Seeking Person-Environment Fit in Community Care Placement

Abstract

Large numbers of patients leave mental hospitals to become residents of community-based sheltered-care facilities, yet little is known about how they use particular local environments to satisfy their needs and wants. This paper considers a crucial issue in community care placement, person-environment fit, using survey data from interviews with 397 residents in 211 sheltered-care facilities, drawn from 157 census tracts in California. It studies how individual characteristics interact with eight environmental contexts to influence sheltered-care residents’ external social integration. The results underline the power of social norms to determine ex-patient outcomes within specific environments. Ex-patients do relatively better in facility/community environments that allow their independent outreach, and also where they either share personal characteristics with the dominant social group or the dominant group is able to tolerate behavioral differences.

Many patients leave mental hospitals to become residents of community-based sheltered-care facilities—i.e., halfway houses, board-and-care homes, or family care homes. However, very little is known about how these former patients use local environments to satisfy their needs and wants, and how a particular placement affects their ability to do so. This paper considers a crucial issue in community care placement: person-environment fit, or “the degree of congruence … between an individual’s needs, capabilities, and aspirations and the resources, demands, and opportunities characteristic of the environment” (Coulton, 1981, p. 26).

This problem has been little studied, in large part because it is difficult to collect data that permit consideration of both differences among ex-patients and among their environments. We present such an analysis here, based on the assumption that individual characteristics, in interaction with characteristics of specific facility and community types, have an important bearing on social integration.

Method

The data came from a 1973 survey of community-based sheltered-care facilities located in 157 census tracts or Census County Divisions (CCDs) in California (see Segal & Aviram, 1978). The facility sampling frame included all family care, board-and-care homes, and halfway houses in California serving at least one qualified resident. It excluded all licensed hospitals, as well as nursing homes and intermediate care facilities.

To obtain the samples, the state was divided into three master strata: Los Angeles County, the nine-county San Francisco Bay Area, and all other counties in the state. We stratified facilities by size in both Los Angeles and the Bay Area, and drew a sample of facilities with probabilities proportionate to bed capacities. For the third stratum, made up of “all other counties,” we designed a cluster sample using counties as primary selection units. Two counties were selected from the north and two from the southern part of the state with probabilities proportionate to capacities. From each pair, a sample of facilities was selected, also with probabilities proportionate to capacities. Finally, using a specially prepared field listing, we randomly selected residents within facilities. (Further details of sampling procedures are reported in Segal & Aviram, 1977.) The samples of residents and facilities were self-weighting probability samples representative of all 12,430 18–65-year-old, nonretarded, formerly hospitalized sheltered-care residents, and all 1,155 facilities in the state during the summer of 1973. This methodology yielded a final sample of 211 facilities and 397 residents.

Data were collected from two sources: (a) face-to-face structured interviews with the residents and the operators (on-site managers) of each facility; and (b) aggregate data on the demographic, housing, criminal justice, and voting characteristics of all 157 census tracts and CCDs that contained sampled facilities, culled from the 1970 census and other public records.

Defining the Environment

The sheltered-care environment includes both the facility in which former patients live and the surrounding neighborhood, for which the census tract served as a surrogate. To control for aspects of the facility and the neighborhood, we created typologies of environments (Tryon & Bailey, 1970), constructed separately for facilities and neighborhoods; that is, the unit of analysis was the facility or the census tract. A principal-components factor analysis of a number of indicators yielded factor scores, each of which represented an important descriptive characteristic of an environment. Facilities or neighborhoods were then clustered together based upon the similarity of their profiles across these factors.

The analysis of facilities distinguished six types or clusters based upon five factor scores (see Fig. 1). The factors were degree of complexity (more complex facilities are larger, more highly staffed, and not regarded by their operators as surrogate families); program orientation (high scorers sponsor frequent social activities and offer in-home rehabilitation programs); control (more controlling facilities have curfews, regular schedules, generally applicable rules, and great authority vested in a facility manager); support (supportive facilities encourage mutual aid among residents and staff); and medical orientation (high scorers view residents as patients, supervise medications, employ a house physician, and shelter a more physically infirm group). (Details of this typological analysis are reported in Segal & Moyles, 1988.)

Fig. 1.

Facility type profiles, plotted in standardized factor score units ($\overline{X}=50$, SD = 10). Factor scores are “reflected” so that higher scores indicate more of each characteristic.

The analysis of communities distinguished five neighborhood types or clusters based upon five factorial dimensions (see Fig. 2): political conservatism (more support for Republican candidates and capital punishment, and opposition to school busing), family orientation (many owner-occupied, single-family structures, many dependent children, and relatively large households); socioeconomic status (high incomes and home values, high levels of education, low unemployment, and high prevalence of managerial and professional employment); criminal activity (high rates of murder, burglary, rape, assault, and theft); and non-traditional orientation (political support for ecological initiatives and legalization of marijuana, large numbers of hotels and rooming houses, and a high percentage of women in the labor force). (Details of this analysis are included in Segal, Baumohl, & Moyles, 1980.)

Fig. 2.

Community type profiles, plotted in standardized factor score units ($\overline{X}=50$, SD = 10). Factor scores are “reflected” so that higher scores indicate more of each characteristic.

The following paragraphs describe five facility types and four neighborhood types, which in combination contained a sufficient sample size for internal analyses to demonstrate significant relationships between resident characteristics and the level of resident social integration.

The five facility types (see Fig. 1) can be described as follows (using our labels rather than facility self-designations):

1. The family care home (17% of California facilities). This facility is small (average bed capacity of 7 residents) and family oriented. It serves more stable older residents (an average age of 46) and lacks a formal program intended to move residents out of the facility. These homes are neither formally controlling nor directive, but are only somewhat supportive, at least as defined above. They are not medically oriented.

2. The group home (20% of California facilities). This facility is not family oriented and is above average in size when compared to the entire sample of 211 facilities (average bed capacity of 30 residents). It has a relatively complex structure, employing staff and using extramural medical, social, and psychological services. It is similar to the family care home in that it does not have a formal outreach program, serves an older population (mean age 49 years), does not emphasize control over its residents, and is somewhat supportive in comparison to other facility types. The group home has a profile closely approximating the average on all five dimensions in Fig. 1.

3. The group home, programmatic (24% of California facilities). The programmatic group home resembles the group home, although it is somewhat larger (average bed capacity of 77 residents), more structured, and as the name suggests, programmatic in orientation, which may account for its younger residents (a mean age of 40). It controls its residents more than the group home, but offers a slightly more supportive environment. Like the group home, it is average in medical orientation.

4. The medically oriented facility (19% of California facilities). This facility is of average size (bed capacity of 17 residents), complexity, and program orientation. The average age of residents is 47. It is distinguished by its highly controlling and directive policies and by its low score on social support. As the name implies, it is very high on medical orientation.

5. The therapeutic community (16% of California facilities). This is the most complex facility. It is large (average capacity of 56), employs staff, makes extensive use of outside services, and is not like a family. It places great emphasis on its program, focusing on rehabilitative activities, and caters to the youngest population (an average age of 39). The therapeutic community is neither controlling nor directive, but also is not particularly permissive. It is slightly supportive and somewhat nonmedical.

The four neighborhood types can be described as follows (see Fig. 2):

1. The average sheltered-care neighborhood (contains 31% of sample facilities). This is not identical to the average California community, although it bears some resemblance. The average sheltered-care neighborhood is less conservative, more family oriented, and less affluent than the state as a whole. On all but family orientation, where it ranks well above all but one other neighborhood type, it falls in the middle of our descriptive dimensions.

2. The liberal, racially mixed, working-class neighborhood (contains 19% of the sample facilities). This neighborhood is only 42% Caucasian (as opposed to 89% of California’s enumerated population in 1970). It is extremely liberal in comparison to the whole state, and it is relatively impoverished (the average family income is 25% below the state average). It appears less family oriented than the state as a whole, principally because it has relatively few single-family dwellings and very few owner-occupied units, but given the lower incomes of neighborhood residents, this characteristic should be interpreted with caution. Crime rates are very high compared with state averages, and higher than those in other sheltered-care community types. The percentage of unaffiliated males (neither working nor in school) is three times the state level (12% versus 4%).

3. The conservative working-class neighborhood (contains 16% of the sheltered-care facilities). This is a predominantly white (89%) conservative neighborhood in which incomes are low (18% below state average) and crime rates are high. Residents tend to be apartment dwellers (58% live in large structures, compared to 20% statewide) with few children (the youth dependency rate of 12% is less than half the state’s 28%).

4. The conservative middle-class neighborhood (contains 24% of the sample facilities). This is a white (95%), very conservative, well-to-do neighborhood (family incomes are 28% above the state average). It contains chiefly single-family structures (76%) occupied by families with children, and its crime rates are somewhat lower than other neighborhood types. The proportion of eligible residents turning out to vote (44%), while nearly identical to the state average (43%), is much higher than in other sheltered-care community types, where from 27% to 30% of all eligible voters cast ballots in the 1972 general election.

Measures of Fit

Social integration.

A previously developed self-report measure of “external social integration” was used as a criterion for matching the fit of individuals with facilities and communities (cf. Segal & Aviram, 1978). External integration is defined as the extent to which an individual participates in and makes use of the community in a self-initiated manner—that is, independent of the facility’s efforts. External integration includes a number of related dimensions: the amount of time a resident spends outside the facility on an independent basis; the ease with which the resident engages in social contacts with people and in work-related tasks, uses community services, or obtains basic resources; the amount of contact a resident has with family, friends, and acquaintances; the involvement of the resident in income-producing activities or in educational activities that might lead toward employment; the amount of time the resident spends on independent consumer activities; and the resident’s access to his or her own bank account.

The external social integration scale consists of seven subscales measuring the above dimensions. The average subscale internal consistency score (alpha) was .78. Our past validity studies and that of Hylton (1981) have shown that the total scale score and subscale scores effectively distinguish between groups of former chronic mental patients, criminal justice halfway-house residents, and a sample of the general population—that is, all differences observed were in the expected direction, with former patients scoring the lowest and the general population scoring the highest.

Individual characteristics.

We chose for study 16 individual characteristics significantly related to social integration in our whole sample and linked to patient outcome in previous investigations (Davis, Dinitz, & Pasamanick, 1974; Miller, 1965). They included age, gender, race, socioeconomic status, education, marital status, sufficiency of money, resident choice in placement, chronicity of hospital experience, conservatorship status, Langner Scale score, paranoid symptomatology, schizoid symptomatology, somatic stress symptoms, social competence, and use of illegal drugs or alcohol. Information on all 16 characteristics was obtained in the resident interview. Nine of the 16 characteristics were single variables and seven were unit-weighted composite scores, made up as follows:

Four psychopathology variables were derived from combinations of items drawn from the Langner Scale (Langner, 1962) and the Brief Psychiatric Rating Scale (BPRS; Overall & Gorham, 1962). They were (1) a somatic stress symptoms composite, which included reports of frequent pains, nervousness, shortness of breath, lack of sleep, feeling ill, inability to sit still, and breaking into sweats; (2) a schizoid symptom composite, which included disturbed affect, withdrawal mannerisms, disorganized thinking, and motor retardation; (3) a paranoid ideation composite, which included expressions of suspiciousness, hostility, grandiose behavior, and unusual thought content; and (4) the 22-item Langner Scale score.

The Socio-Economic Status (SES) composite score was derived from the resident’s and father’s Socio-Economic Index (SEI) status (Reiss, 1961), and the resident’s and father’s education. The chronic hospitalization composite was based on total and continuous time in mental hospitals and the date of first admission. Based on the work of Zigler and Phillips (1961), the social competence scale was derived from a nonlinear composite rating of marital status, employment, and age.

Analyzing Person-Environment Fit

To document that the social integration of the resident depends on his or her environment, it is necessary to show that predictive models differ across environments. Thus, we had to control for the environment when deriving our models, but full accomplishment of this task would require an extremely large sample size. We would need sufficient statistical power to detect significant predictors within environments and to show differences among environments.

Our analysis was the first step toward this end. We first chose eight environments that had a sufficient number of cases (minimum N = 17). Since each community/facility environment had relatively small Ns, the tests of multivariate prediction of external social integration were confined to only five predictors. These five individual characteristics chosen as possible key factors in placement decisions included two major demographic characteristics of the client, age and race; two “red flag” characteristics, paranoid symptomatology and drugs/drinking problems; and one opportunity characteristic, the availability of money. (These predictors were only weakly intercorrelated.) To test whether the eight regression equations were indeed different, or whether the subsamples could be pooled and a single model computed, a generalized Chow test (Hu, 1982) was computed. The test yielded an F of 92 (p < .0001), thus rejecting the null hypothesis that the regression coefficients in the subsamples were statistically identical to those in the pooled one.

Based on this result, here we report the general pattern of associations that appear within each of the eight multivariate models. The purpose of this is to show that there are differences in predictors of fit among environments and that these differences follow meaningful patterns. To further elaborate possible patterns of predictors, we also report the bivariate relationships between external social integration and 11 other relevant individual characteristics. These data, however, do not allow a strong test of the interaction between environment and individual characteristics in any particular model. We discuss the multivariate models and bivariate correlations separately for each environment. However, these results are mainly intended to suggest future research possibilities with a larger body of either survey or quasi-experimental data.

Results

Table 1 displays results of the regression equations predicting external social integration scores within eight facility/community environments from five individual characteristics: race, money, age, paranoid symptoms, and drug/drinking problems. Significant bivariate correlations of variables that were not significant in the multivariate model appear in parentheses. It is evident from these eight regression models that money and age each play a consistent role in many of the individual environments, whereas race and involvement with drugs and drinking interact with the community/facility environment setting to produce different relationships with external social integration in different environments. Table 2 presents the bivariate relationships between external social integration and 11 additional individual characteristics. From the information available in both tables, the following person-environment fit patterns emerge within each community/facility environment.

Table 1.

Predictors of External Social Integration Within Facility/Community Combinations

Facility/neighborhood	Individual characteristics
	N	R2	Racea	Money	Age	Paranoid Symptoms	Drug/Drinkinga
Family care/average	17	.53	−.35	.61			(−.40)
Group/conservative, middle class	17	.78	−.28	.59	−.38
Group, programmatic/average	21	.40		.64
Group, programmatic/liberal, working	25	.34	.28		(−.39)	(−.33)
Group, programmatic/conservative, working	20	.55	−.41	.36		−.48	.41
Group, programmatic/conservative, middle	24	.38		.51	(−.33)
Medically oriented/liberal, working	19	.21		(.44)
Therapeutic community/conservative, working	38	.20			−.37

Note. Table reports partial standardized regression coefficients, p < .10. Values in parentheses are significant (p < .05) bivariate correlations for which the partial regression coefficients were not significant in the multivariate models. All models except for “medically oriented/liberal working class” were significant, p < .05.

^aCoding: Race—coded Caucasian (0) and non-Caucasian (1); a negative relationship means that whites scored significantly higher on external integration. Drug/Drinking—coded (1) if the resident currently used drugs or alcohol and (0) if not; positive relationships indicate higher external integration of users.

Table 2.

Association Between External Social Integration and Eleven Individual Characteristics Within Facility/Community Combinations

Facility/neighborhood	Individual characteristics
	N	Female gender	Schizoid symptoms	Education	Chronic	SES	Choicea	Conservatora	Langner Scale	Single	Somatic symptoms	Social competence
Family care/average	17	−.38
Group/conservative, middle class	17		.33					.40		.35		.55
Group, programmatic/average	21		−.32	.30
Group, programmatic/liberal, working	25			.54		.70	−.45				.35
Group, programmatic/conservative, working	20				.32
Group, programmatic/conservative, middle	24	−.32		.46		.38		−.45	−.42		−.27
Medically oriented/liberal, working	19				.37		−.45				.36
Therapeutic community/conservative, working	38		−.27					−.27	−.54	.36	−.28

Note. Table reports significant (p < .05) bivariate correlations of variables not included in the multivariate models.

^aCoding: Choice—high scores indicate that residents chose the facility in which they were living; a negative relationship indicates that residents who did not choose to live there had significantly higher external integration scores. Conservator—coded (1) if the person had a conservator and (0) if not. Single—coded (1) if the resident was single and (0) if married or divorced.

Family care home/average neighborhood (R² = .53).

Race and sufficient spending money were the two significant individual factors in this context. Probably because the average sheltered-care neighborhood was 82% white, Caucasians were most likely to do better on external integration. While incomes in these neighborhoods were slightly less than average for the state, they were significantly above those of a typical working-class neighborhood and approached those of more conservative upper-middle class areas. Sufficient spending money was an important factor in most types of communities. Based on the bivariate results (Tables 1 and 2), males (r = −.38) and drug or alcohol users (r = −.40) had lower external integration scores in this environment.

Group home/conservative middle-class neighborhood (R² = .78).

In these well-to-do communities, money, youth, and Caucasian race were the important factors promoting external integration. It seems likely that the search for companionship or recreation in conservative middle-class communities often involves travel outside the area. Thus, younger sheltered-care residents and those with money in a relatively unprogrammed group home are more likely to take advantage of such possibilities. Older residents are more likely to be warehoused. Whites are also more likely to do better in these conservative, white, middle-class areas.

Social competence (r = .55), being single (r = .35), being under conservatorship (r = .40), and being schizoid (r = .33) all had significant bivariate relationships with external integration (Table 2). The positive relationships between schizoid behavior, conservatorship, and social integration may indicate that nonthreatening individuals can be best accommodated in this environment.

Group home, programmatic/average neighborhood (R² = .40).

Sufficient spending money was the only significant multivariate relationship promoting external social integration in this environment. Nonschizoid residents (r = − .32) and those with more education (r = .30) were most likely to be externally integrated in these settings, based on the bivariate findings. Programmatic group homes are designed to move more competent adult individuals out into the community; the findings suggest that the program works best for people with financial resources.

Group home, programmatic/liberal, racially mixed, working-class neighborhood (R² = .34).

In this environment, where an average 58% of the population was nonwhite, minority status was the only individual characteristic positively and significantly related to external integration in the multivariate model. Thus, the liberal, racially mixed, working-class community seems to provide a context that supports the minority patient’s integration. Bivariate relationships in these settings indicated that residents with higher external integration scores were younger (r = −.39), higher on somatic complaints (r = .35), of higher socioeconomic status (education r = .54; SES r = .70), and did not choose the facility (a fact that fits a “patient” profile). Those who displayed paranoid symptomatology (r = −.33), however, had lower external integration scores.

Group home, programmatic/conservative, working-class neighborhood (R² = .55).

People with a history of alcohol and drug use or access to money did better, and minorities or people with paranoid symptoms did poorly on external integration in the multivariate analysis of these settings. Bivariate results indicated that a history of chronic hospitalization (r = .32) was also associated with higher levels of external integration in this context. While it is encouraging to find an environment where a history of chronic hospitalization was positively related to external integration, our data suggest that this enhanced social integration was based upon participation in bar life or drug culture, since the association disappeared when the use of drugs or drink was controlled. Thus it is evident that social integration is not always a desirable goal since it may lead to unintended and undesirable consequences in some settings. Future research of this kind must address the normative dimensions of external integration, which we have not considered in detail.

Once again, in this setting, people with paranoid symptomatology or minority status had more limited independent involvements in the environment beyond the facility.

Group home, programmatic/conservative, middle-class neighborhood (R² = .38).

As might be expected, the multivariate model showed that those with enough spending money were most externally integrated. Bivariate relationships showed a pattern of relationships between external integration and competence, and health and resource endowment: residents with high SES (r = .38), education (r = .46), lower Langner scores (r = −.42), no conservatorship (r = −.45), or fewer somatic complaints (r = −.27) were more likely to have high external integration scores. Also, those who were younger (r = −.33) or male (r = −.32) had higher levels of social integration.

Medically oriented facility/liberal, racially-mixed, working-class neighborhood (R² = ns).

While the multivariate model was not significant for this community/facility context, bivariate relationships showed that chronic patients (r = .37), those with sufficient money (r = .44), those high in somatized distress (r = .36), or those who did not choose the facility (r = −.45) had higher external integration scores.

Therapeutic community conservative, working-class neighborhood (R² = .20).

In this context, youth was the only variable in the multivariate model related to enhanced external integration. In the bivariate relationships, single status (r = .36), lower Langner scores (lower amounts of personal distress) (r = −.54), low somatic distress (r = − .28), lack of conservatorship (r = − .27), and few schizoid symptoms (r = −.27) also enhanced external integration.

Discussion

Taken together, the multivariate models and the bivariate correlations of this analysis underline the power of social norms to determine patient outcomes within specific environments. Specifically, patients do better where they can best fit into the social fabric of the environment. Fit seems to depend a great deal on the extent to which community care residents have key characteristics of the dominant social group in the environment, and the willingness of that group to tolerate differences in its midst. In the average neighborhood/family care home, the conservative middle-class/group home and the conservative working-class/programmatic group home, minority racial status was a handicap to resident external integration. Only in programmatic/group homes located in liberal, racially mixed communities (communities where, on average, 58% of the population were minorities) was there a positive relationship between minority status and external integration.

The data also suggest—although provisionally, due to small sample sizes—that residents of certain types of facilities do better or worse in the same community environment; that is, the type of facility interacts with the type of neighborhood. For instance, we might speculate that the failure of minority status to inhibit external integration in programmatic group home environments located in average and conservative middle-class communities indicates that the homes’ programs may somehow diffuse racial stigma.

Interestingly, community care residents who somatized their problems did relatively better on external integration in liberal, racially mixed, working-class communities in programmatic group homes and in medically oriented facilities. Perhaps the greater likelihood of Hispanics and Asians to express mental health problems in somatic terms, and the higher prevalence of physical health problems in black communities, facilitate acceptance of residents with somatic symptoms in minority-dominated communities. Ironically, the white middle-class public may perceive such environments as bad. In an in-service training session we conducted for community care workers, one social worker told us,

Many of the good facilities won’t take “bad” clients. We, therefore, put them in “bad” facilities, where they seem to do reasonably well. I hope in making plans for “better facilities,” they don’t close down all the “bad” facilities, because then there won’t be any place to send a “bad” client.

Following from this observation, it may be important to change perceptions of what is meant by “bad” facilities. The notion that model facilities are those located in white middle-class neighborhoods and ones that are sparkling examples of hygiene and efficiency, seems fundamentally misleading. Our results indicate that liberal, racially mixed, working-class neighborhoods can foster social integration for some clients without necessarily conforming to public stereotypes of “good” facility/community contexts.

The results also suggest precautions about placement in conservative middle-class communities. These neighborhoods seem better able to tolerate the more competent and least threatening community care residents. It is unclear whether these individuals develop contacts within the conservative middle-class neighborhood or whether their youth and access to money, predictors of integration in this environment, enable them to travel farther afield in pursuit of a social life.

Different types of pathology suggest different concerns about environmental placement. Paranoid symptomatology was not associated with higher external integration in any context, and people with these symptoms appear the most difficult placements to make. Schizoid symptomatology, on the other hand, although negatively associated with external integration in programmatic group homes and therapeutic communities in two different community types, was positively associated with social integration in group homes in conservative middle-class neighborhoods. Perhaps the negative effects of schizoid symptomatology in programmatic group homes and therapeutic communities reflect the higher expectations of these programs in comparison to those of group homes. The programs of the former types of facilities may attempt to involve clients in relationships for which they are not ready, thus leading to relatively less benefit from them (Weinman & Kleiner, 1978).

Community care residents with histories of chronic psychiatric hospitalization were found to do relatively better in certain types of facilities in working-class communities. Unfortunately, the data suggest that the chronicity acceptable in conservative working-class environments may be accompanied by drug and alcohol use.

Conclusion

In documenting the impressive role of the environment in determining outcomes in residential care, Cournos (1987) also pointed to the failure of the research she reviewed to identify personal characteristics that “might have more powerful predictive value” (p. 851). The present research suggests that the effect of individual characteristics may be context specific—i.e., that what predicts integration in one facility may not do so in another. Thus, by ignoring the interactions between person and environment, previous research may have overlooked the effects of what would otherwise be significant predictors of integration.

Sheltered care is an important resource for a heterogeneous population of former patients who, when given the opportunity and some sensitive assistance, may be able to find environments suited to their tastes and capacities for improved functioning. It is a mistake to view sheltered care as a monolithic enterprise suited mainly to warehousing phlegmatic, chronic patients who spend their days glued to the television (Van Putten & Spar, 1979). Sheltered-care settings have a salutary diversity of style that potentially permits former patients to find a congenial niche. Our research suggests an approach to determining person-environment fit in placement; future work, perhaps using quasi-experimental designs, can more precisely identify the characteristics of fit. Such research would permit placement in a way that takes greatest advantage of the diversity of sheltered-care environments.

Biography

STEVEN P. SEGAL, Ph.D., is Professor and Director of the Mental Health and Social Welfare Research Group at the University of California, Berkeley, where he is also coleader of a NIMH postdoctoral program on organization and financing of mental health services. His current research has four related strands: long-term adjustment of former psychiatric inpatients to residential care facilities in the community, implementation of psychiatric emergency evaluations in general hospital emergency rooms, and homelessness and self-help among the mentally ill.

CAROL SILVERMAN, Ph.D., is presently a postdoctoral fellow in the NIMH program on financing and service delivery in mental health at the Schools of Social Welfare and Public Health, University of California, Berkeley. Her areas of interest include self-help groups and the mentally disabled, community, and homelessness. She has published in the fields of housing policy and community organization.

JIM BAUMOHL, DSW, is Assistant Professor of Social Work at McGill University, and Affiliate Associate Scientist in the Medical Research Institute of San Francisco/Alcohol Research Group, Berkeley. He has a long-standing scholarly and practical interest in homelessness, mental illness, and addiction. He is currently studying San Francisco’s management of alcoholism and drug addiction between 1849 and 1945, and 19th-century Canadian inebriate homes.