To be included, a community had to be taking significant steps to end long-term street homeless and also had to have at least some data to prove that these efforts were making a difference. It was not easy to find communities that met both criteria.
 The communities involved were four states (Maine, Connecticut, Rhode Island, and Kentucky), two city/county sites (Los Angeles and Portland/Multnomah County, Oregon), and one multi-jurisdiction site (Seattle/King County, Spokane City and County, and the state of Washington).
 A fourth THCH community had these same characteristics initially, but a change in state leadership and direction reduced interest in PSH development and stiffened resistance to system change.
 Jacksonville 2005 Grant Application to HUDs Housing for People Who Are Homeless and Addicted to Alcohol program.
 State of the Plan reports and other relevant documents about the plan and its implementation can be downloaded from the Chicago Department of Housing Web site (www.cityofchicago.org. Click on Departments, then Housing, then "There's No Place Like a Home" section.)
 We discuss THCH findings on the value of the coordinator role because THCH focuses on homeless issues. Dennis, Cocozza, & Steadman (1999) reviewed earlier evidence for the coordinators importance in creating integrated service systems, focusing mostly on ACCESS-related research. Further published research on ACCESS provides additional documentation (Rosenheck et al., 1998, 2001, 2003a and 2003b). Proscio (1997) discusses the coordinators importance in developing an effective supportive housing development alliance, as do Greiff, Proscio, & Wilkins (2003). In addition, the experience of many other systems confirms this importance. See, for example, Burt, Resnick, & Novick (1998) with respect to services integration for high-risk youth, Burt et al. (2000) and Clark et al. (1996) with respect to serving women victims of violence, and Center for Mental Health Services (2003) and Huz et al. (1997) with respect to children and family services.
 Alternatively, guidance materials stress that participating agencies must give their staff time for the planning and coordinative work of system change, often advising that the key person be given at least half time to devote to system change activities (e.g., Proscio, 1997).
 The problem with attempting to submit different structures to an impact evaluation is that each is a unique product of its place, time, and circumstances. No community would ever be able to adopt a particular structure in its entirety, let alone expect it to function as it did in the community from which it was copied. These structures develop in response to the particular attitudes, histories, and cultures of their respective communities, as well as reflecting who cared the most and the agencies or organizations in which those people were situated when the need for mobilization occurred. Another community facing different conditions could not simply decide to adopt a particular structure; it would have to select the one with the best fit, and then modify it until the fit meets community needs and capacities. It would be important to examine whether the structure itself affects what can be accomplished, or whether the strength or weakness of the structure reflects the willingness of local actors to submit themselves to its direction. It would also be important to examine how much power the structure is given and how it wields that power. Further, it would be important to examine the existence and scope of a coordinator role, and whether a clever and dedicated coordinator can make any structure work, albeit using different approaches depending on whether the structure is strong or weak. A good research design would include approaches for sorting out which factors are most important for system change and its ultimate effects, and what is different but does not make a difference.