Predicting Employment Outcomes for People with ...

3 downloads 0 Views 432KB Size Report
(Bond & Friedmeyer, 1987), and the Work Personality Profile (Bolton & ... assessment to be good predictors of employment (Anthony and Jansen,. 1984; Stauffer ...
Predicting Employment Outcomes for People with Chronic Psychiatric Illness Andrew Man Hong Siu, OTR, MSc, MSc

ABSTRACT. This study investigates the factors that contribute to successfd employment of people with chronic psychiatric illness in Hong Kong, and to develop a statistical model that can predict their employment outcomes. Sixty-two clients from three psychosocial rehabilitation programmes were rated by their case therapists on their psychiatric history, mental condition, independent living skills, work performance, and social performance. Employment outcome is measured as the change in employment status, and the total earning achieved within three months. The percentage of the clients who improved their employment status was 21 3%. Diagnosis and demographic variables were not associated with the change in employment status. Total earning has significant correlation with three variables: attendance to program, number of hospital admissions, and relating with authority. Multiple regression analysis identified a sixvariables model that can explain 44% of the variance in total eaming. Path analysis hrther illustrated that attendance, mental condition, and work motivation exert significant effects on total earning directly, and through their influence on the social performance variables. (Article copies availablefor afee fmrn 771eHaworth Docuntent Delivety Service: 1-800-342-9678.E-mail address: geti~~o@huwor/fi,co~n]

People with psychiatric illness often have difficulties in seeking and maintaining a job. Their work history is often short and unstable. Psychiatric illness impairs their cognition, emotion, and motivation. Such impairAndrew Man Hong Siu is Assistant Professor, Department of Rehabilitation Sciences, Hong Kong Polytechnic University, Hunghom, Kowloon, Hong Kong. The author wishes to thank Ado Lo, Daphne Hung, Connie Yiu, Occupational Therapists, who assisted in the data collection, and H. K. Yuen for his invaluable comments on this manuscript: Occupational Therapy in Mental Health, Vol. 13(4) 1997 O 1997 by The Haworth Press, Inc. All rights reserved.

45

46

OCCUPATIONAL THERAPY IN MENTAL HEALTH

ments manifest in the deficits of volition, loss of habits, and deterioration in work and social performance (Rutman, 1994). Employment difficulties could be the result of psychiatric impairment and their associated functional disabilities, as well as institutionalization and being stigmatized. Employment is often seen as a major indicator of success in psychiatric rehabilitation. Improvement in employment status can lead to significant improvement in self-esteem and life satisfaction (Arns, 1992). The types of variables that have been shown to correlate with employment outcomes include: demographic variables (especially work history), indicators of psychiatric impairment (such as symptoms profile and number of admissions), measures of functional ability (task and social performance), and measures of ego strength (Stauffer, 1986; Anthony, 1994). Among these predictors of employrncnt outcome, two types of variables were most relevant to the practice of occupational therapy: measures of functional ability or disability, and indicators of psychiatric impairment. Measures of functional ability include ratings of independent living skills, interpersonal skills, and task performance, that could be obtained from performance testing or structured observation. In work rehabilitation settings, measures of functional ability were often obtained through the use of situational assessment. Situational assessment refers to the systematic observation of task-related and interpersonal behavior in a simulated work environment. Very often, situational assessment is aided by behavioral-anchored rating scales. Notable examples of these rating scales include the Griffiths Scale (Griffiths, 1973, 1977), the Threshold Evaluation (Bond & Friedmeyer, 1987), and the Work Personality Profile (Bolton & Roessler, 1986). Many studies showed that ratings made in situational assessment to be good predictors of employment (Anthony and Jansen, 1984; Stauffer, 1986). Four dimensions of work adjustment, namely work readiness (or motivation), work attitude, interpersonal relations, and work quality (or task competence), were most commonly found in these rating scales (Bolton & Roessler, 1986; Bond & Friedmeyer, 1987; Grifiths, 1973). Interpersonal relations was consistently found to be a more important predictor of employment outcome than the other dimensions (Stauffer, 1986; Anthony, 1994). Degree of psychiatric impairment is indicated by variables like diagnosis, ratings of mental status, number of previous hospital admissions, and the length of last hospitalization. Past research showed that employment outcome is affected by the long-term nature of psychiatric illness rather than diagnosis or specific signs and symptoms. This is supported by the lack of relationship between a person's psychiatric diagnosis or symptomatology and work performance or employment (Anthony, 1994; Antho-

Andrew Marl Hong Siu

47

ny and Jansen, 1984). On the other hand, a number of indicators that reflected the long-term nature of psychiatric illness were found to correlate with employment outcome (Anthony, 1994). Examples of these variables include number of previous hospitalizations, total length of past hospitalizations, and length of last hospitalization. In Hong Kong, the occupational therapist has an important role in prevocational training and vocational rehabilitation of people with psychiatric illness. Occupational thcrapists specialize in the assessment of functional ability and disability. This rcsearch aims to establish thc links between employment outcomes and the assessment of the functional ability. It examines how occupational therapists may contribute to the prediction of employment potential in psychiatric rehabilitation through functional assessment. Occupational therapists have a major role in monitoring the progress of clients in work rehabilitation, and in providing expert opinion on the work rehabilitation status of clients. The results of this research would also improve our understanding of the impact of psychiatric illness on functional ability and employment, as well as the relationship between psychiatric impairment, functional ability, and employment handicap. Therapists may utilize the results to obtain a better focus in assessment and set priorities in work rehabilitation. This study aims to identity the factors that are related to the successful employment of people with chronic psychiatric illness. Demographic variables, indicators of psychiatric impairment, and measures of functional performance are the three groups of factors that were collected in the study. Clients are followed up over a period of three months to note changes in employment status and the total earning they achieved. Relationship between the measures of employment success and the three groups of factors were explored using contingency tables, measures of association, correlation, and regression analysis.

METHOD Participants and Reltabilitation Program

Sixty-two clients from two day training centers and one day hospital participated in the study (32 men and 30 women, mean age = 36.0). Forty-eight clients (77.4%) have a diagnosis of schizophrenia, six (9.7%) clients have a diagnosis of obsessive compulsive disorder, four (6.5%) have mood disorders, and the diagnoses of the rest of clients included behavioral problems, personality disorder, and dementia. Most of the clients (79% ) were singlc. Twelve clients (19.4%) lived in supervised housing, while the rest lived in either rcnted or private apartments. The partici-

48

OCCUPA TlONAL THERAPY IN MENTAL HEALTH

pants have on average received hospital treatment 2.7 times. All of them had a high average attendance (M = 83.2%) to the psychosocial rehabilitation programmes offered by the three centers, and had on average joined the centers for just less than a year. Clients were invited to join the study by case therapists, and participation is voluntary. The three centers involved in the study offered psychosocial rehabilitation programmes on a daily basis. All centers have a multidisciplinary team of psychiatrists, nurses, occupational therapists, and medical social workers. All the participants attended the day occupational therapy programs during office hours, and there were optional adult education sessions and social gatherings in the evening. Participation in these activities was voluntary. The major aims of the occupational therapy program were to re-establish work motivation, habits, and skills, to facilitate social adjustment and maximal competence in independent living. Common training activities include workgroups, task and project groups, developmental groups, social skills training, independent living skills training, social and recreational activities. Procedure

Two recording forms were devised to collect information on the predictors and outcomes of employment. At the start of the study, the therapists used the first rating form to collect demographic variables, indicators of psychiatric impairment, measures of functional performance, and initial employment status. Demographic variables and indicators of psychiatric impairment could be obtained from review of case records or interview with the client. Therapists were also requested to rate the functional ability of the clients based on their observation in the next ten working days. Three months after the completion of the first recording form, the therapists monitored any changes in employment status and recorded the total earning achieved in the sccond rating form. The second rating form collects information on the employment outcomes. Therapists were requested to record any change in employment status, the final employment status, and the total earnings achieved in three months. All therapists who assist in data collection have extensive experience (average 6.5 years, range 6 to 8 ycars) in mental health.

The first recording form collects three groups of variables or possible predictors of employment outcome: demographic variables, indicators of psychiatric impainnent, and measures of functional ability. Demographic variables that were collccted include age, gender, marital status, and accom-

Andrew Ma11 Hong Silt

49

modation. The indicators of psychiatric impairment are the variables that reflect the present mental condition as well as the psychiatric history of the clients. The three variables that were collected are: (I) Primary psychiatric diagnosis, (2) overall evaluation of mental condition, and (3) the number of previous hospital admissions. These variables were chosen because they could be obtained from case records with high reliability and efficiency.

Functional Performance Functional performance is defined as task-related and interpersonal skills which are thc basis for any kind of productive behavior. Seven variables were collected to reflect the ability of the client in independent living, work performance, and social performance. Three variables were used to indicate the ability in independent living, i.e., independent living skills, nuisance behavior, and the need for supervision of medication. Nuisance behavior refers to social behavior that may provoke negative responses from staff or co-worker. These three variables were measured using the corresponding sub-scales of the Missouri Level of Care Instrument (Missouri Department of Mental Health, 1987). The Missouri Level of Care Instrument, a 79-item questionnaire, was developed to determine community placements of psychiatric clients (Kramer, Massey, & Pokorny, 1990). It matches functional abilities, behavior, and care requirements of clients with the requirements of 12 types of accommodation facilities for people with psychiatric illness. Internal consistency of the instrument was high (n = 985, Alpha > 0.78 for all items). The validated instrument has 7 factor-derived subscales: cormnunity skills, self care skills, nuisance behavior, sociability, skilled nursing need, proclivity for violent behavior, control of anger (Massey, Pokorny & Kramer, 1989). Weighted scores of the sub-scale can correctly predict 57% of the residential placements of clients. Task-related and social performance is assessed using a behavior rating scale devised by Grifiths (1973). The Griffiths Scale (Grifiths, 1973) is a 22-item rating scale for the assessment of work behavior, which aims to provide a comprehensive picture of client's assets and problems in the workplace. Past research demonstrated the high inter-rater reliability (r = 0.84) and high test-retest reliability (r = 0.75) of the scale. A principal components factor analysis identified 4 factors of the scale: task competence, response to authority and supervision, social relationship, work enthusiasm (or work motivation). Two studies showed that the scale could correctly predict 69% to 89% of the future vocational placements (Griffiths, 1977; Watts, 1978).

50

OCCUPATIONAL THERAPY IN MENTAL HEALTH

Employnzent Outcomes Employment outcome is measured by noting the change in employment status and the total earning achieved in work or training settings within the three-months follow-up period. The total earning achieved in threemonths is a commonly used measure of success in employment. Seven possible types of employment status are identified as the common work arrangements for clients in psychosocial rehabilitation programmes in Hong Kong: in-patient treatment (readmission),day treatment or trainiig, sheltered employment, vocational training or further education, transitional employment, competitive employment. Thempists were requested to record any change in employment status among these seven levels of employment, as well as to predict the most possible employment status after three months. RESULTS

Descriptive and CorrelationnlAtralysis Employment outcome was measured by the total earning achieved and the change in employment status at the end of three months. Forty-nine (79%) of the 62 clients continued their trainiig at their present psychosocia1 rehabilitationprogrammes. Four (6%) clients started their training in sheltered workshops, six (10%) began supported or transitional employment, and three (5%) clients started competitiveemployment. The occupational therapists were able to correctly predict 61.7% of the change in employment status at the three-months follow-up, although most of the correct predictionswere the cases who maintained status quo. Ten (16.7%) clients achieved a better than expected status, while 13 clients (21.7%) achieved a status lower than the expectation of therapists. Analysis using contingency tables and Chi-square statistic showed that there is no significant relationship between demographic variables (including gender, age, accommodation, and marital status) and the change in employment status. Similar proportions of clients with psychotic illness and non-psychotic illness achieved an improvement in employment status. Total earnings of the clients in the three-monthsfollow-upfall in a wide range from US$0.76 to US$758.10 (M = 466.00, SD = 127.66). The distribution of total earning is positively skewed, and has a wide spread. With a natuml logarithmic transformation, total earning was effectively transformed to a normal distribution. Examining the correlation matrix (Table I), number of admissions, attendance, and relating with authority

Andrew Muti Hong

Sill

51

are the three variables that have significant correlation with total earning. In particular, number of admissions has no strong correlation with other variables except with total earning, while attendancc has strong correlation with total eaming and the ability to relate with authority. Independent living skills, ratings of work performance, social performance, nuisance behavior, and the overall rating of mental condition have high correlations with each other. As expected, the strongest correlations were found among the four factors of the Griffiths Scale. There are also a number of significant negative correlations among the variables. It appears reasonable to see that nuisance behavior increases with decrease & ratings of task competence, ability to relate with authority, work motivation, and mental condition. Relating with authority has two significant negative correlations with attendance and total earning, which is less expected. This means that clients who received a lower rating in relating with authority are more likely to get a higher attendance and total earning. Regression Analysis The regression analysis aims to identify the relative importance of predictors of total earning (one of the employment outcome) at a three months follow-up. The regression analysis started with LO variables (as in Table 1). A forward stepwise method was used to guide the choice of predictor variables for a parsimonious and efficient regression model: (1) Examination of the significance of change in R~ on adding each possible predictor, (2) Comparing the values of the standardized regression coefficients for each predictor. The change in R~ is examined when each predictor variable is entered into the equation. A partial F-test is used to test the null hypothesis whether the change in R~ is significantly different from 0 or not. R~ increases from 0.26 to a maximum of 0.49 as the eleven variables were entered into the equation. The partial F-tests showed that only the change in R* is significant at 5% level for the two variables: attendance and number of admissions. The partial F-tests for the other five variables, including relating with authority, social relationships, medication, work motivation, and overall mental condition, also came close to significance (p > 0.10). The size of standardized regression coefficients (P) suggested that attendance, number of admissions, relating with authority, social relationships, work motivation, and mental condition are the most important variables in the regression. It is clear from these results that attendance, number of admissions, and social relationships are the key predictor variables. Three other variables

TABLE 1. Intercorrelations o f Indicators o f Psychiatric Impairment, F u n c t i o n a l P e r f o r m a n c e , a n d Total E a r n i n g Independent Medication Nuisance Task Relatingwith Social Work Anendance Living Sk~lls S u p e ~ ~ s ~ oBehaviour n Competence Authority Relat~onships Motivat~on Ln (Total earning) IndependentLiving Skills MedicationSupew~sion Nuisance Behaviour Task Competence Relatmgwith Authority Social Relationships Work Motivation

-.I9

-.25

-.08

-

- 17

-33" 33"

-

-.02 .75"

-.31* .71"

-.22

-.07

-.46"

-.46"

-.22 .68" .04 -.I5 .67"

-

Number of Admissions

.04

.49"

.OO

.69"

.08

.57"

.01

.06

.16

-.26 -57"

.52"

Mental Condition

-.lo .OO

.73"

.12

.61"

.64"

- .25'

-

.60"

.08

-

.16

-.41" 58"

.33'

.21 -.06

56"

.ll

.48"

-.03

.43"

-.06

Anendance

.09

-.20

Mental Condition

-

-.06

Number of Admiss~ons 'Two-tailed tests signlflcant at p < .05 "Two-ta~iedtests sgniflcantat p < .Ol

Alldrew

Mart Hang Sit1

53

that may also contribute to the prediction are relating with authority, work motivation, and mental condition. The final model includes six variables and the results of the regression are shown in Table 2. When the six selected variables are entered into the equation, R2 is 0.44 and Adjusted R* is 0.38 for the regression equation. This means that 44% of the variation in total earning could be explained by the regression using six variables. Adding the other five variables only improves the explanation by 5% (i.e., R2 = 0.49). F-tests showed that the regression is significant at 0.001 level. Residual plots showed no significant departures from the assumptions of regression. Regression diagnostics, including partial plots and F-tests, demonstrated the adequacy of the model. Although several outlying observations in the outcome variable werc detected, none of them could be considered influential. Collinearity was detected between two pairs of variables but remedial measures were not considered worthwhile. Path Analysis Although the regression was able to identify six variables (out of 11) that were most important in the prediction of the total earning, the relationship among these six predictors and the outcome variable is not clear. A path analysis is conducted to farther illustrate the causal inter-linkages among the predictors and the outcome variable (Asher, 1983; Retherford & Choe, 1993). Firstly, a saturated recursive model is specified. Number of admissions, mental condition, work motivation are identified as the explanatory variables, as they reflect the impairment brought about by psychiatric illness. Attendance, relating with authority, and social relationships werc specified as the response variables as they reflect the present functioning and disability of the client. Total earning that reflects the TABLE 2. Results of Regression Analysis for Variables Predicting Ernployrnent Outcomes (N = 62) Variable

b

Altendance

0.02

Number of Admissions

-0.10

Relating with Authority

-0.11

Social Relationships

-0.13

Work Motivat~on

0.07

Mental Condition Note

R2= .44 (p c .05). 'p

0.43 c 05

SE b

fi

54

OCCUPATIONAL THERAPY IN MENTAL HEALTH

degree of employment handicap is specified as the final response variable. The standardized regression coefficients of the initial model were estimated and tested for their significance at the 5% level. Insignificant paths were dropped from the model. Initial results showed that all the paths arising from number of admissions were insignificant, and therefore this explanatory variable was removed from the model. Attendance was then introduced as an explanatory variable as it only has significant paths coming out of it, but not leading to it. This is also theoretically sound as attendance could be seen as an indication of both psychiatric impairment or functional disability. Figure 1 shows the fitted unsaturated model formed after re-estimation, and el represents the residual variables in the model. Except mental condition, all variables have significant direct effects on the outcome variable. Social performance variabIes, including relating with authority and social relationships, mediate between the explanatory variables and employment outcome. Attendance has the strongest direct influence on total earning, and its indirect effects are relatively small (Table 3). Mental condition only exerts indirect effects on total earning through relating with authority. On the other hand, it is interesting to note that work motivation has a positive direct effect (0.33) that is effectively canceled out by its indirect effects (0.32) through the variables relating with authority and social relationship. DISCUSSION Demographic variables such as age, gender, accommodation, and marital status were not associated with the change in employment status. High TABLE 3. Direct, Indirect, and Total Effects of the Fitted Causal Model Exolanatotv Variable

Source of Elfectson Ln (Total Earninal

Value

Total Eflect

Anendance

Direct A d R d L AAR->S-->L

0.41 0.07 0.02

0.49

Mental Cond~tion

Direct M->R-->L M->R->S->L

0.00 4.11 4.04

4.15

Wolk Motivation

Direct W-AS->L W->R->L W->R->S->L

0.33 4.10 4.16 4.05

0.02

-

--

-

Nole.A->R~LmeansthepathfromAltendancetoRelalingwilhAuthoritytoLn(TotalEarning),andsimilaly other indirect paths.

forthe

FIGURE 1. Fitted Path Model on Predictors and Outcomes of Employment

56

OCCUPATIONAL THERAPY IN MENTAL HEALTH

correlations were noted between measures of functional performance. It suggested that highly consistent and reliable measures of functional performance could be devised in a smaller set of dimensions in functional performance. In addition, only two measures of functional performance (attendance and ability to relate with authority) and a measure of psychiatric impairment (number of previous hospital admissions) were significantly correlated with employment outcomes when a logarithmic data transformation is performed. This poses a doubt in the link between functional perfonnance and employment outcomes, and also reminds us of the lack of relationship between measures of psychiatric impairment and employment outcomes. However, the lack of relationship between demographic variables and employment outcomes, or between functional perfonnance and employment outcomes, could be the result of a small sample size. The regression analysis reached a final mulitiplicative model with six variables that can predict about 44% of the total variation in total earning. Two indicators of psychiatric impairment (overall mental condition and number of admissions) and four measures of functional performance (work motivation, social relationships, relating with authority, and attendance) were identified as the key predictors of employment outcome. This model reconfirms the importance of medical history, mental condition, work motivation, and social performance in employability. The predictive power of situational assessment, however, was not as strong as described by previous research (Grifiths, 1973; Watts, 1978). To obtain a model with higher predictive power, it appears that some other variablcs need to be included. Possible predictors that were omitted in this model include work history variables and measures of ego strength (Theodorakopoulos, 1990). It is also expected that employment outcome is affected by characteristics of clients, the training programmes and centers, as well as social and economic factors in the society. The predictive power of the present model is bound to be limited by the lack of consideration to program variables and societal factors that may alter the outcomes of employment. The causal path model constructed has three explanatory variables that include attendance, mental condition and motivation. These three variables exert their effects on employment outcome (total earning) through two intervening variables on social performance. Perhaps the most puzzling finding in the causal model are the ncgative paths between social variables and employment outcomes. This means that a client who may not have a good relationship with the supervisor, or is passive and not sociable, may in fact achieve a higher earning than another client who is sociable and has a good relationship with the supervisor. This finding could bc the result of spurious relationships among the predictors.

Andrew Man Hong Siu

57

Over 20% of the clients improved their employment status at the end of three months. This suggests that the predictive model may actually reflect the criteria of success in training rather than in open employment. The employment rate of 20% for people with psychiatric illness is largely consistent with the results of earlier studies (Hong Kong Council of Social Service & Rehabilitation Development Co-ordinating Committee, 1987; Hong Kong Council of Service, 1993). On the other hand, males, the married, and the 25 to 34 age group have better employment prospects than females, the unmarried, or other age groups in the general population in Hong Kong (Census and Statistics Department, 1991). In this study, demographic variables did not take an important part in employment. It appears that the factors affecting employment in the general population are not the same as in the study sample. Lastly, it was noted that therapists had to exclude some clients with low attendance from the sample, as it was difficult to conduct accurate assessment and observation of these clients. It is desirable that the study could be replicated with a larger sample. A longer period of follow up could be used, as it is noted that clients on average stayed at these centers for nearly a year. REFERENCES Anthony, W.A., Cohen, M.R., & Farkas, M.D. (1990). Psychiatric rehabilitation. Boston, MA: Boston University, Center for Psychiatric Rehabilitation. Anthony, W.A., Jansen, M.A. (1984). Predicting the vocational capacity of chronically mentally ill: research and policy implications. American Psychologist, 39,537-544. Asher, H.B. (1983). Camal modeling. Newbury Park: Sage. Bond, G.R., & Friedmeyer, M.H. (1987). Predictive validity of situational assessment at a psychiatric rehabilitation center. RehabilifafiorfPsychology. 3, 99-1 12. Census and Statistics Department (1991). Hor~gKoug 1991 population ceilsus: sunlrnary resuNs. Hong Kong Government. Cook, J.A., & Roussel, A.E. (1989). Who works and what works?: effects of race, class, age and gender 011 emnploymer~tanlorlg the psychiatrical[v disabled. Paper presented at the Annual Meeting of American Sociological Association, Chicago, IL. Douzinas,N., Carpenter,M.D. (1981). Predicting the Community Performance of Vocational Rehabilitation Clients, Hospi~aland Cornmunip Psychiat~v,32, 409-413. Griftiths, R.D.P. (1973). A Standardized Assessment of the Work Behavior of Psychiatric Patients. British Jour71alofPsychiat~y,123, 403-408. Griffiths, R.D.P. (1977).Thc Prediction of Psychiatric Patient's Work Adjustment in Community. British Jofinlal of Social ar~dClinical psycho log^^ 16, 165-173.

58

OCCUPATIONAL THERAPY IN MENTAL HEALTH

Hong Kong Council of Social Service (1993). ,411explorato~ystrrdyon sheltered enrployn~enfof psj~chiatricaNydisabled persons in Hong Korrg. Hong Kong: Hong Kong Council of Social Service. Hong Kong Council of Social Service & Rehabilitation Development Co-ordinating Committee (1987). Survey on emploj~nientsituu~ionof the disabled. Hong Kong: Hong Kong Council of Social Service. Jacobs, H.E., Wissusik, D., Collier, R., Stackman, D., Burkeman, D. (1992). Correlations between disabilities and vocational outcome. Hospital arrd Cornt~twtif), Psycltiat~y,43, 365-369. Kramer, H.B., Massey, O.T., Pokorny, L.J. (1990). Development and validation of a level-of-care-instrument for predicting residential placement, Hospital and Commru~tifyPsychiatry, 41, 407-412. Massey, O.T.,Pokomy, L.J., & Kramer, B.H. (1989). The development of factorbased level of functioning scalcs from a level of care instrument. Jownnl of Ciittical Psychology. 45, 903-909. Mintz, J., Mintz, L.I., & Phipps, C.C. (1992). Treatments of mental disorders and the functional capacity to work. In R.P. Liberman (Ed.), Har~dbookof Psychiatric Rehabilitation, pp. 290-316. New York: Macmillan. Missouri Department of Mental Health (1987 edition). Missorrri Level of Care /~rstrumetrt.Missouri: Author. Refherford, R.D., & Choe, M.K. (1993). Stalisticai modelsfor causal attalysis, NY: John Wiley and Sons. Rutman, I.D. (1994). How psychiatric disability expresses itself as a barrier to employment. Psychosocial Rehabiliration Journal, 17(3),15-35. Schultheis, A.M.M., & Bond, G.R. (1993). Situational assessment ratings of work behaviours: change across time and between settings. Psychosocial Rehabililatio~rJournal, 17, 107-119. Stauffer, D.L. (1986). Predicting successful employment in the community for people with a history of chronic mental illness. Occupational Therapy in Mental Healllr, 6, 3 1-49. Theodorakopoulos, P. (1990). The relationship of depetldency, ego-sfrerrgth,and of clients wilh attitude towards work lo vacaliotral rehobilitatio~rfi~~tctioni~rg psycltiatric disubilities. unpublished Ph.D. thesis, New York University. Watts, F.N. (1978). A study of work behavior in a psychiatric rehabilitation unit. British Jounral of Social atrd Clirtical Psycltologv, 17, 85-92.