program directors engage most frequently (completing paperwork) is cited as one of the least important things they do. Likewise, securing funds from funding sources, an activity that is relatively infrequent, is cited as the second-most important activity. Although ranked fourth in importance, meeting with community members is sixth in terms of amount of time spent. Perhaps most notable is the relative isolation of program directors; they spend little time talking with professional colleagues in similar programs, and rank such conversations low in importance. Further, they spend little time at conferences that focus on collaborative services.

Table 3.14 reports some basic facts about the tenure of the program directors in our sample. Fifty percent of the directors have served in their present position for five years or less; the modal response category for tenure in present position is two years or less. Some directors did report that they had been associated with their projects prior to becoming director, and some may have served as collaborative program directors in other venues prior to serving in their present positions. Overall, however, this is a relatively newly hired group of individuals who began working in their present positions between 1992 and 1994 (recall that the survey was administered in late 1994). Experience appears to be associated with the perception of stresses to collaboration, given that the correlation between stress and the variable measuring tenure in present position is -.30 (p = .002). Thus, it appears as though the more experience a given director had, the less likely he or she was to report that there were problems with the collaborative activity. We cannot discern from these data if this perception of stress is due to a relatively shorter tenure in the position, or if collaborative efforts experiencing problems tend to have frequent staff turnover or hire less experienced directors.

Finally, we asked program directors what kinds of resources they had used to solve technical problems in their programs. Table 3.15 shows the percentage of program directors who reported that they used particular sources of assistance, and further delineates programs by location. Perhaps the most notable point with regard to this table is that while directors reported that conferring with collaborating professionals is a valuable source of assistance, they also reported receiving little to no help from professional associations that are dedicated to their needs. Similarly, it was reported that universities and private consultants provide assistance to less than a third of the programs, with the bulk of the resources used coming from the literature on service integration or the personnel provided by funding agents.
Table 3.16 shows another interesting phenomenon worth exploring: as the number of services in a program increases, so does the number of resources that are used for assistance by the program director. It is possible that an expansion of services creates more possibilities for professional communities to provide assistance, or perhaps provides more resources with which to seek assistance. This could also mean that more complex programs have more technical problems, thereby impelling directors to seek more help than do directors in programs with relatively smaller scopes.

The purpose of this report is three-fold. First, we described fully the genesis of this study, including the conceptual framework that guided the development of the survey instrument. Second, we presented the technical details of the methodology used to create a comprehensive database of programs and to collect data from appropriate respondents. Third, we presented some of the basic findings of our study in each of four key areas: general information about programs, the collaborative process, relationships with the community, and the role of the program director. Major findings in each of these areas are briefly summarized below.

With respect to general information about programs, we noted that most of our sample is located in urban areas. The range of services offered by individual programs varies greatly, with most programs offering at least 2 different services, and some over 30. In general, health and family services are offered more frequently than academic interventions, despite the fact that schools are most often the location of these programs. We will continue to explore hypotheses about the extent to which the programs under study represent "co-location" of services rather than genuine collaboration.

The structures and factors affecting the process of collaboration appear to be quite pragmatic in the view of the program directors who responded. Although the literature details the advantages of formal agreements and incentive structures that promote participation by professionals, the program directors cited flexible time and space arrangements as being more critical to the success of their programs. While their perceptions of collaborative activity in their own programs were generally quite high, they showed variation in the degrees to which they believed the programs have contributed to real change in the delivery of services, especially with respect to transformations of fundamental attitudes among professionals and the development of shared languages. With respect to relationships with communities, the directors were generally positive about the benefits and extent of community involvement. To generalize from the present sample, it appears as though the development of positive relationships with communities is associated with lower degrees of stress on the collaborative process.

Finally, the survey depicted the role of the program directors as being extremely complex, challenging them to manage a variety of time-consuming, often competing tasks. Many of the most common activities in which the directors reported engaging were administrative in nature, even though they reported that spending time on such tasks did not warrant a high priority. Additionally, most of the surveyed directors are relatively new to their positions and the least experienced of these reported the greatest stresses in the collaborative process. In a future paper, we will continue to explore the outlets directors use for problem solving and the importance of emerging collaborative networks as they seek out new resources for assistance.

Capper, C. (1994). "We're not housed in an institution, we're housed in the community": Possibilities and consequences of neighborhood-based interagency collaboration." Educational Administration Quarterly, 30(3), 257-277.

Comer, J. (1980). School power. New York: Basic Books.

Crowson, R. L, & Boyd, W. L. (1993). Coordinated services for children: Designing arks for storms and seas unknown. American Journal of Education, 101(2), 140-179.

Heath, S., & McLaughlin, M. (1994). The best of both worlds: Connecting schools and community youth organizations for all-day, all year learning. Educational Administration Quarterly, 30(3), 278-300.

Kirst, M. (1991). Improving children's services. Phi Delta Kappan, 72(8), 615-618.

Koppich, J. (1994). The politics of policy making for children. In L.S. Adler & S. Gardner (Eds.), The politics of linking schools and social services (pp. 51-62). Washington, DC: The Falmer Press.

Mitchell, D. E., & Scott, L. (1994). Professional and institutional perspectives on interagency collaboration. In L. S. Adler & S. Gardner (Eds.), The politics of linking schools and social services. Washington, DC: The Falmer Press.

Schorr, L. B. (1988). Within our reach: Breaking the cycle of disadvantage. New York: Doubleday.

Smylie, M., Crowson, R., Chou, V., & Levin, R. (1994). The principal and community-school connections in Chicago's radical reform. Educational Administration Quarterly, 30(3), 342-364.