ImummmmWinnlWTTTI i/ 3 1293 008773 1 This is to certify that the dissertation entitled PRETREATMENT DROPOUT AMONG PERSONS REFERRED TO PSYCHOLOGICAL SERVICES FROM EMPLOYEE ASSISTANCE PROGRAMS presented by William Robert Meermans, Jr. has been accepted towards fulfillment of the requirements for Ph.D. degreein Counseling, Educational Psychology, and Special Education Firm ELM/cut PM) . Major professor Date January 9, 1992 MS U is an Affirmative Action/Equal Opportunity Institution 0-12771 “ I LIBRARY 1 Mlchlgan State ‘ Unlversity PLACE IN RETURN BOX to remove We checkout from your record. TO AVOID FINES return on or before clue due. DATE DUE DATE DUE DATE DUE ll 4995 A99“? 1&ng I 5 i995 9 ”NJ? :“ I . 7W Wm 7 ‘ fi—W II MSU Is An Affirmative Action/Equal Opportunlty InetItutIon emote”! PRETREATMENT DROPOUT AMONG PERSONS REFERRED TO PSYCHOLOGICAL SERVICES FROM EMPLOYEE ASSISTANCE PROGRAMS By William Robert Meermans, Jr. A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Counseling, Educational Psychology, and Special Education 1992 ABSTRACT PRETREATMENT DROPOUT AMONG PERSONS REFERRED TO PSYCHOLOGICAL SERVICES FROM EMPLOYEE ASSISTANCE PROGRAMS By William Robert Meermans, Jr. The primary objective of this study was to identify variables related to pretreatment dropout among clients referred from employee assistance programs (EAPs) to psychological services. Hypotheses were tested to analyze relationships between pretreatment dropout and (a) client satisfaction with the EAP assessment and referral process, (b) attitudes toward seeking professional psychological help, (c) psychiatric helpseeking motivation, (d) previous counseling experience, and (e) client demographic characteristics. Hypotheses pertaining to motivation and attitudes were based on Krause’s (1966) theory of treatment motivation and Fischer, Winer, and Abramowitz’s (1983) psychiatric helpseeking model. The sample consisted of 64 female and 43 male clients who had scheduled an intake appointment to meet with a treatment provider. Following a routine EAP assessment and referral interview, subjects completed the researcher’s questionnaire, which included the Attitudes Toward Seeking Professional Psychological Help (ATSPPH) scale, the Client Satisfaction Questionnaire (CSQ-B), demographic Items and II‘C‘I'. (or; imp: ref IIII mu m VIE William Robert Meermans, Jr. items, and scales devised by the researcher to measure motivation and social pressure to seek treatment. Chi-square tests, one-way ANOVA tests, and discriminant function analyses were used to analyze the data. The results of this study indicated that psychiatric helpseeking motivation accounted for 79% of the total explained variance between the referral-noncompletion and the referral- completion groups. Confidence in treatment emerged as the most important motivational difference between the two groups. The referral-noncompletion Igroup expressed slightly less satisfaction with the EAP assessment and referral process than did the referral- completion group. Among demographic variables only age and marital status differentiated the two groups. No significant differences were found regarding psychiatric helpseeking attitudes, social pressure to seek treatment, or previous counseling experience. Finally, 25% of the sample dropped out before they completed their mental health referral. Seventy-four percent of this group could be identified retrospectively from their responses to the scale created to measure Krause’s theory of treatment motivation. Fischer, E. H., Winer, D., & Abramowitz, S. I. (1983). Seeking professional help for psychological problems. In A. Nadler, J. D. Fisher, & B. M. DePaulo (Eds.), New directions in h in : Vol. 3. Applied perspectives on help-seeking and :rgggjxjng (pp. 163-185). New York: Academic Press. Krause, M. S. (1966). A cognitive theory of motivation for treat- ment. [he Jggrnal of General Psychology, 15, 9-19. This dissertation is dedicated to my parents, who dedicated themselves to the betterment of our family. Thank you, Mom and Dad. iv ACKNOWLEDGMENTS I wish to express my appreciation for the contributions each of the following individuals has made to my dissertation research: To Dr. Linda Forrest, chairperson of my dissertation committee, whose dedicated, intelligent, and patient guidance of this project over many months and many hurdles was essential to its completion. To Dr. William Hinds, dissertation committee member, who helped me launch this project when it was hardly more than an intriguing, but undeveloped, idea. To Dr. Richard Johnson, dissertation committee member, whose knowledgeable and helpful discussions were just what I needed when my ‘thoughts became Inired 'hi statistical quandaries and research design problems. To Dr. John Powell, dissertation committee member, who first provided me with expert clinical supervision during my employment as an EAP practitioner, and who later directed that same thoughtful attention toward improving this study. To Dr. Jun Shin, my statistical and computer consultant, who spent many, many hours processing questionnaire data to provide me with results to ponder. To Susan Cooley, my editorial consultant and word processor, who combined remarkable productivity with meticulous attention to the preparation of each successive draft. To Dr. Anne Meermans, my wife and greatest supporter throughout this long project, whose love, sense of humor, and interest in my work helped me to see this dissertation through. I also wish to express my appreciation to the administrators, clinicians, support staff, and clients of the four Michigan employee assistance programs that took part in this research. I am deeply grateful for the effort they put forth to generate the data needed for the analyses of my research hypotheses. The four programs were: the Employee Service Program (State Government of Michigan, Lansing and Detroit), the Associate Assistance Program (Sparrow Hospital, Lansing), the Employee Assistance Center (Grand Rapids), and the Michigan State University EAP (East Lansing). Finally, I want to acknowledge the financial support given to me by the American Psychological Association. A portion of the cost of this research investigation was defrayed with an 1991 APA Dissertation Research Award. vi TABLE OF CONTENTS LIST OF TABLES ..... . .................. Chapter I. THE PROBLEM ..................... Statement of the Problem ............. Purpose of the Study ............... Importance of the Study .............. Overview of the Study ............... II. REVIEW OF THE LITERATURE .............. Premature Termination Research .......... Pretreatment Dropout Rates ........... Etiology of Pretreatment Dropout ......... Critique of the Etiological Research Literature . . Dropout Prevention Research ............ ATSPPH Scale Studies of Psychiatric Helpseeking Attitudes .................... Pretreatment Dropout: A Theoretical Perspective The Fischer, Winer, and Abramowitz Model The Krause Cognitive Theory of Motivation for Treatment ................... Research Questions ................ Summary ...................... III. DESIGN AND METHODOLOGY ............... Description of the Setting ............ Assessment and Referral Interviewing ....... Interviewer Characteristics ............ The Sample .................... Procedures for Collecting Data .......... Determination of Subject Follow-Through Status . Procedures to Protect the Anonymity of Subjects . Instruments .................... Researcher’s Measure of Krause’s Conditions for Seeking Mental Health Treatment ........ vii Researcher’s Measure of Subject-Reported Social Pressure ................... Client Satisfaction Questionnaire ........ The Attitudes Toward Seeking Professional Psychological Help Scale ........... Subject Personal Data Sheet ........... Research Design and Statistical Procedures for Analysis of the Data .............. Hypotheses .................... Discriminant Function Analysis .......... Summary ...................... IV. ANALYSIS OF THE DATA ................ Overview of the Research Questions ........ Preliminary Analysis of the Data ......... Description of the Research Sample ....... Correlations Among All Research Variables . . . . Analysis of Classification Precision ...... Main Analyses ................... Hypothesis 1 .................. Hypothesis 2 .................. Hypothesis 3 .................. Hypothesis 4 .................. Hypothesis 5a .................. Hypothesis 5b .................. Hypothesis 6a .................. Hypothesis 6b .................. Hypothesis 7a .................. Hypothesis 7b .................. Hypothesis 7c .................. Discriminant Function Analysis .......... Summary ...................... V. SUMMARY AND CONCLUSIONS ............... Summary of Findings ................ Demographic Variables .............. Satisfaction With EAP Service .......... Helpfulness of Previous Counseling Experience . . Attitudes Toward Seeking Mental Health Treatment ................... Psychiatric Helpseeking Motivation ....... Perceived Social Pressure to Seek Professional Mental Health Care .............. Prediction of Pretreatment Dropout in EAP Mental Health Referrals ............ Discussion of Findings .............. viii 121 122 122 122 123 124 Page Demographic Variables .............. 124 Satisfaction With EAP Service .......... 126 Previous Counseling Experience ......... 127 Attitudes Toward Seeking Mental Health Treatment ................... 129 Psychiatric Helpseeking Motivation ....... 133 Perceived Social Pressure to Seek Professional Mental Health Care .............. 138 Prediction of Pretreatment Dropout in EAP Mental Health Referrals ............ 141 The Decision to Seek Help--The Helpseeking Model ..................... 145 Limitations of the Study ............. 148 External Validity ................ 148 Design and Methodology ............. 149 Implications for Practice ............. 152 Recommendations for Future Research ........ 154 APPENDICES A. INSTRUCTIONS FOR EAP INTERVIEWERS .......... 158 B CLIENT COVER LETTER ................. 160 C. FOLLOW-UP LETTER .................. 161 D CLIENT INCENTIVE MEMO ................ 162 E INTERVIEWER INCENTIVE MEMO ............. 163 F. QUESTIONNAIRE .................... 164 REFERENCES ......................... 173 ix Table \IOTU'I-wa 10. 11. 12. 13. 14. 15. 16. LIST OF TABLES Correlations Among Main Research Variables ...... Results of Subgroup Classification Analysis ...... Demographic Summary of the Research Sample ...... Satisfaction Score (CSQ-8 Scale) by Subject Group . . . Previous EAP Use by Subject Group ........... Previous Counseling Experience by Subject Group . . . . ATSPPH Scale Composite and Factor Scores by Subject Group ........................ Krause’s Conditions of Psychiatric Helpseeking Motivation: Dichotomous Scale Scores ........ Krause’s Conditions of Psychiatric Helpseeking Motivation: Continuous Scores ........... Perceived Social Pressure Level by Subject Group Subject-Perceived Social Pressure Levels Among Subjects With Less Favorable Psychiatric Helpseeking Attitudes ................ Subject-Perceived Social Pressure Levels Among Subjects Not Satisfying Krause’s Conditions of Psychiatric Helpseeking Motivation ........ Discriminant Analysis Summary: All Significant Variables ...................... Classification Results: All Significant Variables Discriminant Analysis Summary: Krause Variables Classification Results: Krause Variables ....... Page 85 87 90 92 93 95 97 99 101 103 105 106 108 110 112 114 CHAPTER I THE PROBLEM For a growing number of individuals and their families, the route to professional psychological help passes through an employee assistance program, or EAP (Sonnenstuhl, 1986). Along with its other services, the modern EAP provides assessment and referral assistance to persons seeking relief from emotional and behavioral problems (Franz, 1986; Masi, 1984; Wrich, 1980). Although persons request help with other types of problems, surveys of a number of programs have indicated that the majority ask for assistance with mental health concerns (Presnall, 1981; Sonnenstuhl, 1986). These helpseekers are turning to EAPs for reasons that include uncertainty about the potential helpfulness of psychological services, confusion about available treatment options, and, in some instances, the need to obtain EAP authorization before using managed employee mental health care benefits (Roman & Blum, 1988). Statement of the Problem EAPs have flourished since the early 19805, when an estimated 2,500 or more were in operation in the United States (Byers, 1979; Roman, 1988; Witti, 1980). Larger programs respond to the needs of hundreds of clients a year. Collectively, EAPs annually assess and refer a considerable number of troubled individuals to mental health treatment programs. Indications are that the EAP assessment and referral process successfully guides the majority of these helpseekers into appropriate professional services (Brennan, 1982). Nevertheless, a sizable minority who express the intention of seeking treatment, and who stand to benefit from doing so, apparently fail to follow through. As the review of the literature in Chapter II will show, perhaps as many as one in four does not persist to the point of beginning with a treatment provider as planned. The pretreatment attrition of this group is the subject of this research investigation. Pretreatment attrition (commonly referred to in the literature simply as "dropout") can create significant problems for EAP clients, EAP programs, and the mental health care centers that serve as EAP referral resources. For EAP clients, dropping out may mean that they will fail to receive timely professional help with psychological and social conflicts that they have been unable to manage on their own. Those who fail to get needed professional aid may deteriorate until an outside intervention becomes imperative and more involved. Moreover, while these individuals remain in an impaired state, they are likely to have a negative effect on family members, friends, co-workers, and even members of their community (Hochstadt & Trybula, 1980). As for difficulties created for EAPs, the failure of clients to begin recommended psychological treatment may be evidence that their readiness for 'treatment, was misjudged during the assessment and referral process. The review of literature, however, will show that past pretreatment dropout studies have been largely inconclusive concerning the factors that thwart the desire to obtain psychological help. Consequently, EAP assessment and referral interviewers have had to rely on their best "clinical judgment" to anticipate what is apt to deter a particular client from completing a referral for needed assistance. Further research is needed to understand these deterrents to successful referral so that they can be anticipated and addressed systematically with clients during the assessment and referral process. Finally, problems are created for treatment agencies and those whom they serve when prospective clients miss (u: belatedly cancel intake appointments (Farrell, 1981). These schedule changes waste the valuable time of professional and support staff alike. In the current era. of' cost accountability and budget reductions, these disruptions drag down efforts to hold the line on the cost of delivering treatment services (Swenson & Pekarik, 1988). No less important, they also delay the provision of (flinical services to others in need who must spend time on a waiting list (Kluger & Karras, 1983). Purpose of the Study This researcher examined pretreatment dropout among EAP clients who agreed to be referred to professional mental health treatment. Those who began treatment were compared with those who did not, in terms of variables that have been linked to pretreatment dropout among psychiatric helpseekers. The findings of this research should contribute to the understanding of pretreatment dropout and should indicate which aspects of the decision to seek treatment deserve special attention during the EAP assessment and referral process. Importance of the Study The findings of this investigation should help to depict a clearer picture of pretreatment dropout among EAP clients than currently exists. In particular, this research was designed to identify variables involved in psychiatric helpseeking that are correlated in varying degrees. with pretreatment. dropout in this group. This knowledge, in turn, should help sensitize EAP practitioners to potential deterrents to the completion of appropriate referrals, deterrents that they would want to try to eliminate or minimize during the assessment and referral process. The findings of this study are likewise potentially useful to professionals in non-EAP settings who want to understand and improve the treatment readiness of their clients before referring them to psychological services. Medical personnel, members of the clergy, and school psychologists are all examples of allied providers of human services whose work with people sometimes involves preliminary assessment and referral for mental health treatments. The counselors and clinicians to whom these clients are actually referred also have a potential interest in the findings of this study. It is their professional efforts, after all, that may be assisted or hindered by the degree of their clients’ receptiveness to treatment. Finally, the findings of this study may be useful to persons involved in the promotion of healthy lifestyles and mental health education. A clearer understanding of how troubled individuals decide to seek or reject mental health treatment may help these specialists deliver more persuasive messages about the potential benefits of timely involvement in mental health interventions. The present study is also important because the data that were collected have the potential to shed light on several issues of theoretical interest. First of all, the relationship between psychiatric helpseeking attitudes and actual helpseeking behavior has received scant attention in past research investigations (Fischer, Winer, & Abramowitz, 1983). Knowledge of these attitudes ought to help in predicting such behavior; however, little is known about the extent to which attitudes toward seeking professional psychological help do, in fact, relate to subsequent helpseeking behavior. Second, Fischer et al. (1983) formulated a model of psychiatric helpseeking. Empirical research is needed to evaluate the soundness of their model. Third, Krause (1966) proposed a theory of psychotherapy motivation that explicitly set forth what he believed to be the three necessary conditions for deciding to seek psychotherapy. No» later studies could be found, however, that examined the validity of his theory, in spite of its potential value in the refinement of conceptualizations about psychiatric helpseeking and its obvious practical applications in mental health prevention, referral, and treatment services. Finally, the data obtained in the present study should help to clarify the role of social pressure in the EAP assessment and referral process. Overview of the Study This research investigation examined pretreatment dropout among a sample of EAP clients who were referred for professional mental health treatment. Pretreatment dropout tends to create significant problems for the referred client, the referring EAP, and the scheduled treatment provider. The present study was also undertaken to clarify several important theoretical issues that pertain to psychiatric helpseeking generally and to pretreatment dropout spe- cifically. In Chapter II the review of the literature is presented. Particular attention is given to research that has addressed the incidence and etiology of this behavior. In addition, theoretical considerations are taken up and a number of research hypotheses advanced. In Chapter III the methodology for this study is discussed. This discussion is organized around a description of the setting, the description and selection of the sample, instrumentation, the procedures for collecting data, the design of the research, and the analysis of data. The results of the data analyses are presented in Chapter IV. The results of three preliminary analyses of the data are presented, after' which the results of hypothesis testing and discriminant function analyses are presented. In Chapter V the study is summarized and the research questions are discussed in light of the findings. Theoretical and practical implications of the findings are discussed as well. Limitations of the study are identified, and recommendations for future research are offered. CHAPTER II REVIEW OF THE LITERATURE In this chapter, research relevant to pretreatment dropout in EAP clients is reviewed. Because almost no researchers have investigated attrition irI this group, the literature review draws primarily on the findings of related research. The review begins with an examination of the extent of the dropout problem across a variety of outpatient mental health care settings. Following that, studies that have explored the etiology of pretreatment dropout are reviewed. These studies are organized around the major variables of the etiological research. Next, several pertinent dropout prevention studies are discussed, followed by a review of selected psychiatric helpseeking attitude research. This chapter concludes with the presentation of a theoretical perspective and a series of research questions based on 'that perspective and the literature review. Premature Termination Research Premature termination research has focused primarily on the frequency, etiology, and prevention of client departure from mental health treatment before the completion of services. Most investigators have examined premature termination in relation to clients who complete one or more outpatient counseling or psychotherapy sessions (Baekland & Lundwall, 1975; Swenson & Pekarik, 1988). Comparatively few have looked at clients who complete an initial screening interview or intake, but drop out before their first treatment session (e.g., Betz & Shullman, 1979; Epperson, 1981). A few researchers also have focused on persons who inquire about treatment services but do not follow through to an initial intake interview (e.g., Hochstadt & Trybula, 1980; Kluger & Karras, 1983). Rodolfa, Rapaport, and Lee (1983) suggested that these three groups of premature dropouts probably represent three distinct client populations. Consequently, research findings associated with one group may not be generalizable to the other two. Several investigators have noted, for example, that persons who drop out before beginning treatment may do so for reasons different from those that cause premature termination during treatment (Baekland & Lundwall, 1975; Betz & Shullman, 1979; Kirk & Frank, 1976). In keeping with the time frame of persons referred to treatment by EAPs who Idrop out before 'treatment begins, only pretreatment dropout studies are discussed in this literature review. Pretreatment Dropout Rates According to two recent reports (Lowman, DeLange, Roberts, & Brady, 1984; Swenson & Pekarik, 1988), reliable estimates are not yet available concerning the percentage of persons who cancel, miss, or decide not to schedule an initial intake appointment after 10 contacting a provider about mental health treatment. It seems predictable, however, that varying dropout percentages would be found to be associated with preintake dropout, given differences in the client samples and the treatment settings that have been examined. After reviewing the literature, Lowman et al. (1984) concluded that 15% to 40% of all persons who contact a mental health treatment source (h) not follow through to an intake interview. Several older studies have reported higher dropout percentages (Krause, 1966; Overall & Aronson, 1963; Raynes & Warren, 1971a, 1971b), whereas a more recent quasi-experimental research investigation (Hochstadt & Trybula, 1980) found that 55% of a nonintervention control group scheduled, but did not attend, an intake interview. II recent study conducted at an urban community mental health center reported a dropout rate of 34% before the initial intake interview (Kluger & Karras, 1983). Only a few researchers have examined the dropout rate of clients who completed an intake interview but failed to follow through to the first session of individual counseling. Betz and Shullman (1979) examined a sample of clients who sought assistance at a large university counseling center. They found that 24% did not follow through to recommended counseling. Krauskopf, Baumgardner, and Mandracchia (1981) reported that 19% of their university counseling center client sample prematurely terminated after“ an intake interview' and before the first session of recommended counseling. Epperson (1981) partially replicated these two investigations and found that 25% of a sample of applicants for 11 university counseling center service failed to return for the first or second counseling session following an intake interview. Rodolfa et al. (1983), using a more restrictive definition of referral recommendations than those in the three previous investigations, recorded only a 7% dropout rate between the intake interview and the first session of counseling at a university counseling center. Turning to data on premature termination and the EAP client, Brennan (1982) reported on the dropout percentages of several groups of EAP clients who were referred to fOrmal treatment services but did not follow through to an intake interview. In one sample of 317 clients, approximately 30% did not follow through to attend an intake session with the recommended treatment provider. In a second sample, the dropout rate was 18% for subjects who complied with his request to complete a pre- and posttest attitude measure. For subjects who did not cooperate in completing the attitude measure a second time (20% of his sample), the dropout rate was 37%. Taken as a whole, the findings of these studies suggest that a sizable proportion of the group who contact a mental health treatment center regarding possible treatment do not follow through to treatment for one reason or another. It is important to keep in mind, however, that an incomplete picture of premature termination may result when dropout figures are based on data gathered only between the intake interview and the beginning of treatment or, for that matter, only between the beginning and end of treatment. In either case, by not reflecting client dropout that may have occurred 12 earlier, dropout figures recorded during these later intervals could significantly underreport the population that might have gone on to benefit from treatment. In the studies reviewed above, the observed dropout rates probably underreport the actual extent of client attrition before treatment at the data collection sites. The dropout problem is muddled further because figures are unavailable regarding clients who drop out but return at a later time for assistance (del Gaudio, Carpenter, Stein, & Morrow, 1977), clients who drop out after deciding that they do not need or want to begin treatment (Glassman, 1975; Noonan, 1973; Rodolfa et al., 1983), and clients who drop out but seek treatment elsewhere (Lowman et al., 1934). Etiology of Pretreatment Dropout Researchers have examined a variety of variables to understand why prospective counseling and psychotherapy clients prematurely end their contact with treatment providers. Quite a few have looked for answers among certain client demographic variables. Raynes and Warren (1971b), for example, examined client age and found that, except for white females, persons past 40 years of age were more likely than those under 40 to attend scheduled intake interviews at a city hospital psychiatric outpatient department. Other research- ers, however, have failed to uncover an association between age and attendance at scheduled intake appointments (Noonan, 1973; Rosenthal & Frank, 1958; Williams & Pollack, 1964). The research data regard- ing client sex and follow-through show a similar pattern. Noonan 13 (1973), Rosenthal and Frank (1958), and Williams and Pollack (1964) found no relationship, whereas Raynes and Warren (l971b) found women to be more likely than men to attend an initial intake session. Raynes and Warren (1971b) also observed a relationship between race and follow-through, with a higher percentage of whites keeping an intake interview than blacks (66% versus 54%). King (1979) found that racial similarity between an intake team and juvenile offenders and their families had a significant impact on whether black families began a program of outpatient counseling. Other investi- gators, however, have reported no differences concerning the follow- through rates of black and white applicants for treatment services (Rosenthal & Frank, 1958; Williams & Pollack, 1964). The same inconsistent pattern of ‘findings has emerged when other demographic variables have been examined. Socioeconomic status has failed to predict reliably who will attend and who will cancel or miss an initial appointment. Burgoyne, Acosta, and Yamamoto (1983), Errera, Davenport, and Decker (1965), Grieves (I978), Rosenberg and Raynes (1973), Rosenthal and Frank (1958), and Weighill, Hodge, and Peck (1983) all reported that middle- and upper-class clients were more likely than lower-class clients to follow through to an intake interview. Other researchers, however, have found no such association (Carpenter, Morrow, del Gaudio, & Ritzler, 1981; Gaines, 1978; Gould, Paulson, & Daniels-Epps, 1970; Kluger & Karras, I983; Noonan, 1973; Williams & Pollack, 1964). Raynes and Warren (l971b) and Noonan (1973) found no ewidence to link. marital status to follow-through. Previous mental health 14 treatment has been related to keeping an intake appointment in two client samples (Carpenter et al., 1981; Rogawski & Edmundson, 1971), but not in three others (Errera et al., 1965; Raynes & Warren, 1971b; Swenson & Pekarik, 1988). In addition to demographic variables, researchers have examined several personality and clinical variables. In a study of missed intake appointments at 21 large university counseling center, Kirk and Frank (1976) studied selected personality characteristics of students applying for service. They' were able to discriminate students who did not attend their intake appointment from those who did only on the basis of impulsiveness, higher impulsiveness being positively correlated with nonattendance. In a similar study conducted by Gaines (1978) at a child guidance clinic, anecdotal evidence suggested that the level of applicant motivation for treatment might--with more precise measurement--discriminate between the attending and the early terminating groups. Pfouts, Wallach, and Jenkins (1963) also looked at pretreatment motivation and found that low levels of motivation interacted with client attitude toward the referral source. More of the minimally motivated persons in their sample who had positive attitudes toward the referral agent followed through to treatment than did persons who were only minimally motivated and had negative attitudes. Similarly, Baekland and Lundwall (1975) reported that patient refusal of physician referrals to mental health services was correlated with negative attitudes on the part of the physician toward the patient. Raynes 15 and Warren (1971a) found in one study that patients simply referred by a physician or a social agency were less likely to follow through to treatment than patients who referred themselves. In another study, however, they found source of referral to be unrelated to attendance at an outpatient psychiatric clinic intake session (Raynes & Warren, 1971b). Rosenthal and Frank (1958) found that patients referred by a psychiatrist or a psychiatric facility were likely' to follow ‘through to outpatient therapy. Finally, both Noonan (1973) and Gould et al. (1970) reported that persons who gave specific reasons for wanting professional help were more likely than persons with vague or evasive reasons to attend intake interviews. Noonan’s (1973) study is also interesting in that persons who failed to attend intake interviews were queried and gave the following reasons for ‘their absence: (a) no reason, including alleged forgetfulness--39%; (b) improvement in symptoms--35%; and (c) fear and anxiety about the intake and treatment process--23%. Although the remission of symptoms may influence a client to decide against treatment, several studies have suggested that the severity of symptoms and presenting problems at intake is unrelated to the return rates of clients for treatment (Epperson, 1981; Gould et al., 1970; Kirk & Frank, 1976; Rosenthal & Frank, 1958). Calhoun and Selby (1974), however, found evidence in an analogue study that more severely disturbed individuals tend to avoid professional mental health care. In recent years, a small group of researchers has investigated the influence of the experience level and/or sex of the intake 16 interviewer and the scheduled counselor on premature client termination. The results of a study conducted by Betz and Shullman (1979) indicated that both male and female clients were less likely to return for counseling when initially interviewed by a male as opposed to a female intake interviewer. Specifically, clients referred by male intake interviewers to male counselors were significantly less likely to begin counseling than clients who had been assessed by female intake interviewers and/or referred to female counselors. Betz and Shullman did not, however, find that the client dropout rate was related to the experience level of the intake interviewer. By comparison, Epperson (1981) found that clients who saw male intake interviewers returned at higher rates than those seen by female interviewers. Krauskopf et al. (1981), on the other hand, found that neither the sex of the intake interviewer nor the sex of the scheduled counselor had a significant effect on the dropout rate. Rodolfa et al. (1983) sought to clarify these discrepant findings concerning sex in a carefully designed study. The results of their research indicated that the length of the intake interview, the number of days from the intake interview to the first counseling appointment, and the experience level of the scheduled counselor were the only factors significantly related to the client dropout rate. Intake interviewer sex and experience level did not affect, dropout rates. 17 Finally, several investigators have looked at accessibility (Albers & Scrivner, 1977) and administrative variables. In contrast to the findings of Rodolfa et al. (1983), other studies of the length of time between initial contact with the treatment source and the scheduled intake appointment have generally revealed nonsignificant correlations. Gaines (1978), Gould et al. (1970), Kluger and Karras (1983), Noonan (1973), Rosenthal and Frank (1958), and Swenson and Pekarik (1988) all found no significant association between time on a waiting list and attendance at a scheduled intake interview. Raynes and Warren (1971a), on the other hand, found that scheduling delays of more than 15 days were associated with increased attrition at a hospital outpatient psychiatric program, findings that were supported by a second study at the same facility (Raynes & Warren, 1971b). Coombs (1986) also found that clients who were put on the waiting list of an outpatient treatment center tended to discontinue. Surprisingly, almost no investigators have accounted for the cost of treatment to the client. One who has, Solovei (1987), discovered that intake and session fees were correlated with discontinuing treatment after intake, session fees explaining the greatest amount of variance-~though still small--in a discriminant analysis involving 13 variables. Lack of attention to fees may reflect the possibility that many of the data collection sites, particularly university counseling centers and community mental health centers, might have offered treatment services free of charge or on an ability-to-pay basis. 18 Last of all, the inconvenience and the expense of traveling to clinics for service have been examined in at least two studies. Gaines (1978) reported that distance traveled ix> a child guidance clinic had no effect on attendance at the initial intake interview. This same finding was reported by Swenson and Pekarik (1988) in an investigation of clients scheduled for intake at a community mental health center. To summarize, researchers have examined a variety of demographic, personality, accessibility, interviewer, and adminis- trative variables as potential factors in pretreatment client termination. In the studies reviewed here, most of these variables were found to have statistically nonsignificant or inconsistently positive and negative correlations with client dropout. The empirical evidence to date is tentative but 'suggests that the following variables are associated with pretreatment dropout: client pretreatment motivation, client attitudes toward the referral source, and the specificity of client reasons for seeking help. In addition, the anticipated cost of treatment appears to be a consid- eration for some individuals. The studies by Noonan (1973) and Rodolfa et al. (1983) also suggested that pretreatment dropout may be related to spontaneous improvement, fear and anxiety about the intake and treatment process, the length (H: the intake interview, and the experience level of the scheduled therapist. 19 Critique of the Etioloqical Research Literature The pretreatment dropout research suggests that perhaps one- third of all persons who contact mental health clinics about obtaining services do not persist to the point of entering formal treatment. More than three decades of investigation have produced rather limited insight into the behavior of this group. Several shortcomings of the research seem to have slowed progress toward a better understanding of this phenomenon. To begin with, much of the research consists of descriptive, correlational studies that have focused on demographic variables. Demographic variables have understandably been attractive to investigators of pretreatment dropout. Demographic data such as age, sex, race, education, income, and prior treatment experience are all generally easy to collect from client records (n1 file at mental health centers. Furthermore, the collection, analysis, and publication of this type of client data are unlikely to harm the interests of the research subjects and, therefore, impose relatively easily managed ethical strictures on researchers (e.g., American Psychological Association, 1982). The discovery of correlations between demographic variables and pretreatment dropout irI a clinic sample also may increase the ability of treatment providers to identify potential dropouts. The limitation of analysis involving demographic variables, however, is that, in and of themselves, demographic variables do not explain why one person or group terminates before treatment begins and another does not. For example, discovering that young adults 20 tend to terminate before treatment at a: rate higher than that of middle-age adults (e.g., Raynes & Warren, 1971a) suggests that a relationship exists between age and dropout behavior. This finding does not, however, reveal what the actual age-related factor is that influences these two groups to drop out or follow through to treatment at different rates. The same ambiguity is true of correlations observed between follow-through and sex, income level, race, or any other demographic variable. If demographic variables are related in) pretreatment dropout merely in an adventitious way, then other kinds of variables must be more centrally involved in an individual’s decision to drop out before treatment begins. This review (H’ the literature suggests that these nondemographic variables are related to pretreatment dropout in ways that have yet to be adequately explored and understood. Several. studies, for' example, found no association between dropping out and the length of time clients had to wait for their first treatment session (e.g., Gaines, 1978; Gould et al., 1970; Kluger & Karras, 1983). It is reasonable to imagine, however, that most individuals have an upper limit to their tolerance for treatment delay. Did the clients in these studies persist because they did not perceive the delay to be excessive by local standards or because they knew of'ru) acceptable, alternative treatment providers who could see them sooner? 'These particular conditions were not explored and make it questionable to generalize the findings of these studies to situations of longer delay or greater 21 freedom to choose a treatment provider. In all fairness to these researchers, it must be noted that their interest lay in examining naturalistic client data, not in experimentally manipulating a variable such as tolerance for treatment delay. Gaines’s (1978) research also points up the omission of another important comparison group from the research on pretreatment dropout (Albers & Scrivner, 1977). These are the potential clients who can be imagined to have considered using the services of a clinic, only to have decided not to. To be sure, it might be difficult to study this hidden group, and it could be argued that they are not, strictly speaking, dropouts from treatment anyway. But including representatives of this unaccounted-for group of potential clients in the research sample, whenever feasible, would permit. a more thorough examination of the relationship between the use of services and deterrents such as distance and treatment delay. Finally, Rodolfa et al. (1983) pointed out that comparisons among many of the studies are suspect because intake procedures and the definition of premature client dropout were not the same in all cases. As already noted, this lack of uniformity may have led to the selection of substantially dissimilar samples and the investigation of nonequivalent client populations. Under these conditions, it should not be surprising to discover that apparently similar investigations yield dissimilar findings. 22 Dropout Prevention Research The body of research that has arguably come closest to identifying the variables most responsible for pretreatment dropout is the work of several investigators who have experimented with strategies for reducing the rate of missed intake appointments. These investigators have examined the effect of telephone prompts, letter prompts, orientation groups, and orientation pamphlets and letters in a series of quasi-experimental studies of clients scheduled for intake sessions at community mental health treatment centers. One group, Hochstadt and Trybula (1980), compared the effectiveness of a reminder letter received 3 days before the scheduled intake, a reminder telephone call 3 days before the intake, and a reminder telephone call 1 day before the scheduled intake. All of these prompts simply reminded the prospective client of the date and time of the intake appointment and the name of the intake worker. Hochstadt and Trybula found that all three interventions reduced the rate of preintake dropout from the 55% observed for a nonintervention control group. Both types of 3-day reminders reduced the dropout rate to 32%, while the I-day telephone prompt reduced the rate to just 9%. Kluger and Karras (1983) took Hochstadt and Trybula’s interventions a step further by providing clients with information about what would take place at the intake session, a variation of an approach for modifying client expectancies about psychotherapy for which there is empirical support (Tinsley, Bowman, & Ray, 1988). 23 Kluger and Karras compared the effectiveness of a brief orientation statement prepared for this purpose and read to clients during the scheduling of their intake appointment with the effectiveness of a simple l-day phone reminder and a l-day orientation statement plus phone reminder. Unlike Hochstadt and Trybula, they found that only the orientation statement read during the scheduling of the intake appointment improved the dropout rate in comparison with that of a nonintervention control group--28% versus 56%. However, they were unable to reach most clients by phone. Had they been able to reach more clients by phone, they believed that the 20% dropout rate recorded for the l-day orientation statement plus phone reminder would have been statistically significant and the largest improvement over the control group rate. Swenson and Pekarik (1988), noting the possible difficulty and expense of trying to contact clients by phone, attempted to improve the preintake dropout rate through the use of inexpensive mailed appointment reminders and orientation letters. As in the Kluger and Karras (1983) study, the orientation letter described what a prospective client could expect to take place at the intake interview. Swenson and Pekarik investigated the effect of timeliness by sending reminder letters and orientation letters that were received 3 days or 1 day before the intake appointment. They found that clients who received the 1-day orientation letter had the lowest dropout rate, 17%. The control group, who received neither an orientation letter nor a reminder, posted the highest dropout 24 rate of 43%. The dropout rates for clients sent the 3-day reminder, the 3—day orientation letter, and the 1-day reminder also were all lower than the 43% dropout rate of the control group; however, these differences were not statistically significant. The data did not support Swenson and Pekarik’s prediction that clients who received either a reminder or an orientation letter 1 day before their appointment would have lower dropout rates than clients whose letter arrived 3 days before their appointment. In other words, the length of time between receipt of a letter and the intake appointment was not a factor in the dropout rates observed in this study. What is noteworthy about all three of these investigations is that one or more of the experimental interventions significantly reduced the pretreatment dropout rate (compared to control group base rates). These interventions apparently acted upon one or more of the factors responsible for dropping out or following through to treatment. The researchers associated with these investigations speculated about what their treatment conditions may have affected. Orientation statements may have improved attendance at initial intake interviews by reducing unfavorable client misconceptions regarding the intake and treatment process, thereby reducing anxiety and its inhibiting effects (Kluger & Karras, 1983; Swenson & Pekarik, 1988). Telephone and letter appointment reminders may have served as cues for clients who otherwise would have forgotten their scheduled appointment time. Hochstadt and Trybula (1980) also suggested that telephone reminders may have reduced the dropout rate by extracting an overt commitment to attend the intake interview. 25 They noted that research on attitude-discrepant behavior in other settings has implied that clients who commit themselves to attending the intake interview will tend to conform their behavior to their stated intentions in spite of discrepant attitudes. Swenson and Pekarik (1988) also suspected that spontaneous improvement and environmental obstacles account for a percentage of pretreatment dropouts. Also noteworthy is that all of these researchers tested for interactions involving variables examined in the descriptive studies reviewed earlier in this chapter. Kluger and Karras (1983) tested and found no significant difference between their intervention groups and dropout rate as a function of race, sex, or age of clients; Hochstadt and Trybula (1980) found none for sex or age; and Swenson and Pekarik (1988) found none for sex, age, residence distance from the clinic, previous psychiatric treatment, referral source, waiting time until the intake appointment, or request for individual or family or couple intake. The absence of such interactions provides further evidence that these particular variables, at least under the circumstances of these investigations, seem to have little direct effect on whether a prospective client drops out or follows through to an initial intake interview. ATSPPH Scale Studies of Psychiatric Helpseeking Attitudes In addition to the preceding quasi-experimental studies, another area of research is particularly relevant to the present investigation. The studies in question have explored psychiatric 26 helpseeking behavior with a scale developed by Fischer and Turner (1970) to assess psychiatric helpseeking attitudes. With the development and publication of their scale, Fischer and Turner hoped "to clarify an attitude and personality domain which applies to one’s tendency to seek or to resist professional aid during a personal crisis or following prolonged psychological discomfort" (p. 79). The present investigation was intended 11) contribute to the growth of knowledge in this area as well. The Fischer and Turner instrument is known as the Attitudes Toward Seeking Professional Psychological Help (ATSPPH) scale. Since its publication, the ATSPPH scale has been used to clarify the relationship between psychiatric helpseeking attitudes and a variety of clinical, demographic, and behavioral variables. With it, Calhoun, Dawes, and Lewis (1972) and Calhoun and Selby (1974) demonstrated that severity of psychological distress and positive psychiatric helpseeking attitudes were inversely related in separate urban outpatient and undergraduate student samples. Kligfield and Hoffman (1979) used it to study the development of psychiatric helpseeking attitudes during the educational careers of medical students. They were surprised to find that male students developed significantly more positive attitudes as they progressed through the first 2 years of medical school. Female students, by comparison, entered school with more positive attitudes than their male classmates, but did not increase their positive attitudes during their 4 years of medical education. 27 As with the previously reviewed research on pretreatment dropout, demographic variables have received the most attention in research involving the ATSPPH scale. Lorion (1974), for example, examined socioeconomic level and psychiatric helpseeking attitudes and, contrary to prediction, found no class differences. Pfeiffer (1976) also found no statistically significant class differences in attitudes in a predominantly white sample of individuals not in therapyu Fischer and Cohen’s (1972) research likewise provided evidence that psychiatric helpseeking attitudes were not closely tied to socioeconomic class. Contradicting this trend, Wolkon, Moriwaki, and Williams (1973) found that middle-class blacks had significantly more positive attitudes toward psychotherapy than lower-class blacks. Considered as a whole, the findings of these studies are consistent with those of the previously reviewed studies that, by and large, showed no relationship between socioeconomic level and the actual use of mental health services. In other studies involving the ATSPPH scale, psychiatric helpseeking attitudes have been examined in relation to demographic variables that have included college major, religious orientation, and gender. Fischer and Cohen (1972) found in their sample of high school and college students that college juniors and seniors had more favorable attitudes than high school students and college sophomores and freshmen; that Jewish students had more favorable attitudes than either' Catholic CH“ Protestant students; and that psychology students had more favorable attitudes than students majoring in the humanities, physical sciences, and applied programs. 28 Beginning with Fischer and Turner’s (1970) original work to validate the ATSPPH scale, women have generally been observed to have more favorable attitudes than men toward seeking professional care (e.g., Dadfar & Friedlander, 1982; Kligfeld & Hoffman, 1979; Puig, 1979; Sanchez & Atkinson, 1983; Surgenor, 1985). Exceptions to this pattern were observed by Lorion (1974), Todd and Shapira (1974), and Zeldow and Greenberg (1979). Several investigators have related culture and nationality to psychiatric helpseeking attitudes. Sanchez and Atkinson (1983) found that Mexican-American college students who had a strong commitment to the Mexican-American culture expressed less favorable attitudes toward using professional counseling services than did their peers who had a weak commitment to both the Mexican-American and Anglo-American cultures. In a study similar to Sanchez and Atkinson’s, Rivas (1981) also found that acculturated Mexican- Americans had more positive helpseeking attitudes than traditional Mexican-Americans. In an investigation involving international university students, Dadfar and Friedlander (I982) found that the attitudes of European and Latin American students were significantly more positive than those of Asian and African students. In another study of student nationality and attitudes, Todd and Shapira’s (1974) analysis revealed that American college students had more favorable attitudes toward psychotherapy than did their British counterparts. 29 In the context of the present investigation, the most pertinent findings to emerge from ATSPPH scale research are those that relate global and subscale attitude scores to actual psychiatric helpseeking behavior. Most reports have indicated that the ATSPPH scale can discriminate persons who have previously sought professional help from persons who have not. Global-score differences between these groups, which strengthen the known-group validity of the scale, have been reported by a number of investigators (e.g., Cash, Kehr, & Salzbach, I978; Cooperman, 1983; . Dadfar & Friedlander, 1982; Fischer & Turner, 1970; Surgenor, 1985). Global-score differences, however, must be broken down into their component factors to observe more fundamental differences in attitudes between these two groups. Unfortunately, these details have been omitted from most published research reports. During the development of the ATSPPH scale, Fischer and Turner (1970) conducted three independent factor analyses that consistently identified four interpretable dimensions within the item responses of their samples: (1) recognition of personal need for professional psychological help, (2) tolerance of stigma associated with receiv- ing psychiatric help, (3) interpersonal openness regarding one’s problems, and (4) confidence in the mental health profession. Fischer and Turner suggested that the first factor, recognition of personal need for professional psychological help, together with the fourth factor, confidence in the mental health profession, probably represents the "essence of the attitude toward seeking professional help" (p. 88). Stigma tolerance and interpersonal openness, they 3O believed, are important aspects of the helpseeking orientation; however, they may not be as important in deciding to seek profes- sional help as the need and confidence factors. A study by Magnavita (1981), however, casts doubt on the order of importance of the ATSPPH subscale factors suggested by Fischer et al. (1983). Magnavita used discriminant analysis to analyze the variance associated with premature termination between the intake and first counseling sessions at a university mental health center. He found that the variable, tolerance of stigma associated with receiving psychiatric help, accounted for most of the explained variance (with a standardized discriminant function coefficient, or SDFC, of .547). Next in importance was recognition of personal need for psychotherapeutic help (SDFC = .386), followed by interpersonal openness regarding one’s problems (SDFC .346) and confidence in the mental health profession (SDFC = -.O77). With regard to the prediction of attrition, all of the subscale scores of the group who remained for counseling were slightly more positive than those of the group who dropped out, but these differences were nonsig- nificant. Similarly, a quasi-experimental research investigation by Brennan (1982) resulted 'hi inconclusive findings about the relationship between psychiatric helpseeking attitudes (assessed with the ATSPPH scale) and early attrition from psychotherapeutic assistance. Brennan discovered that members of his sample of persons referred by an EAP to treatment providers dropped out or 31 followed through to treatment virtually irrespective of their measured psychiatric helpseeking attitudes. To explain this unexpected finding, Brennan suggested that the motivating influence of the EAP assessment and referral interviews was so strong as to override the inhibiting effect of negative attitudes toward obtaining professional assistance. However, an equally if not more plausible explanation is that many persons who consult an EAP are experiencing some degree of external pressure to reform unacceptable behavior of one kind or another (see Brennan, 1982, pp. 26, 49, 54, and 57). Such social pressure may help to motivate them to seek professional assistance despite their negative helpseeking attitudes. In summary, a substantial body of research has linked favorable psychiatric helpseeking attitudes with (a) the predisposition to seek professional assistance to deal with psychological problems and (b) personal experience with such assistance. The ATSPPH scale and its subscales appear to be the most sophisticated measure of these attitudes currently available (the development and psychometric qualities of the scale are discussed in detail in Chapter III). Despite its potential significance, there has been a dearth of research involving the use of the ATSPPH scale scores as dependent variables in studies of actual psychiatric helpseeking. The use of the ATSPPH scale in the present investigation was intended to help fill in the limited existing knowledge about the attitude-behavior relationships in this area (Fischer & Turner, 1970; Fischer et al., 32 1983). The role of extrinsic motivation in this attitude-behavior relationship was also investigated. Pretreatment Drooout: A Theoretical Perspective In view of the overall inconclusive state of the etiological research literature, researchers need to give further attention to identifying the factors that actually determine whether a prospective mental health client will begin treatment their membership in the referral-noncompletion group or the referral-completion group. The predictor variables used in the discriminant analyses were selected from 5 of the 23 variables examined in the study. These five were selected for analysis because both hypothesis testing and correlation testing had established that they were associated with statistically significant differences between the two groups. The five variables were the subjects’ (1) confidence in treatment, (2) need for treatment, (3) tolerance for treatment demands, (4) age, and (5) satisfaction with the EAP assessment and referral service. Summary Volunteer subjects for this study were recruited from the clientele of four EAP programs in southern Michigan. EAP interviewers asked the subjects, who had received and agreed to pursue a referral for mental health treatment, to complete a questionnaire that surveyed their satisfaction with EAP service, their attitudes toward seeking professional psychological help, their motivation to seek treatment, their previous experience with counseling, and their feelings of being pressured to seek treatment. They were also asked to provide basic demographic information about themselves and to authorize their EAP to confirm whether they completed the referral with the intended provider. These data were used to test a number of hypotheses developed to answer the research questions on pretreatment dropout posed in Chapter II. Discriminant 79 function analyses, one-way ANOVA tests, correlational tests, and chi-square tests were performed to analyze the data. co AII si ma COI CHAPTER IV ANALYSIS OF THE DATA In this chapter, the results of the analyses of the data are reported. The research questions, around which the investigation was organized, are presented first. Next, the research sample is described, the correlations among the main research variables are presented, and the precision of the criterion variable is discussed. The results of the main analyses of the data are presented in the remainder of the chapter. Overview of the Research Questions This research investigation was undertaken to shed light on a number of variables thought to be related to completion or noncompletion of EAP-recommended mental health referrals. To explore some of the connections suggested by the review of EAP and related research, the following research questions were framed. 1. Are client demographic characteristics related to referral completion or noncompletion? To examine this question, one-way ANOVA or chi-square analyses were used to test for statistically significant differences in age, education, income, sex, race, and marital status between research subjects who did and did not complete an EAP-recommended referral. 80 81 2. Is client satisfaction with the EAP assessment and referral process related to referral completion or noncompletion? Data for the examination of this question were gathered through the administration of a standardized client satisfaction scale and a two-part questionnaire item. The data were analyzed with chi-square and one-way ANOVA tests. 3. Are EAP-recommended mental health referrals less likely to be completed when the client has previously had unfavorable experiences with mental health treatment? Chi-square and one-way ANOVA were used to test the data gathered to examine this relationship. 4. Do attitudes concerning individual beliefs about the need for treatment, the stigma of treatment, the effectiveness of treatment, and the demands of treatment predict follow-through behavior as postulated by the psychiatric helpseeking model of Fischer et al. (1983)? In the context of the present investigation, do negative attitudes about mental health treatment predict noncompletion of an EAP-recommended mental health referral? One-way ANOVA and discriminant function analysis were used to answer these questions. 5. What is the role of client-perceived social pressure in the completion or noncompletion of EAP-recommended mental health referrals? More specifically, does such pressure override unfavorable treatment attitudes and/or weak intrinsic helpseeking motivation in the completion of EAP referrals? One-way ANOVA tests were used to analyze these relationships. 82 6. Are the prerequisite ideational conditions of motivation to seek. professional psychological help specified by Krause (1966) present in persons who complete an EAP-recommended referral and absent in persons who do not? Chi-square and one-way ANOVA tests were performed on the data to answer these questions. 7. To what extent can EAP referral dropout be predicted at the conclusion of the assessment and referral interview? To what extent can clients who are at risk of not completing a recommended referral be identified at the conclusion of the assessment and referral interview? Which variables measured in this study are the best predictors of referral dropout? Discriminant function analysis was used to answer these questions. Preliminary Analysis of the Data Description of the Research Sample Two hundred fifty-six clients of the four cooperating EAPs agreed to fill out and return a copy of the questionnaire developed for this study. Of this group, 121 actually completed and returned their questionnaires, a 47.3% return rate for all four programs. Questionnaires received from 14 members of this group could not be used, primarily because of a breakdown in record-keeping at one of the EAPs. Questionnaires from the remaining 107 respondents supplied the data analyzed in this study. Twenty-seven of these 107 respondents were reported not to have completed their EAP referral, making the dropout rate for the group 25.2%. 83 As a group, the 107 individuals who completed and returned the researcher’s questionnaire had the following demographic profile. Sixty percent of them were women, 40% men. Most (82.2%) were Caucasian. They ranged in age from 19 to 59, with a mean age of 38.2. The majority (54.7%) were married. Their median 1990 family income was in the range of $25,000 to $35,000. They were employed in a wide variety of occupations, including corrections officer, university professor, cook, and shift supervisor. Most (86.9%) had continued their education after completing high school, 29% had earned a 4-year college degree, and 14% had completed a postgraduate degree. For 83%, this was their first experience with an EAP assessment and referral interview. Fifty-four percent, however, reported that they had previously received mental health counseling. Limited demographic data (sex, age, race, marital status, and prior counseling experience) were also obtained for the 135 EAP clients who accepted, but did not complete a questionnaire. As a group, these individuals differed appreciably from the respondents only in relation to two recorded characteristics: They were younger (33.7 versus 38.2), and they included a somewhat higher percentage of African Americans (27% versus 19%). The groups were very similar with respect to sex (both groups: 60% women, 40% men), marital status (respondents: 54.2% married, nonrespondents: 53% married; respondents: 16.8% never married, nonrespondents: 20% never married; respondents: 22.5% divorced, nonrespondents: 13% divorced), and prior counseling experience (respondents: 55.1% "yes"; nonrespondents" 53% "yes"). 84 Correlations Among All Research Variables A correlation matrix was computed, based on data from all variables included in the research investigation. 'The correlation matrix revealed that three main variables were significant at the .01 level in relation to the criterion variable: the CSQ-8 Satisfaction scale score (p = .26), the Krause Need for Treatment scale score (m = .33), and the Krause Confidence in Treatment scale score (3 = .42). One additional main variable, the Krause Tolerance for Treatment Demands scale score (3 = .22), was significant at the .05 level. Of the remaining 17 variables, only Education was significant at the .05 level. The complete correlation matrix is available for inspection in Table 1. _malysis of Classification Precision As described in Chapter III, the sample was analyzed to examine the homogeneity of the referral-completion group and the homogeneity of the referral-noncompletion group. This analysis was also performed to make an indirect assessment of the precision of the procedure for classifying referral outcome. For' the analysis, the sample data. were organized into two subsamples. One subsample was composed of data from the two EAPs that were well known to the researcher. The second subsample was composed of data from the other two EAPs. The data were then further broken down so that the referral-completion portions of the subsamples could be compared with each other and the referral- noncompletion portions compared with each other. The comparisons 85 mocm u FN> .oocooeoaxm mcmpomcsoo m30m>oem n n—> .oeooc_ n op> .maumum Foumcmz n mp> .oomm .8..v mi. .3. v m... amcmgonsoz aneco u ~N> .oocowconxm o~> .voucoeeooom neoc30nom _mccowom we Lonssz n m—> .Acozo*>eouc*o xom u op> .comuooavm :P> .oo< u mF> .xom u NP> .Aopmom Immmh .Aopmom Immwp.Ao_m0m Immmh .Ao_oom :mmm» .Aopmom Imamh .o_oom mmocwcmucaoo> ucom_o u o> .o_oom ocammoam _mmoom vo>moocom u m> .Aomsmcxv upmom mucmsoo acoEuoocp cop oocncoPOP n :> .Aomamcxo o_mom acoEucoch c? oocoomocoo n m> .Aomaoexv opmom ucoEuooLh Lou oooz u N> .Aopmom oIOmoo oom>com m .xox oo.P ~N> mo. 00.? —N> mo. eemm. oo.P om> mo. Fo.u mo.u oo.F mp> mo. or. mo. «Fm. oo.~ m~> kpw. mo. kkom. up.n up. oo.F n—> :o. m—. n—.n mo. NF.» *kom. oo.F o—> :—.n m9. emu. ——.u «P. :o.. m—.- oo.F m—> ~—.u mF.u m—.n mp. e—N. m—.n eom.n mo. oo.~ :F> n—.u m—. No. mp. op. :o.n new. eom. mo.u oo.— m—> mo. PF.- 90.- oo.. so. o—.u PF.u FF. oo. Po. oo.— ~F> mo.- FF. *kom. mo. mo.- mo.u #0.- kFN. oo. oo. mm. oo.— FF> wo.u no. {*mN. mo. 30.: mo. mo.n Mr. —P. so. ksm. seam. oo.— oF> ~—.a :o. mp. Fo.u mo.u Fo. mo. NF. oo.- oo. mo. ekom. seam. oo.F m> mo.u mo. FF. Fo. Pp.i mo.n ~o.u or. po.u mo. no. *knn. *eoq. aeoo. oo.P o> No. mp. *om. oo. :o. mo.- no.1 how. :P. so. «*Fm. «*Fm. «emc. heme. *kme. oo.— n> mo.u —o. o—.u Fo. ~o.: no. mo. mF.n no.u Fe. mo.u mp.u No.4 Po.n up.u :F.n oo.~ o> No. mo.u op. op. Po.n :o.n me.n No.- no.: co. m—.u mp.n oF.u mo.i *PN.I oF.u «o:. oo.F m> *um. po. ¥FN.n No. :o.: mo. eemm. emN.u so. mo.u No. mp. oo.n m—. no. mp. mo.u 30.: oo.P ¢> ewe. 50.: mo. . Po. mp.u Fo.o Po. mo.n :o. m—.. Po. up. NF. mF. op. up. «-.n mo.o emu. oo.P m> eemm. mo. s—. w—. oo. Fo.u 50.: ~—.u Fo.n mo.i NF. or. FF. oo.n mo. ekqm. mo.n no. eon. neon. oo.F u> emu. No.1 —o. eeom. Fo. No.1 mo. mo. mo. mp. mo. «Fm. no. N—. m—. NP. No. up. or. mp. FF. oo.F F> N~> FN> o~> mp> mp> 59> op> mp> 39> mF> ~F> ~F> op> m> o> ~> o> m) :> m> ~> F> .mo_nmmca> cecmomoc come ocOEm mcomuopoccooun.p o_nmh 86 involved scores from the ATSPPH scale, the CSQ-B scale, and each of the three Krause scales. One-way ANOVA tests performed on the scores for each scale revealed no significant subsample differences at the .05 level between either the two referral-completion portions or the two referral-noncompletion portions of the subsamples. The results of these tests and the associated descriptive statistics are reported in Table 2. These results showed that relative to the ATSPPH scale, the CSQ-8 scale, and the Krause scales the two referral- noncompletion portions were statistically equivalent samples from the same population. The results showed that this was also true of the two referral-completion portions. This comparability, in conjunction with the researcher’s confidence 'hi the outcome classifications provided by the two EAPs whose interviewers he knew to be diligent, suggested that subjects from all four EAPs had been classified with an acceptable degree of accuracy as well. Further evidence of the classification precision of referral outcomes is presented in the prediction-rate findings discussed later in this chapter. Tat 87 Table 2.--Results of subgroup classification analysis. Group Sites 1,2 Sites 3,4 Total E Sig. of E ATSPPH Scale Noncompletion .5640 .4597 N 20 7 27 Mean 66.48 62.79 SD 11.77 9.07 Completion 1.7600 .1881 N 46 34 80 Mean 61.83 65.43 SD 12.96 10.52 Krause Need for Treatment Subscale Noncompletion 4.1374 .0527 N ’ 20 7 27 Mean 3.75 2.29 3.37 gm 1 1.62 1.70 1.74 Completion .1083 .7430 N 46 34 80 Mean 4.62 4.51 4.58 SD 1.37 1.46 1.40 Krause Confidence in Treatment Subscale Noncompletion 3.4946 .0733 N 20 7 27 Mean 4.10 2.71 3.74 SD 1.59 1.98 1.77 Completion .4875 .4871 N 46 34 80 Mean 4.87 5.00 4.93 SN .98 .55 .82 88 Table 2.--Continued. Group Sites 1,2 Sites 3,4 Total E Sig. of E Krause Tolerance for Treatment Demands Subscale Noncompletion .1324 .7190 N 20 7 27 Mean 3.55 3.29 3.48 SD 1.64 1.70 1.63 Completion .0513 .8214 N 46 34 80 Mean 4.28 4.21 4.25 §Q 1.44 1.57 1.49 CSQ-B Scale Noncompletion .0015 .9691 N 20 7 27 Mean 24.40 24.29 24.37 SD 7.24 4.23 6.51 Completion .0716 .7897 N 46 34 80 Mean 27.15 27.40 27.26 SD 3.90 4.23 4.02 89 Main Analyses In this section, each hypothesis is stated, followed by a description of the statistical analysis performed and the results obtained. Hypothesis 1 There will be no difference between the group that drops out before the initial intake interview with a treatment provider and the group that attends the interview with regard to the following demographic variables: (a) age, (b) sex, (c) race, (d) educational level, (e) marital status, and (f) income level. Hypothesis 1 was analyzed to determine whether demographic variables were related to the completion and noncompletion of EAP- recommended mental health referrals. As indicated in Table 3, one- way ANOVA or chi-square tests were used to test the data associated with each demographic variable. Hypothesis 1 was rejected at the .05 level of significance for subject age and marital status. As shown in Table 3, subjects who completed the referral were younger (N = 37.08, E = 6.86) than those who did not (N = 41.54, SD = 8.03). Table 3 also reveals that the referral-completion group had a higher percentage of subjects (N = 17) who never married--21.8% as opposed to 3.9% of the referral- noncompletion group (N = 1). The referral-noncompletion group, however, had a higher percentage of subjects (N 4) who were separated--15.4% as opposed to 3.9% of the referral-completion group (N. = 3)- Hypothesis 1 was retained relative to respondent sex, race, educational level, and income level. None of the observed difI SigI dif I'ES Tab differences in these four demographic variables were statistically significant at the .05 level of significance. These observed differences are summarized in Table 3 and in the description of the research sample in the preceding section. Table 3.--Demographic summary of the research sample. Completion Noncompletion Test Group Group Total F Sig. of F Age Mean 37.08 41.54 38.19 £(1,103)=7.564 SQ 6.858 8.031 7.388 p = .007 Range 19-53 26-59 19-59 N 78 26 103 §e_. Male N = 31 N = 12 N = 43 N2 1)=.2723 Female N = 49 N = 15 N = 64 p = .60178 Race White N = 68 N = 20 N = 88 N?- 1)=1.65013 Black N = 12 N = 7 N = 19 p = .19894 Other N = 0 N = 0 N = 0 N_rital Status Never married N = 17 N = 1 N = 18 N;(4)=10.8020 Married N = 44 N = 14 N = 58 Separated N = 3 N = 4 N = 7 Divorced N = 16 N = 7 N = 23 Education Some high school _ = 1 N = o N = 1 N§(7)=6.84082 High school diploma N = 9 N = 4 N = 13 Some college N = 35 N = 10 N = 45 Z-year degree N = 13 N = 4 N = 17 4-year degree N = 8 N = 2 N = 10 Some grad. school N = 2 N = 4 N = 6 Master’s N = 6 N = 2 N = 8 Doctorate N = 6 N = 1 N = 7 Tabl Q‘VNil 91 Table 3.--Continued. Completion Noncompletion TESt Group Group Total E Sig. of E Income Below $15,000 N = 5 N = 1 N = 6 N;(8)=7.02344 $15,000-24,999 N = 13 N = 4 N = 17 p = .53411 $25,000-34,999 N = 23 N = 10 N = 33 $35,000-44,999 N = 14 N = 1 N = 15 $45,000-54,999 N = 8 N = 4 N = 12 $55,000-64,999 N = 4 N = 1 N = 5 $65,000-74,999 N = 5 N = 1 N = 6 Above $75,000 N = 7 N = 3 N = 10 Hypothesis 2 As measured by the CSQ-8 scale, the group that drops out before the initial intake interview with a treatment provider will report less satisfaction with their EAP assessment and referral experience than the group that attends the interview. Hypothesis 2 was tested to examine the data in light of findings from related research that suggested compliance with a referral recommendation for mental health treatment may be influenced by the helpseeker’s feelings about the source of the referral recommendation. A one-way ANOVA test was performed on the data to test the difference in satisfaction between subjects who completed the referral and those who did not. The results showed that there was a statistically significant difference between the two groups: E(1,104) = 7.410, p = .008. The results of this test are presented in Table 4. 92 Table 4.--Satisfaction score (CSQ-8 scale) by subject group. 3 Mean SQ E Sig. of E Noncompletion group 27 24.37 5.52 Completion group 80 27.26 4 02 7°4l° °°°8 Total 157 26.53 4.91 For the entire research sample (N = 107), the mean satisfaction score was 26.53 (SE = 4.91). Broken down by group, the mean satis- faction scores were 27.26 (SD = 4.02) for subjects who completed a treatment referral (N = 80) and 24.37 (g0 = 6.52) for subjects who did not (N = 27). The higher mean score of the referral-completion group represents greater reported satisfaction with the EAP assess- ment and referral service. Hypothesis 3 There will be no difference with respect to the previous use of an EAP between the group that drops out before the initial intake interview with a treatment provider and the group that attends the interview. Hypothesis 3 was tested to explore the possibility that previous use of an EAP, particularly among persons whose reaction to the assistance was comparatively less favorable, might be related to the noncompletion of a current referral recommendation for mental health treatment. A chi-square test of association was performed on the data to test the difference in the previous use of an EAP between subjects who completed the mental health referral and those who did not. The 93 results showed that there was no statistically significant difference between the two groups: £2“) = .31754, p = .57309. Hypothesis 3 was retained at the .05 level of significance. The results of this test are presented in Table 5. Table 5.--Previous EAP use by subject group. Noncompletion Completion Row Group Group Total Prior N 5 19 24 experience Cell % 4.7% 17.8% Row % 20.8% 79.2% 22.4% Col. % 18.5% 23.8% No N 22 61 83 experience Cell % 20.6% 57.0% Row % 26.5% 73.5% 77.6% Col. % 81.5% 76.3% Column total N 27 80 107 Row % 25.2% 74.8% 100.0% For previous users of an EAP, a second analysis was performed to test the group difference in subject opinion about the helpfulness of prior EAP assistance. A one-way ANOVA performed on the data showed that there was no statistically significant differ- ence between the two groups: E(1,23) = 2.55, p = .125. The helpfulness of prior EAP assistance was rated by subjects on a 4-point scale with a range from Not Helpful At All (1) to Extremely Helpful (4). Subjects who completed a mental health referral (N = 19) had a mean score of 3.11 (E = .809), which 94 slightly surpassed a 'rating of’ Helpful. Subjects who did not complete a treatment referral (N = 5) had a mean score of 2.40 (SD = 1.140)--about midway between scale criteria Somewhafit Helpful and Helpful. Hypothesis 4 There will be no difference in previous counseling experience between the group that drops out before the initial intake interview with a treatment provider and the group that attends the interview. Hypothesis 4 was tested to examine the relationship between prior and current use of psychological counseling services. As with prior EAP experience, there was the possibility that a portion of the variance in this investigation could be explained in terms of subject feelings about previous counseling experiences. A chi-square test of association was performed on the data to test the difference in previous counseling experience between subjects who completed the referral and those who did not. The results showed that there was no statistically significant difference between the two groups: NgU) = 1.50072, p = .6821. Hypothesis 4 was retained at the .05 level of significance. The results of this test are presented in Table 6. For subjects with counseling experience, a second analysis was performed to test the group difference in subject opinion about the helpfulness of previous counseling. A one-way ANOVA test performed on the data showed that there was no statistically significant difference between the two groups: E(1,58) = .933, p = .338. 95 Table 6.--Previous counseling experience by subject group. Noncompletion Completion Row Group Group Total Prior N 13 46 59 experience Cell % 12.4% 43.8% Row % 22.0% 78.0% 55.1% Col. % 48.2% 57.5% No N 14 34 48 experience Cell % 13.3% 32.4% Row % 29.2% 70.8% 44.9% Col. % 51.9% 42.5% Column total N 27 80 107 Row % 25.2% 74.8% 100.0% The helpfulness of previous counseling was rated by subjects on a 4-point scale with a range from Not Helpfulayat All (1) to Extremely Helpful (4). Subjects who completed a mental health referral (N = 46) had a mean score of 2.52 (§Q = 1.05). Subjects who did not complete a mental health referral (N = 13) had a mean score of 2.85 (E = 1.14). In relation to scale criteria, these mean scores fell between Somewhat Helpful and Helpful. Hypothesis 5a As represented by the ATSPPH scale composite score, the group that drops out before the initial intake interview with a treatment provider will have a less positive global attitude toward seeking professional psychological help than the group that attends the interview. Hypothesis 5a is one of two hypotheses that were tested to examine the relationship between attitudes toward seeking 96 professional psychological help and the completion or noncompletion of the EAP mental health referral. On the basis of prior research, it was anticipated that more positive attitudes would be related to referral completion, more negative attitudes to referral noncompletion. A one-way ANOVA test was performed on the data to test the difference in ATSPPH scale composite scores between subjects who completed the referral and those who did not. The results showed that there was no statistically significant difference between the two groups: E(1,105) = .6758, p = .4129. The ATSPPH scale has a range of O to 87. High scores are indicative of favorable attitudes toward seeking and using professional mental health services; low scores are indicative of unfavorable attitudes. For the research sample as a whole (N 107), the mean composite score on the ATSPPH scale was 63.90 (E = 11.80). For the subsample that did not complete a referral (N = 27), the mean composite score was 65.52 (E = 11.09); for the sub- sample that completed a treatment referral (N 80), the mean composite score was 63.36 (SQ = 12.05). These results are presented in Table 7. 97 Table 7.--ATSPPH scale composite and factor scores by subject group. Factor Composite 1 2 3 4 Mean Noncompletion group 18.37 10.69 16.02 10.89 65.52 Completion group 18.51 10.06 15.01 10.44 63.36 Total 18.47 10.22 15.27 10.56 63.90 SD Noncompletion group 4.15 2.74 3.83 2.21 11.09 Completion group 3 93 3.22 3.73 2.35 12.05 Total 3.97 3.10 3.77 2.31 11.80 Range Noncompletion group 12-24 5-15 7-21 6-14 38-84 Completion group 9-24 1-15 6-21 3-15 34-82 Total 9-24 1-15 6-21 3-15 34-84 E-Value .0235 .8275 1.4453 .7471 .6758 Sig. of E .8785 .3652 .2320 .3892 .4129 Key: Factor 1 Factor 3 Need for Help, Factor 2 = Stigma Tolerance, Interpersonal Openness, Factor 4 = Confidence Referral-completion group N = 80, referral-noncompletion group N = 27, Total N = 107 Hypothesiaf5b The group that drops out before the initial intake interview with a treatment provider will have a lower mean score than the group ‘that attends the interview on each of 'the factorial subscales of the ATSPPH inventory: (a) recognition of need for psychotherapeutic help, (b) stigma tolerance, (c) interpersonal openness, and (d) confidence in mental health professionals. 98 Hypothesis 5b was tested to examine the relationship between the completion or noncompletion of the EAP mental health referral and each of four factors of the helpseeking attitude domain measured by the ATSPPH scale. Separate one-way ANOVA tests were performed on the data to test the differences in the four factors between subjects who completed the referral and those who did not. As presented in Table 7, the results showed that there was no statistically significant differ- ence between the two groups for any of the factors: Factor 1-- E(1,105) = .024, p = .879; Factor 2--E(l,105) = .828, p = .365; Factor 3--E(1,105) = 1.445, p = .232; Factor 4--E(1,105) = .747, p = .389. Hypothesis 6a Compared with the group that attends the initial intake inter- view with a treatment provider, the group that drops out before the interview will have a lower Ete of aqreement with all three of Krause’s conditions for seeking mental health treatment. The three conditions are: a. They consider their present situation to be intolerable without professional psychotherapeutic assistance. b. They expect to be able to resolve their intolerable situa- tion with effective psychotherapeutic assistance. c. They expect to find the costs of obtaining or using profes- sional psychotherapeutic assistance to be tolerable. Hypothesis 6a is the first of several hypotheses designed to examine Krause’s (1966) view of psychiatric helpseeking motivation. Hypothesis 6a tested the theoretical proposition that persons who complete an EAP-recommended mental health referral will endorse all three prerequisite helpseeking conditions at a rate higher than that 99 of persons who do not complete such a referral. For this first analysis, the data were coded into agree/disagree categories to reflect Krause’s conception of the prerequisite conditions as dichotomous variables. A chi-square test of association was performed on the data to test the group difference in agreement with all three prerequisite conditions. The results showed that there was a statistically significant difference between the two groups: N;(l) = 11.03, p < .001. The results of this test are presented in Table 8. Table 8.--Krause’s conditions of psychiatric helpseeking motivation: Dichotomous scale scores. Noncompletion Completion Row Group Group Total Agree N 7 49 56 Cell % 6.5% 45.8% Row % 12.5% 87.5% 53.3% Col. % 25.9% 62.8% Disagree N 20 29 49 Cell % 18.7% 27.1% Row % 40.8% 59.2% 46.7% Col. % 74.1% 37.2% Column total N 27 78 105 Row % 25.7% 74.3% 100.0% 213(1) =11.03, p < .001. As shown in Table 8, the rate of agreement with all three conditions was substantially lower for the referral-noncompletion 100 group. Approximately 26% of that group agreed with all three conditions, whereas nearly' 63%. of ‘the referral-completion group agreed with them. Conversely, 74% of the referral-noncompletion group disagreed with one or more conditions, as predicted by the Krause theory. Hypothesis 6b The group that drops out before the initial intake interview with a treatment provider will have lower mean scores than the group that attends the interview for each of Krause’s three conditions for seeking mental health treatment. a. They consider their present situation to be intolerable without professional psychotherapeutic assistance. b. They expect to be able to resolve their intolerable situa- tion with effective psychotherapeutic assistance. c. They expect to find the costs of obtaining or using profes- sional psychotherapeutic assistance to be tolerable. Hypothesis 6b was analyzed to examine the degree of subject agreement or disagreement with each of Krause’s theorized conditions of psychiatric helpseeking motivation. For this analysis, the data were coded to allow the calculation and comparison of mean score differences between the referral-completion and referral- noncompletion groups. Separate one-way ANOVA tests were performed on the data to test the group difference associated with each of the three conditions. The results showed that the difference between the groups for each condition was statistically significant: Need for Treatment-- E(1,105) = 13.18, p = .0004; Confidence in Treatment--E(l,105) = 101 22.06, p = .0000; Tolerance for Treatment Demands--E(1,105) = 5.14, p = .0255. The results of these analyses are reported in Table 9. Table 9.--Krause’s conditions of psychiatric helpseeking motivation: Continuous scores. Condition Composite 1 2 3 Mean Noncompletion group 3.37 3.74 3.48 10.59 Completion group 4 58 4 93 4.25 13.75 Total 4.27 4.63 4.06 12.95 SD Noncompletion group 1.73 1.77 1.63 3.84 Completion group 1 40 .82 1 49 2.55 Total 1.57 1.24 1.55 3.22 Range Noncompletion group 1-6 1-6 1-6 2—16 Completion group 1-6 1-6 1-6 5-18 Total 1-6 1-6 1-6 2-18 E-Value 13.18 22.06 5.14 23.61 Sig. of E .0004 .0000 .0255 .0000 Key: Condition 1 = Need for Treatment, Condition 2 = Confidence in Treatment, Condition 3 = Tolerance for Treatment Demands Referral-completion group N = 80, referral-noncompletion group N = 27, total N = 107 102 The descriptive statistics presented in Table 9 are based on the following scale points: Very Strongly Disagree (1), Disagree (2), Slightly' Disagree (3), Slightly Agree (4), Agmaa, (5), and Strongly Agree (6). The condition with which there was the highest degree of agreement from both the referral-completion and referral- noncompletion groups was Confidence in Treatment (N = 4.63). This condition also evoked the least varied reaction from both groups (SQ = 1.24), particularly among subjects who completed the mental health referral (SE = .82). For the referral-noncompletion group, the condition with which there was the lowest degree of agreement was Need fOr Treatment (N = 3.37); for the referral-completion group, that condition was Tolerance for Treatment Demands (N = 4.25). The degree of agreement with all three conditions was uniformly lower for the referral-noncompletion group. Mean scores for this group did not reach the level of Slightly Agree for any of the three conditions. Hypothesis 7a There will be no difference in the level of perceived social pressure to seek professional psychological help between the group that drops out before the initial intake interview with a treatment provider and the group that attends the interview. Hypothesis 7a is one of three hypotheses that were tested to examine several aspects of the relationship between subject- perceived external pressure to seek professional mental health care and completion or noncompletion of the EAP referral to seek such assistance. 103 A one-way ANOVA was performed on the data to test the difference in perceived social pressure between subjects who completed the referral and those who did not. The results showed that there was no statistically significant difference between the two groups: E(1,105) = .0396, p = .8426. Hypothesis 7a was retained at the .05 level of significance. The results of this test are summarized in Table 10. Table 10.--Perceived social pressure level by subject group. N Mean §Q E Sig. of E Completion 9rOUp 80 1.86 1.08 Noncompletion group 27 1.81 1.08 “0396 ‘8426 Total 157 1.85 1.07 Subject-perceived social pressure was measured by a one-item, 4-point scale on which 1 = No Pressure, 2 = A Little Pressure, 3 = More Than a Little Pressure, but Less Than a Lot, and 4 = A Lot of Pressure. For the entire research sample (N = 107), the mean subject-perceived social pressure score was 1.85 (SD = 1.07). Broken down by group, the mean score was 1.86 (SE = 1.08) for subjects who completed a treatment referral (N = 80) and 1.81 (§Q = 1.08) for subjects who did not (N = 27). In relation to scale criteria, these scores indicate that most subjects in both groups felt only slight pressure from others to seek professional mental health care. CC SC 104 The sources of social pressure reported by the referral- completion group (and the percentage of subjects reporting each source) were: spouse (21.3%), EAP interviewer (16.4%), parent(s) (13.1%), supervisor* at work (9.8%), friend(s) outside of work (9.8%), co-worker(s) (8.2%), son(s) or daughter(s) (6.6%), medical doctor (6.6%), other* person(s) (6.6%), and union representative (1.6%). The sources of social pressure reported by the referral- noncompletion group were: spouse (17.3%), supervisor at work (22.7%), EAP interviewer (13.6%), other person(s) (9.1%), parent(s) (4.5%), co-worker(s) (4.5%), brother(s) or sister(s) (4.5%), son(s) or daughter(s) (4.5), medical doctor (4.5%), and union representa- tive (4.5%). Hypothesis 7b Among subjects whose ATSPPH scale composite score is less than the mean of the research sample, the group that drops out before the initial intake interview with a treatment provider will have a lower perceived social pressure mean score than the group that attends the interview. Hypothesis 7b was tested to determine whether greater social pressure to seek professional mental health care was associated with referral completion among subjects who had comparatively less favorable attitudes toward seeking such care. A one-way ANOVA test was performed on the data to test the difference in subject-perceived social pressure between these two referral-completion and referral-noncompletion subgroups. The results showed no statistically significant difference between these 105 two subgroups: E(1,44) = .0825, p = .7753. These results are summarized in Table 11. Table 11.--Subject-perceived social pressure levels among subjects with less favorable psychiatric helpseeking attitudes. fl Mean SQ E Sig. of E Completion subgroup 36 1,39 1.12 Noncompletion subgroup 10 2.00 .94 '0825 '7753 Total 40 1.91 1.07 As shown in Table 11, the mean perceived social pressure score of subjects (N = 10) who did not complete the referral was slightly higher (N = 2.00) than that of subjects (N = 36) who completed the referral (N = 1.89). These mean scores indicate that most subjects in both subgroups felt only slight social pressure to seek professional mental health care. Hypotheaia 7c Among subjects who do not satisfy all three of Krause’s conditions for seeking mental health treatment, the group that drops out before the initial intake interview with a treatment provider will have a lower perceived social pressure mean score than the group that attends the interview. Hypothesis 7c was tested to determine whether there was a subject-perceived social pressure effect on the completion of a mental health referral for the subjects who did not satisfy all thI noI di su cc sI 106 three of Krause’s theorized conditions of psychiatric helpseeking motivation. A one-way ANOVA test was performed on the data to test the difference in subject-perceived social pressure between the two subgroups whose subjects did not satisfy all three theorized conditions of psychiatric helpseeking motivation. The results showed no statistically significant difference between these two subgroups: E(1,48) = .4056, [a = .5272. These results are summarized in Table 12. Table 12.--Subject-perceived social pressure levels among subjects not satisfying Krause’s conditions of psychiatric help- seeking motivation. l2 Mean SD F Sig. of E Completion subgroup 30 1,30 1.03 Noncompletion subgroup 20 2.00 1.17 '4056 '5272 Total 50 1.88 1.08 As shown in Table 12, the mean perceived social pressure score of subjects (N = 20) who did not complete the referral (N = 2.00) was slightly higher than that of subjects (N = 30) who completed the referral (N = 1.80). These mean scores indicate that most subjects in both subgroups felt only slight social pressure to seek professional mental health care. res dat ev; st; in to Fi e)- he 107 Discriminant Function Analysis Discriminant function analyses were performed to answer several research questions not addressed in the preceding analyses of the data. First of all, discriminant function analysis was used to evaluate the relative importance of the five variables that statistically differentiated the referral-noncompletion group from the referral-completion group. This analysis provided information to evaluate the decision-making stage of the helpseeking model by Fischer et al. (1983). A second, similar analysis focused exclusively on the three Krause conditions of psychiatric helpseeking motivation to determine the relative importance of each condition in differentiating the groups. As part of each analysis, the combination of variables was also put through a classification procedure to evaluate how' well the combination predicted group membership. In both of the analyses, a stepwise selection method, MAXMINF, was used to enter the variables into the discriminant function. MAXMINF (SPSS, Inc., 1986) was chosen because the researcher wanted to determine the relative importance of each variable. The MAXMINF method evaluated the unique contribution each variable made to the discriminant power of the fUnction and tested the significance of that contribution by performing an F test. By focusing on the largest E value, MAXMINF sequentially selected the predictor variable that, when combined with previously selected predictor variables, produced the largest additional increment in discrimi- nation. ii di CC 108 In the first analysis, four variables were selected from the five variables for which there were statistically significant differences between the referral-noncompletion and referral- completion groups. The selection of variables stopped with the fourth variable because the remaining variable did not possess the minimum E needed to enter the discriminant function (i.e., the remaining variable accounted for a negligible, additional amount of explained variance). The first variable entered was the Krause Confidence in Treatment subscale score, followed by the CSQ-8 scale score, age of subject, and the Krause Need fer Treatment subscale score. The results of this selection process are summarized in Table 13. Table 13.--Discriminant analysis summary: All significant vari- ables. Standardized Step Discriminant Wilks’s Sig. No. Function Variable Lambda E of E Coefficient 1 .608 Confidence .826 22.063 .0000 in Treatment (Krause) 2 .493 Client satis- .785 14.279 .0000 faction (CSQ-8 scale) 3 -.342 Age of subject .761 10.759 .0000 4 .326 Need for Treat- .744 8.755 .0000 ment (Krause) 109 To help in interpreting Table 13, the standardized discriminant function coefficient indicates the relative importance of each variable in contributing to group differences. More specifically, each coefficient indicates the relative importance of the associated variable in predicting the criterion, in this case referral completion or noncompletion. The coefficient is functionally equivalent to a beta weight computed in regression analysis. For the four predictor variables in this first analysis, a subject’s score on the Krause Confidence in Treatment subscale had the largest effect on predicting that person’s referral behavior. The Krause Need for Treatment subscale score had the smallest effect. Finally, Wilks’s lambda is the proportion of unexplained variance remaining after the inclusion of each successive predictor variable. Each successive decrease of the Wilks’s lambda was tested for statistical significance by the E test. The canonical correlation between group membership (referral completion or noncompletion) and the combined four predictor variables was 0.51. Canonical correlation is a measure of the relationship between a criterion variable and a combination of predictor variables. The square of the canonical correlation is the proportion of explained variance in the total variance of the criterion variable. In this analysis, the four predictor variables explained 26% of the total variance of group membership. A classification procedure was also performed to predict the group membership of each subject. The results of this procedure 110 were used to gauge the potential of the discriminant function composed of the four predictor variables to predict group membership. The procedure first computed group centroids for the referral-completion and referral-noncompletion groups. A group centroid is the mean of the discriminant scores within a group. For the referral-completion group, the group centroid was 0.33721. For the referral-noncompletion group, the group centroid was -0.99913. These values were then compared with the discriminant score calculated for each subject. Subjects were predicted to be members of the group whose centroid was closer to their discriminant score. The procedure correctly classified 70.1% of the subjects (75 of the 107 subjects). As shown in Table 14, 13 of the 27 subjects who did not complete the referral (48.1%) were correctly classified; 62 of the 80 subjects who completed the referral (77.5%) were correctly classified. Table 14.--Classification results: All significant variables. Predicted Group Actual Group N Noncompletion Completion No. % No. % Noncompletion 27 13 48.1 14 51.9 Completion 80 18 22.5 62 77.5 To put these prediction (classification) rates into a meaningful perspective, they were compared with the percentage of 111 correct classifications expected on ‘the basis of chance alone. Based on chance, the prediction rate for the referral-completion group was 74.8%. Discriminant analysis increased the prediction rate for this group to 77.5%, thereby correctly classifying an additional 2.7% of the group. Similarly, the chance prediction rate for the referral-noncompletion group was 25.2%. Discriminant analysis correctly classified 48.1% of this group, or an additional 22.9% of the group. The second discriminant function analysis was performed using only scores from the Krause psychiatric helpseeking motivation subscales: Need for Treatment, Confidence in Treatment, and Tolerance for Treatment Demands. This second analysis was done to determine the percentage of total variance of group membership attributable to these three motivational variables. In addition, it provided the means to examine the structure of a discriminant function based on Krause’s theory. It also permitted the prediction rate based on these three variables to be determined and compared with the rate based on the four variables in the first discriminant function. Results of the analysis are presented in Table 15. 112 Table 15.--Discriminant analysis summary: Krause variables. Standardized Step Discriminant Wilks’s Sig. No. Function Variable Lambda E of E Coefficient l .701 Confidence .826 22.063 .0000 in Treatment (Krause) 2 .367 Need for Treat- .806 12.555 .0000 ment (Krause) 3 .255 Tolerance .795 8.840 .0000 for Treatment Demands (Krause) The canonical correlation for the combination of these three predictor variables was 0.453. The square of this correlation, or 20.48, represents the proportion of explained variance in the total variance accounted for by the three Krause variables. By compari- son, the previous four predictor variables accounted for 26.01% of the total variance of group membership. Thus, the three Krause variables together accounted for 78.74% of the total explained variance of group membership (20.48/26.01 = .7874). Breaking down the total known variance explained by the Krause variables, 84.78% was accounted for by the condition Confidence in Treatment, 10.18% by the condition Need for Treatment, and 5.04% by the condition Tolerance for Treatment Demands. (The Wilks’s lambda associated with each condition provided the information for computing these percentages.) The dramatic decrease in the variance 113 accounted for by the second and third conditions shows that the three variables shared much of the discriminating information contained in the discriminant function. This shared information was accounted for when the first variable was entered into the function. The same ranking of the Krause variables is evident in the size of the standardized discriminant function coefficients: Confidence in Treatment, (.701), Need for Treatment (.367), and Tolerance for Treatment Demands (.255). It has been pointed out, however, that the standardized discriminant function coefficient does not always provide the clearest indication of the importance of a predictor variable (Betz, 1987; Klecka, 1984). Structure coefficients, or discriminant loadings, may be more useful values for interpreting the meaning of a canonical discriminant function, specifically the dimensionality of group differences. The structure coefficient is simply the product-moment correlation between a predictor variable and a discriminant function. The structure coefficient reports how closely the variable and function are related. In relation to Krause’s conditions, the structure coefficients were: Confidence in Treatment (.903), Need for Treatment (.698), and Tolerance for Treatment Demands (.255). These figures revealed that confidence in treatment was still the most important difference in motivation between the referral-noncompletion and referral- completion groups; however, they also show that need for treatment was a more important difference than previously suggested by the standardized discriminant function coefficient. Tolerance for tre con re' th 114 treatment demands also increased in importance as the third component of the Krause composite variable, but it clearly played a relatively minor role in the motivational differences reported by the two groups. The classification procedure 'was also performed to predict group membership for each subject using the combination of the three Krause predictor variables. The group centroid for the referral- completion group was 0.29202. For the referral-noncompletion group, the group centroid was -O.86524. Comparison of each subject’s discriminant score with these centroids produced the predictions shown in Table 16. Table 16.--Classification results: Krause variables. Predicted Group Actual Group N Noncompletion Completion No. % No. % Noncompletion 27 15 55.6 12 44.4 Completion 80 15 18.8 65 81.3 Table 16 shows that the classification procedure performed with the three Krause subscale scores correctly classified 74.8% of the sample (80 of 107 subjects). By comparison, 70.1% of the sample was correctly classified on the basis of the four predictor variables. For the referral-completion group, the Krause composite-score prediction rate was 81.3%, versus 77.5% for the four predictor vari nonc vers of c EAP que Mos sut 00 di ar 115 variables and 74.8% on the basis of chance. For the referral- noncompletion group, the Krause score prediction rate was 55.6%, versus 48.1% for the four predictor variables and 25.2% on the basis of chance. Summary The research sample of this investigation was composed of 107 EAP clients who returned usable copies of the researcher’s questionnaire. Women outnumbered men in the sample three to two. Most subjects were married and Caucasian. The mean age of the subjects was 38. Slightly more than half of the subjects had been in counseling previously. Hypothesis testing revealed that, as a group, subjects who did not complete an EAP-recommended mental health referral (N = 27) differed from subjects who did (N = 80) in the following respects: 1. They were several years older, more likely to be separated, and less likely to have remained single. 2. They were somewhat less satisfied with their EAP assessment and referral service. 3. They were less likely to be in agreement with each of Krause’s prerequisite conditions for seeking mental health treat- ment. 4. They expressed a lower degree of agreement with each of Krause’s conditions. Nearly 75% of the group that did not complete a referral were identified correctly when Krause motivational data were analyzed 116 with a 2 x 2 contingency table. Slightly more than 50% of the group were correctly identified by a discriminant function analysis of Krause motivational data. Twenty-five percent of the group could be classified correctly in regard to their referral behavior on the basis of chance alone. Discriminant function analysis also determined that confidence in 'treatment was the most important measured difference between this group and the group that completed a referral. CHAPTER V SUMMARY AND CONCLUSIONS EAPs have become an important bridge to the organized mental health care system. Many troubled individuals now consult an EAP interviewer to decide whether to seek professional mental health treatment. Findings from past EAP and EAP-related research suggest that the majority of these individuals will follow the recommendation of their interviewer to seek professional help. Others--expressed intentions notwithstanding--will not. The present study showed that this outcome can occur 25% of the time. Unless their need for treatment subsides, persons who do not complete these referrals; miss out on the potential benefits of professional help. A primary objective of this study, therefore, was to identify factors that distinguish this group from the group that completes a referral. The identification of these factors would help EAP interviewers anticipate and respond to client concerns known to undermine follow-through. In the present investigation, as many as three of every four subjects who did not complete their mental health referral could have been identified during their EAP interview if certain motivational factors had been assessed. 117 118 Several matters of theoretical interest were also examined as part of this research investigation. First, evidence was examined pertaining to the soundness of the Krause theory of treatment motivation as applied to initial helpseeking efforts (Krause, 1966). Second, the data provided the means for a similar analysis of the decision-making stage of the Fischer et al. (1983) model of psychiatric helpseeking. Third, it was possible to study the relationship between measured attitudes and both past and present behavior associated with seeking professional mental health care. Fourth, it was possible to analyze EAP referral outcomes in relation to an important assumption of the EAP assessment and referral model --the assumption that social pressure helps to motivate efforts to seek mental health care among persons who would not otherwise be so inclined. . The setting for this investigation involved the programs and facilities of four EAPs in Lansing and Grand Rapids, Michigan. All four EAPs had been in operation for more than 5 years and were well- established programs. Staff interviewers who volunteered to take part in the study had considerable experience conducting assessment and referral interviews. They held or were completing graduate degrees in psychology, counseling, or social work. At each of the EAPs, the assessment and referral of individuals with mental health concerns occupied a major portion of staff time. The research sample was recruited from 256 clients from these four programs. Clients were included in the volunteer sample if they' expressed the intention to seek. professional mental health 119 care, agreed to let their EAP interviewer confirm their completion of an intake interview with a treatment provider, and completed a copy of the researcher’s questionnaire. The final sample was composed of 107 clients who fulfilled these requirements. Women made up 59.8% of the sample, Caucasians 82.2%, and persons with more than a high school education 86.9%. In addition, 54.7% of the sample members were married. The mean age of sample members was 38.19 years. The criterion variable in this study was the completion of an initial intake interview with a mental health treatment provider or program. The main independent variables included client satisfaction with the EAP assessment and referral process; attitudes toward seeking professional mental health care; intrinsic and extrinsic motivation to seek professional mental health care; previous use of an EAP; previous counseling experience; and demographic characteristics, including sex, age, race, marital status, income, and education. The completion (or noncompletion) of an initial intake interview with a treatment source was verified by a telephone call from the EAP interviewer to the treatment provider. Client satisfaction with the EAP assessment and referral process was measured with the Client Satisfaction Questionnaire (CSQ-8) (Larsen et al., 1979). Attitudes toward seeking professional mental health care were measured with the Attitudes Toward Seeking Professional Psychological Help (ATSPPH) scale (Fischer B Turner, 1970). 120 Intrinsic motivation to seek professional mental health care was measured by questionnaire items devised to assess motivational factors conceptualized by Krause (1966). Extrinsic motivation, or client-perceived social pressure to seek professional mental health care, was measured by questionnaire items devised for this study. All other variables were measured by items created expressly for this study. All scales and personal information questions were presented to the volunteer subjects in a confidential questionnaire. The basic design of the investigation was correlational. This design allowed relationships between all variables of interest and the outcome of the EAP referral to be examined to determine the statistical significance and the strength of these relationships. It did not permit conclusions to be drawn about cause-and-effect relationships. Chi-square tests, one-way ANOVA tests, and discrimi- nant function analyses were used to analyze differences observed between subjects who did and did not complete an initial intake interview with a treatment provider. Summary of Findings Qamographic Variables Because findings from previous research were ambiguous, it was hypothesized that there would be no significant differences between the referral-noncompletion and referral-completion groups in relation to sex, age, race, marital status, income, or education. No significant differences were found between the groups with respect to sex, race, education, or income. Age and marital status 121 differences, however, were significant at the .05 level. The referral-noncompletion group was several years older than the referral-completion group. The referral-noncompletion group also contained a higher percentage of persons who were separated, but a smaller percentage of persons who had not married. Satiafaction With EAP Service On the basis of findings from related research, it was hypothesized that subjects in the referral-noncompletion group would express less satisfaction with their present EAP assessment and referral service than subjects in the referral-completion group. As predicted, the mean score of the referral-noncompletion group was lower on the CSQ-8 scale, and this difference was significant at the .05 level. There was no significant difference between the groups in the proportion of group members who had previously received help from an EAP. There also was no significant difference between the groups in the rated helpfulness of prior EAP assistance. Helpfulness of Previous Counseling Experience It was suspected that considering a previous counseling experience to be unhelpful or marginally helpful would discriminate the referral-noncompletion group from the referral-completion group. However, there was no significant difference between the groups in the rated helpfulness of previous counseling. There was also no 122 significant difference between the groups in the proportion of group members who had previously received counseling. Attitudes Toward Seeking Mental Health Treatment Based on earlier research findings, it was hypothesized that the attitudes toward seeking professional mental health care of the referral-noncompletion group would be less favorable than those of the referral-completion group. However, there were no significant differences in attitudes between the groups, as measured by the ATSPPH scale. Psychiatric Helpseeking Motivation It was predicted that the referral-noncompletion and referral- completion groups would show significant differences in relation to the prerequisite conditions of Krause’s theory of treatment motivation. In particular; it was hypothesized that. a smaller proportion of the referral-noncompletion group would endorse all three of Krause’s theorized conditions. This hypothesis was retained at the .05 level. It also was hypothesized that the degree of agreement with each of the three conditions would be lower for the referral-noncompletion group. This hypothesis, too, was retained at the .05 level. Perceived Social Pressure to Seek Professional Mental Health Cage The researcher speculated that the referral-noncompletion group might differ from the referral—completion group with respect to 123 perceived social pressure to seek professional mental health care. However, the difference in social pressure level between the groups was not significant. It also was predicted that comparatively higher levels of social pressure would be associated with subjects who completed a referral, but had comparatively less favorable attitudes toward seeking professional assistance. However, there was no significant difference in social pressure level when subjects in the referral- completion group were compared with subjects in the referral- noncompletion group who scored below the ATSPPH scale sample mean. Similarly, it was predicted that comparatively higher levels of social pressure would be associated with subjects who completed a referral, but were comparatively less intrinsically motivated to seek professional assistance. However, there was no significant difference in social pressure level when subjects in the referral- completion group were compared with subjects in the referral- noncompletion group who did not fulfill all three of Krause’s prerequisite conditions of motivation. Prediction of Pretreatment Dropout in EAP Mental Health RefeNNals The final area of investigation addressed the question: Which combination of variables examined in the study most accurately classified the pretreatment dropouts in the research sample? Contingency table analysis of the three Krause motivational conditions showed that 74.1% of the subjects who did not complete 124 thEir mental health referral could be identified from their agreement or disagreement with these three variables. A discriminant function analysis showed that 48.1% of the referral-noncompletion group could be identified correctly using discriminant information pertaining to four variables. The four variables were: (1) confidence in the effectiveness of treatment (Krause scale), (2) satisfaction with current EAP assessment and referral service (CSQ-8 scale), (3) age of subject, and (4) need for treatment (Krause scale). The 48.1% prediction rate achieved with discriminant function analysis and these four variables was also a substantial improvement over the 25.2% prediction rate expected on the basis of chance alone. Discussion of Findings In this section, the results of the analyses of the data are discussed in relation to the research questions posed in Chapter 11. Special emphasis is given to understanding the subjects who did not complete mental health referrals. For continuity and coherence, the discussion is presented in the order in which results were presented in the preceding section. Demographic Variables Age. As a group, subjects who did not complete their mental health referral were 4.46 years older than those who did. This difference may be a reflection of the weak negative correlation (N = -.13, p > .05) between age and confidence in treatment, given that confidence in treatment (as measured by the Krause scale) was more 125 closely related to referral outcome (N = .42, p < .01) than any other variable examined. Similarly, the weak negative relationship (N = -.09, p > .05) between age and another key variable, felt need for treatment (as measured by the Krause scale), suggests that younger members of the research sample felt slightly less able than older members to cope with their problems without professional help. Until these results are replicated with an acceptable level of significance, however, no sound conclusions can be drawn from them. Narital status. The referral-noncompletion group and the referral-completion group contained virtually the same percentage of married subjects. Therefore, marriage per se did not differentiate subjects who completed a referral from those who did not. The referral-noncompletion group, however, contained a higher percentage of individuals who were separated. The referral-completion group had a higher percentage of individuals who had not married. How these differences may have been related to referral follow-through behavior, though, is not clear from the analysis of the data. Taken as a whole, the demographic differences that distinguish the referral-noncompletion group from the referral-completion group support the general conclusions made about past research findings in Chapter 11. Those conclusions were that connections between demographic variables and referral outcome were inconsistent across studies and were more descriptive than explanatory. In view of the unacceptable significance levels and the weak correlations associated with most of the demographic differences found in the 126 present study, replication with a new sample would probably reveal somewhat different demographic distinctions between the referral- noncompletion_and referral-completion groups. By themselves, demographic variables do not explain much about the behavior of the subjects who did not complete their referral. Satisfaction With EAP Service Pretreatment dropout was statistically related to the subjects’ level of satisfaction with their current EAP assessment and referral experience. The mean score of subjects in the referral- noncompletion group was approximately three points lower on the CSQ-8 scale than the mean score of the referral-completion group. Feelings about earlier experiences with an EAP followed a similar pattern: Subjects in the referral-noncompletion group rated the helpfulness of previous EAP assistance lower than subjects in the referral-completion group did. These reactions to past assistance, however, were provided by only five subjects and were not statisti- cally significant. With 32 points possible on the CSQ-8 scale, the comparatively high ratings given to EAP service by both the referral-completion group (N = 27.36) and the referral-noncompletion group (N = 24.37) indicate that most or all rated aspects of EAP service were pleasing, though a little less so in the estimation of the referral- noncompletion group. Most clients reported that they received the kind of service they wanted, the amount of help they wanted, service that helped them cope with their problems, and service that met 127 their needs. Unfortunately, these categories of service were stated in rather general terms in the scale (the preceding categories are the actual terminology) that did not permit these reactions to the EAPs to be analyzed in more detail. A separate questionnaire item, however, showed that a specific source of satisfaction with EAP assistance was the number of referral resources recommended (1; = .30, p < .01). It would be well to note that the mean satisfaction-score difference between the referral-noncompletion and referral- completion groups was rather small, viewed in terms of CSQ-8 scale units. The 2.89-point difference translates into an average difference of only 0.36 points (2.89 points divided by eight items) per four-point item. The predictive utility of these relatively minor differences in satisfaction is likely to be realized only when clients actually complete the CSQ-8 scale and statistical methods of prediction are applied to their responses, as was done in this study. Prediction of pretreatment attrition will be considered in more detail later. Previous Counseling Experience Previous counseling experience was not statistically related to referral outcome in the research sample. Compared with the referral-completion group, a somewhat smaller percentage of the referral-noncompletion group had been in treatment previously (48.25% versus 57.5%). Chi-square analysis (Table 6) determined that this difference was not significant. 128 Similar results were obtained when the ratings subjects gave to the helpfulness of their previous counseling experience were analyzed with one-way ANOVA. The mean helpfulness rating of the referral-noncompletion group (N = 2.85, fl = 1.14) was actually slightly higher than that of the referral-completion group (N = 2.52, SQ . 1.05). However, the mean score difference between the groups, 33/100ths of a point, was not statistically significant. Expressed in terms of scale values, both groups rated their previous counseling closer to Nelptpl than §pmewhat Helpful; in other words, they regarded their previous experience with counseling as having been more than somewhat helpful to them. The similarity between the groups with regard to these two variables suggests that self-selection affected the composition of the research sample. Persons who volunteered for the study were already members of a largely self-selected group--troubled individuals who had turned to one of the four EAPs for help. Presumably, they believed that the EAP, or mental health care recommended by the EAP, could help them resolve their concerns. It is understandable that most members of this group who had had previous counseling rated that experience favorably. Had they not felt this way, they would probably have been disinclined to seek further involvement with EAP and mental health care services. In short, self-selection apparently gave rise ix: a sample of EAP users who were rather homogeneous with respect to their feelings about seeking professional assistance. Furthermore, the narrow 129 range of their feelings about previous counseling made these feelings useless as a predictor of their referral behavior. Attitudes Toward Seeking Mental Health Treatment The results of one-way ANOVA tests showed that the referral- noncompletion group did not differ significantly from the referral- completion group in their attitudes toward seeking professional mental health care. The mean composite score on the ATSPPH scale was actually slightly higher (more pro-helpseeking) for the referral-noncompletion group. It had been hypothesized, based on the review of literature, that higher attitude scores would be associated with the referral-completion group. The absence of a significant difference between the groups suggests that the helpseeking attitudes measured by the ATSPPH scale may not be useful in discriminating EAP clients who will complete a mental health referral from those who will not. Table 7 shows how similar the referral-noncompletion group and the referral-completion group were in terms of the means, standard deviations, and ranges associated with the ATSPPH scale composite score and each of the factorial scores. This similarity restricted the range of responses to the ATSPPH scale to the point that the scale lost whatever effectiveness it might have as a predictive device with more attitudinally diverse groups. It is possible that significant. differences in helpseeking attitudes actually did exist between the referral-noncompletion group and the referral-completion group, but that the ATSPPH scale 130 failed to detect those differences. This seems an unlikely possibility, though. For one thing, as discussed in the preceding section, the sampling procedure used in the present study virtually guaranteed that there would be a high degree of similarity between the groups. All subjects were recruited only from individuals who had turned to an EAP for assistance, made a verbal commitment to seek treatment, and volunteered to participate in the study. For another thing, as reviewed in Chapter III, the ATSPPH scale was constructed with appropriate attention to accepted principles of test construction and standardization. The internal and test-retest reliability coefficients, as well as the concurrent and known-group validity coefficients, are evidence that the scale is a valid measure (If an individual’s attitudes toward seeking professional psychological assistance. The lack of differentiation between the groups on these attitudes could also mean that subjects who dropped out rapidly lost any improvement in attitude gained during their EAP assessment and referral interview. Such a decline could have caused their true attitudes at the time they decided not to complete their referral to be overstated by their earlier responses to the researcher’s questionnaire. In the present investigation, preinterview and serial postinterview administrations of the ATSPPH scale were not used, so it is not possible to analyze whether the attitudes of the research subjects actually did increase and then fall during this interval. Brennan (1982) reported that the ATSPPH scale scores of 131 both the referral-completion group and the referral-noncompletion group in his study did increase about five points from just before to just after the interview. Assuming the subjects in the present study who did not complete a referral did experience a short-lived improvement in their helpseeking attitudes, the loss of that improvement might have been associated with the halo effect (Mehrens, 1973). Because most subjects in the present study were pleased with their experience at the EAP (as indicated by CSQ-8 scale group means and standard deviations), their satisfaction might have colored their responses to the ATSPPH scale for a brief period after their EAP interview. The correlation between ATSPPH scale composite scores and CSQ-8 scale scores (E = .21, p < .05) suggests that client satisfaction was indeed related to attitude level. Closer to the time of their scheduled treatment intake, however, this sense of satisfaction could have waned, allowing attitudes to recede to preinterview levels. Had it been possible to measure the attitudes of subjects when they changed their minds about completing the referral, their potentially lower ATSPPH scale scores at that time might have discriminated them from the group that completed their referrals. For this explanation to be tenable, any such improvement in helpseeking attitudes would have had to be lost quickly and disproportionately' by ‘the subjects in the referral-noncompletion group before they changed their minds about seeking professional care. Recent research on the Elaboration Likelihood Model of attitude change has indicated that such a rapid reversion to a 132 baseline level of attitudes is possible (Heesacker, 1986; Petty & Cacioppo, 1984). All that can be concluded from the present study, however, is that further research using multiple observations is needed to determine whether and how helpseeking attitudes fluctuate during the assessment and referral process. Given the lack of predictive validity of the ATSPPH scale with the research sample, it is interesting to note that, consistent with other studies (e.g., Dadfar & Friedlander, 1982; Kligfeld & Hoffman, 1979; Surgenor, 1985), the data were supportive of the known-group validity of the scale. Previous counseling experience could be detected with the scale, even though future mental health care helpseeking behavior could not be predicted. The correlation between previous counseling experience and the ATSPPH scale composite score was .30 (p < .01). This correlation indicates that, as a group, subjects who had had previous counseling experience reported significantly higher pro-helpseeking attitudes than subjects without this experience. Even with their previous experience and favorable attitudes, however, they were no more likely to complete their EAP referral than the group that lacked experience and had somewhat less favorable attitudes. In summary, no relationship was evident between the psychiatric helpseeking attitudes of the subjects in this study and their decision to drop out or complete their mental health care referral. Once they completed their assessment and referral interview, other factors evidently were more important in maintaining or disrupting 133 their stated intention to seek professional mental health care. In the' next section, helpseeking attitudes are further examined in relation to the conceptually similar Krause motivational conditions to try to explain why the latter seem to be more central to the completion or noncompletion of referrals. Psychiatric HelpseekingMotivation Several researchers cited in the review of literature suggested that motivation level may be an important variable in the completion and noncompletion of nental health referrals (e.g., Gaines, 1978; Noonan, 1973; Pfouts et al., 1963). In the present study, motivational factors were found to be key predictors of referral behavior. In the discriminant function analysis of the data, more than half of the explained variance of the criterion variable was accounted for by Krause’s three prerequisite conditions of pretreat- ment motivation. Fischer and Turner (1970) speculated that future empirical research would reveal that an individual’s felt need for professional mental health care and confidence in the mental health profession are the key components of the decision to seek professional mental health care. The analysis of the data in this study supports their belief in the importance of these two factors. Of all the variables examined, the most important difference between the referral-noncompletion group and the referral-completion group, as determined by discriminant function analysis, was confidence in mental health care (Table 13). When the three Krause conditions 134 were analyzed separately from other variables, the ranking was Confidence in Mental Health Care, then Need for Professional Mental Health Care, then Tolerance for the Demands of Treatment. To be sure, mean-score differences between the groups for each of the Krause conditions (construed as continuous variables) were modest, viewed in relation to the six scale values (Table 9). For the referral-noncompletion group, mean scores for all three conditions were from the interval between scale points Slightly Disagree and Slightly Agree. For the referral-completion group, mean scores for all three conditions were from the next higher interval, between scale points Slightly Agree and m. These modest differences were sufficient, however, to differentiate members of the sample to the extent that 74.8% of them were classified correctly as to the outcome of their EAP referral. An unexpected finding of this study was the degree to which the researcher’s Krause scale surpassed the ATSPPH scale in predicting referral behavior. Considering that the Krause scale duplicates a number of the ATSPPH scale items in substance, strong correlations between the Krause scale scores and the ATSPPH scale scores were anticipated, as well as similar prediction rates. The weak correlations between the scales may have resulted from a difference in the orientation of the scales toward similar ideas about seeking professional psychological help. Many of the items included in the ATSPPH scale were written in the subjunctive mood; they called upon the research subjects to imagine how they might feel or act if ever the need arose to consider seeking professional 135 mental health care. These questions were worded: "I would willing confide. . . ," "If I believed I was having. . . ," "I would want to get. . . ," and ". . . I might want to have. . . ." and the like. In contrast, the researcher’s Krause scale items were written in the present tense and inquired about the subjects’ view of their current situation. The subjects were asked not what they would do regarding hypothetical mental health crises, but rather what they were going to do about their immediate and actual mental health concerns. The pro-helpseeking attitudes measured by the ATSPPH scale probably predispose troubled individuals ix: consider using professional mental health care services. Their actual decision to seek or not seek such help, however, probably depends primarily on their assessment of their immediate situation. It is reasonable to suppose that changes in their perceived need for treatment, confidence in treatment, or concern about the demands of engaging in treatment would affect their helpseeking intentions and actions. A rapid change in symptoms, for example, could either hasten or stop their efforts to get help. The changeableness of personal circumstances and the need to adjust to such changes may explain the behavior of the subjects in this study whose referral behavior was not predicted correctly by their responses to the Krause scale. Predictions of referral outcomes might have been more accurate had it been possible. to account for changes in subject outlook that occurred after the questionnaire was completed, but before the initial session with a 136 treatment provider took place. Even so, the ability of the Krause scale to classify referral behavior correctly in 7 out of 10 cases implies that the scale possesses both construct and predictive validity. Furthermore, this predictive accuracy is evidence that Krause identified important cognitive correlates of mental health care helpseeking behavior. The number of correct classifications also suggests that in the majority of cases in this study, the subjects’ assessment of their situation, and their resultant helpseeking motivation, remained stable while they waited to begin treatment services. It was anticipated that strong social pressure might help explain the behavior of some subjects who sought treatment despite their measured lack of intrinsic helpseeking motivation. .Analysis of the perceived social pressure levels of these persons, however, did not reveal a high level of social pressure in comparison with the remainder of the research sample. ‘Therefore, social pressure does not appear to explain the behavior of this group or the misclassification of these subjects based on their responses to the Krause scale. In retrospect, it would have been useful to collect data to analyze the reactions of subjects toward each of the costs or demands of treatment referred to in the measure of Krause’s Tolerance for Treatment Demands. It is possible, however, with the help of the correlations reported in Table 1, to examine the relationship between the financial cost of professional help and referral outcome. Table 1 shows that there was a significant 137 relationship between income level and scores on the Tolerance for Treatment Demands subscale (t = .35, p < .01). At first glance, then, it appears that subjects with higher income were more likely, possibly more able, than subjects of lesser means to seek treatment and manage the financial costs. The relationship between income and actual follow-through behavior, however, gives a different impression; the correlation between these two variables was only .04 and was not statistically significant. These findings suggest that even lower income subjects had the means to pay for treatment if they felt a strong need for it (as employees, their health insurance may have paid all or most of the cost). These findings imply that concern over the monetary cost of treatment may not be the chief concern of EAP clients contemplating treatment. Concern about what others may think, the inconvenience of attending weekly treatment sessions, or other unspecified concerns may be more important considerations. To conclude this section, two further speculations are offered concerning the correlations between Krause motivational conditions and other helpseeking variables. First, there was only a nonsignificant correlation of .14 between prior counseling experience and scores on the Krause Need for Treatment subscale. It might be expected that subjects with prior counseling experience would be somewhat more likely than inexperienced subjects to know whether they needed professional help. This weak and nonsignificant correlation might be an indication that discussions of the treatment 138 option during the EAP interviews were effective in helping inexperienced subjects appreciate their need for treatment. Second, scores on the Krause Tolerance for Treatment Demands subscale and prior counseling experience were significantly correlated (N = -.21, p < .05). The negative coefficient indicates that subjects without counseling experience were more likely than those with experience to think that they could tolerate the costs of treatment. Perhaps subjects with prior experience were more realistic about the demands of treatment, whereas those without prior experience underestimated the severity of the demands they could encounter in treatment. Perceived Social Pressure to Seek Professional Mental Health Care Social pressure to seek professional mental health care appears to have played no role in the referral behavior of the subjects in this study. There was no significant difference between the referral-noncompletion group and the referral-completion group in the level of pressure reported. The low level of pressure reported by both groups indicated that other persons in the lives of the subjects were indeed applying some pressure to encourage them to seek professional help. This finding suggests that the subjects felt the pressure, but they did not view it as an important factor in their decision to seek or not seek treatment. Spouses were the most frequently reported source of social pressure. It was anticipated that two groups within the sample would report significantly greater social pressure to seek help. This proved not to be the case. The first group, subjects who completed 139 their referral despite their lack of agreement with one or more of Krause’s. motivational conditions, was not found to have a significantly* higher 'level of‘ reported social pressure than the group that agreed with all three conditions and completed their referrals (Table 12). The other group hypothesized to be under greater pressure to seek help were subjects who completed their referral even though they were in the bottom half of the sample with respect to their ATSPPH scale composite attitude score. Their level of social pressure to seek assistance was not significantly different from the level reported by subjects whose composite attitude score was in the top half of the sample (Table 11). For the sample as a whole, however, the relationship between social pressure and the ATSPPH scale factor Stigma Tolerance (N .21, p < .05) suggests that subjects who felt less concerned about what others might think of their use of mental health services also felt less pressured to seek such help. Conversely, subjects who felt more concerned about the reactions of others also felt more pressure from others to seek help. Ironically, persons who applied social pressure to these resistant clients might have been unaware that the resistance of these individuals was partly due to their fear of how' others would react to their use of mental health services. A possible explanation for the low levels of reported social pressure in both the referral-noncompletion and referral-completion groups is that clients of the EAPs who felt high levels of social LALQ- i, 140 pressure to use the EAP and to seek professional help simply did not volunteer for this study. Clients who had not sought the help of the EAP on an essentially voluntary basis might have expressed their resistance or resentment by rejecting their interviewer’s request to become a volunteer participant. No data were available, however, to determine whether the group that flatly refused to participate or, for that matter, the group that did not return their questionnaires did in fact feel more pressured than the subjects who took part in the study and reported their feelings. Another possible explanation for the low levels of reported pressure is that the researcher’s one-item scale might not have been sufficiently sensitive to detect actual differences among the subjects. During the development of the scale, it was thought that subjects would find it difficult to relate their feelings of being pressured to seek help to more than four or five levels of pressure, one of which was the absence of felt pressure. A scale with a different format, however, might have been more sensitive to individual differences--possibly a scale that required subjects to estimate the relative proportions of intrinsic and extrinsic motivation involved in their decision to seek professional assistance. In addition, a more in-depth, useful, and reliable assessment of their perceived social pressure might have been obtained had they been asked to rate separately the pressure they felt was being applied by various significant persons in their lives. 141 In conclusion, social pressure to seek professional help did not differentiate the referral-noncompletion group from the referral-completion group. It appears to have been an unimportant variable in the decisions subjects made to seek or reject professional help. The comparatively low levels of pressure reported by both the referral-noncompletion and referral-completion groups imply that the helpseeking behavior of the research subjects was essentially self-motivated. ‘This information suggests that their decisions to seek or reject treatment were freely made--a condition that needed to be met for their helpseeking behavior to be analyzed meaningfully in terms of the Krause (1966) pretreatment conditions and the helpseeking model by Fischer et al. (1983). Prediction of Pretreatment Dropout in EAP Mental Health Referrals The results of hypothesis testing and discriminant function analysis showed that it was possible to differentiate the referral- noncompletion group from the referral-completion group on the basis of certain variables examined in this investigation. In particular, discriminant function analysis demonstrated that the information provided by the subjects’ responses to four variables (Table 14) could be used to classify correctly 77.5%. of 'the subjects who completed their mental health referral. This percentage of correct classifications was a small improvement of 2.7% over classification based solely on chance (74.8%). By using this classification procedure, however, it was possible to discriminate 48.1% of all 142 subjects who did not complete their referral, a rate almost double that of classification based on chance (25.2%). Discriminant function analysis provides the means for making statistically sound predictions concerning client referral behavior from ‘the information contained in a combination of helpseeking variables. An EAP interviewer relying on clinical judgment would probably be unable to consistently match the accuracy of this statistical method of prediction. Even based on considerable clinical experience, it would be difficult, if not impossible, for the interviewer to know exactly how much weight to give to the various helpseeking factors involved in a client’s decision to complete or not complete a referral. This weighting process is easily accomplished, however, through a discriminant analysis of the relevant clinical (predictor) variables. The discriminant functions computed for this study should not be adopted directly by other EAPs for prediction purposes. Interviewers interested in using discriminant function analysis to predict the referral behavior of their own clients should compute a discriminant function or functions from data supplied by clients from their program. The reason for doing so is that discriminant analysis. generates predictions by maximizing differences between groups. The differences between clients who do and do not complete referrals recommended by one EAP program will not be exactly the same differences found between these groups in another EAP program. The accuracy of predictions about referral behavior will be diminished to the extent that these unique group differences are not 143 taken into account. In addition, before these functions are used to predict the referral behavior of other clients from the same program, the functions should be cross-validated by means of one of the methods described by Betz (1987). To make the best possible predictions using discriminant function analysis, EAP interviewers will have to address several problems in collecting the necessary data from their referral clients. First of all, interviewers hoping to identify as many potential mental health referral dropouts as possible will need to obtain data representative of their entire mental health care referral group. Clients who would have resisted participating in the present study will therefore need to be convinced to supply the data required to generate optimum discriminant functions. A related problem is that the quality of data provided even by cooperative clients routinely completing an EAP mental health referral may not be as good as the quality of data collected during this independent research investigation. The subjects in the study were guaranteed anonymity in completing the researcher’s questionnaire. Completion of the questionnaire scales under identi- fiable conditions may decrease frankness and increase social desirability bias in client responses. This variation in responding would result in some loss of discriminant power and accuracy in predicting referral outcome behavior. To determine how much discriminant power might be lost, individual programs could compare the results of prediction based on client samples collected under anonymous versus identifiable conditions. 144 In view of these difficulties, an intriguing alternative for assessing the likelihood of pretreatment attrition is simply to ask clients whether they agree or disagree that Krause’s three conditions of psychiatric helpseeking motivation reflect their feelings about their own situation. As predicted by Krause’s (1966) theory, nearly 75% of the subjects in this study who did not complete their referral could be identified by their disagreement with one or more of the three conditions (Table 8). This percentage of correct "hits" was actually superior to the percentage of subjects classified correctly by the discriminant function analysis of their graduated responses to the three conditions (48.1%). Whether the superiority and potential utility of this simple assessment approach can be replicated is well worth investigating. In summary, the data analyses have shown that a significant percentage of the research subjects who did not complete their mental health referrals potentially could have been identified as being at risk. Client reactions to the Krause conditions may be especially useful interview information for identifying these clients. Under identifiable conditions of collection, the client information available for outcome prediction may' yield somewhat lower prediction rates than those obtained under the anonymous collection conditions used in this study. Similarly, it is not known whether simply asking clients for their responses to scale items as part of the EAP interview routine would have an undesirable or a negligible effect on their responses. 145 The Decision to Seek Help-- The HelpseekingModel In addition to the findings of past research, the Fischer et al. (1983) model of psychiatric helpseeking was consulted for guidance in selecting the variables examined in the present study. Many of the helpseeking variables specified in the third stage of the model ("The Decision to Seek [or Accept] Help") were included in the study and used to contrast the referral-noncompletion group with the referral-completion group. In this section, the hypothesized relationships between those variables and the decision to seek or not to seek professional mental health care are reviewed in light of the data analyses. Fischer et al. (1983) hypothesized that troubled individuals decide whether to seek professional help by weighing the anticipated benefits and costs of treatment. The results of the present study suggest a refinement of their model: Individuals who are channeled into treatment by an EAP interviewer tend to give more weight to certain helpseeking variables ("factors" in the language of Fischer et al.) than to others during their deliberations. Using discriminant function analysis, it was possible to determine the relative importance of these variables for the research sample. The ranking was the same based on either the standardized canonical discriminant function coefficients or the structure coefficients (the pooled within-group correlations between the variables and the discriminant function). The most important variable was the anticipated effectiveness of available treatment, 146 which Fischer et al. (1983) labeled a "therapist or agency factor." Next in importance was the severity of the problem or symptoms, a "personal factor." The least important variable was the combination of two cost factors: (1) the monetary cost of treatment and (2) the practical inconveniences associated with receiving treatment, both of which Fischer et al. referred to as "therapist or agency factors." This ranking was computed from scores on the three Krause subscales: Confidence in Treatment, Need for Treatment, and Tolerance for Treatment Demands (Table 15). The lack of significant group differences represented by ATSPPH scale scores suggests that a number of factors proposed in Stage 3 of the model do not apply to all groups of psychiatric helpseekers. The decisions of the subjects in this study to complete or not complete their referral did not depend on these factors. The factors in question are the helpseeker’s (a) attitude toward mental health practitioners, (b) embarrassment and perceived stigma, (c) anticipated loss of self-reliance (dependency) and feelings of indebtedness, and (d) belief that persons who solve their own problems are worthier than those who cannot. The one-way ANOVA test performed on the data (Table 7) indicated that there was no significant difference between the referral-noncompletion and referral-completion groups with respect to any of these factors. As previously discussed, however, these factors may be important in the case of populations that are more heterogeneous than the research sample. 147 Further examination of the data using discriminant function analysis indicated that the CSQ-8 scores ranked between the first and second Krause subscale scores irI their contribution to differentiating the referral-noncompletion group from the referral- completion group (Table 13). The information represented by these scores, however, is not specifically described and included in Stage 3 of the model. It would thus appear that the model would be improved by the addition of a factor that reflects the helpseeker’s reaction to any referral program that he or she has consulted for assistance. Three other observations are offered to conclude this discussion. First, the factors outlined in Stage 3 of the model accounted for only 20% of the variance in this research investigation. The amount of variance unaccounted for suggests that other unspecified variables make important contributions to the completion or noncompletion of EAP-recommended mental health referrals and presumably enter into the decision to pursue or not pursue a referral. Second, the dropout prevention research reviewed in Chapter 11 indicated that several simple strategies were effective in reducing pretreatment attrition. The success of these strategies suggests that the variables involved need to be identified and included in the model, if they are not already part of it. Finally, it would be enlightening to interview members of other samples who drop out, particularly if they could be interviewed soon after they make up their minds not to complete their referral. This approach would allow information about their det Fu‘ fa 148 decision to be gathered while it is fresh in their minds. Furthermore, these individuals may be able to point out additional factors involved in their decision if they do not have to limit their remarks to the variables specified in the model. Limitations of the Study EMT Validity Problems of external validity may exist in relation to this study because a volunteer sample was used. The EAP clients who took part in the study might have differed from those who did not in ways that could weaken or invalidate the generalization of findings to the target population--all EAP clients who do not complete an EAP- recommended mental health referral. Although limited demographic data were obtained regarding the client volunteers who did not return their questionnaires, the analysis of these data revealed only minor differences between this group and the volunteers who did return their questionnaires. Moreover, demographic differences accounted for a comparatively small portion of the variance in this study. Consequently, it is impossible to say whether the addition of questionnaires from clients who did not return them would have significantly changed the findings. This uncertainty in generalizing to the target population is compounded by the absence of data from other clients who openly declined to participate in this research. Completed questionnaires from these individuals would have helped to confirm whether the findings of this study 149 apply to all EAP clients who do not complete a mental health referral. Because of the lack of representation of these two groups in the research sample, the findings of this study should be considered tentative and generalized cautiously to EAP clients. These findings should likewise be considered tentative with regard to the behavior of mental health care helpseekers in other settings, such as medical facilities and public schools. Design and Methodology Several potential threats to internal validity might have affected the findings of this study. 'H: begin with, the research subjects might have been affected by the researcher’s questionnaire. The referral behavior of subjects who completed the questionnaire might have been influenced by something in the scales that would not be present to affect the behavior of clients under normal circumstances. For instance, by drawing attention to various aspects of seeking and receiving professional help, the questionnaire might have influenced some clients either to complete or not complete a referral. Another potential source of unexplained variance, the idiosyncratic effect of meeting with one EAP interviewer as opposed to another, could not be examined in this study. This particular variable might be difficult to study in future investigations, as well, because a relatively large number of subjects are needed from each interviewer (which necessitates a prolonged commitment to the rese invo nay inf' COITI' we le 16 an 150 research on the part of the interviewer) to insure that hypotheses involving small subgroups can be tested. In addition, interviewers may be reluctant to cooperate, knowing that their individual influence or performance will be singled out for scrutiny and comparison. A third shortcoming in the design of this study was the researcher’s limited control over the timing of the subjects’ completion of their questionnaires. Subjects were encouraged by their interviewer to return their questionnaires promptly. They were further' encouraged not ix: delay by the thank-you/reminder letter they received 1 week after their EAP interview. Once they left the interview, however, they themselves decided when and whether to fill out and return their questionnaires. Comparison of the dates on which the researcher received completed questionnaires with the dates on which they were given out indicated that most subjects who returned a questionnaire did so within 1 or 2 weeks of their EAP interview. The longer clients waited to complete their questionnaires, the more their thoughts and feelings about seeking professional help might have changed from what they were at the end of their EAP interview. Any delay potentially allowed various factors beyond those present at the time of the interview to influence their final decision about seeking professional help. Some delay, however, might actually have enhanced the accuracy of the data. Compared with questionnaires completed promptly, questionnaires completed closer to the time of an intake appointment might have provided a 151 more accurate reflection of the personal decision making involved in following through with or calling off the intake appointment. In future research, it would be useful to examine whether data collected right at the conclusion of the EAP interview are more predictive or less predictive of referral attrition and follow- through than data collected closer to the time of the intake session with the treatment provider. The potential problem posed by inaccurate classification of client referral behavior has already been noted. To reiterate, the subjects in 'this study appear' to have been classified into ‘the referral-noncompletion and referral-completion groups with the necessary degree of precision. Nevertheless, there might have been a. small number' of subjects identified as referral dropouts who subsequently did begin treatment elsewhere, unbeknownst to their EAP interviewer. Interviewers tracked their clients closely enough that several instances of switching to an intake with a new provider were discovered and accounted for. Finally, the researcher’s relationship to the EAPs as an outsider amounted to a limitation of sorts in the design and execution of this study. Although this arrangement fostered a certain objectivity concerning the collection and analysis of data, it precluded other kinds of analyses that might have been employed to learn more about the referral-noncompletion and referral- completion groups. In particular, the promise of anonymity and the preservation of client confidentiality prevented the researcher from 152 having first-hand contact with any of the research subjects. Had the researcher been a staff member of one of the EAPs, it might have been possible for him to conduct follow-up interviews with clients who did not complete their referrals to learn more about what caused them to change their minds. With the limited time EAP staff members had available to help with the study, it was not reasonable to ask them to become even more involved than they already were in the collection of data. Implications for Practice Several practical implications flow from the results of this investigation. An important implication concerns the finding that the dichotomized scale used to measure the Krause pretreatment motivational conditions correctly classified three-quarters of the referral-noncompletion group. This scale offers a potentially simple and effective means of identifying EAP clients 'hi need of mental health services who are at risk of not following through to receive them. If the present findings regarding this scale can be replicated with new samples, EAP interviewers would appear to have an effective set of questions for assessing the helpseeking motivation of their clients. Medical personnel, clergy, school psychologists, and other referral agents also might find these questions useful in their efforts to encourage persons in their care to seek professional mental health services. As a practical matter, interviewers using this scale for assessment should allow sufficient time during an interview to deal 153 with client concerns about treatment that are revealed by the scale. In addition, since some clients may be reluctant even to acknowledge their concerns, interviewers should take it upon themselves to open a. discussion of these concerns with all clients who are being assessed for mental health reasons. Interviewers would also be wise to stay in touch with clients during their wait for treatment so that any concerns that might develop then can be dealt with before the client prematurely decides to quit the referral. A second implication of practical significance to EAP interviewers involves the finding that the subjects’ satisfaction with their experience at the EAP was related to referral completion. In view of this finding, EAP interviewers should keep abreast of research developments concerning the determinants of client satisfaction with the assessment and referral process. They might also wish to obtain more specific feedback from their clients than the CSQ-8 scale provides. They could do so by developing their own client satisfaction scale based on a review of the literature, input from colleagues, and the observations of their clients. Interviewers should strive to meet their clients’ needs and expectations, particularly in the aspects of client satisfaction found to be most important. Third, the finding that no statistically significant relationship existed between the» cost of 'treatment services and referral outcome obviously applies only to persons who have the means to pay for professional services. It is probably safe to say that EAP clients, as employed persons, generally have employee nedi prof pruc cl i1 an COI sh 154 medical insurance that covers some or all of the costs of professional mental health treatment. Nevertheless, it would be prudent of EAP interviewers to review insurance benefits with their clients to be certain that financial considerations will not become an obstacle to completing a recommended mental health referral. Finally, the largest source of variance in this study, client confidence in the effectiveness of professional mental health care, should be a concern of all mental health professionals, not just EAP professionals. For their part, EAP interviewers have a considerable interest in referring their clients to treatment providers and treatment modalities of demonstrated helpfulness for the problems presented by each client. Given the apparent importance of client confidence in treatment, EAP interviewers and mental health providers alike must communicate to clients that duly qualified providers and their methods can be relied on to produce favorable results. The point is that professional mental health care must be worthy of consumer trust and confidence. Recommendations for Future Research Replication of all or various portions of this study is recommended to help confirm that the statistically significant findings reported do, in fact, hold up across new samples of EAP clients. Replication, particularly' with a larger subsample of clients who do not complete their EAP-recommended mental health referrals, might also uncover other significant differences between the referral-noncompletion and referral-completion groups. Future his pari adv; qua pro whc ve‘ th ti 155 researchers should also consider how they could encourage certain potential subjects to participate in their research. In particular, they will need to address the problem of recruiting subjects like those in this investigation who openly declined to participate or did not return a completed questionnaire. As mentioned above, this researcher was handicapped somewhat in his efforts to collect data by his status as an outsider to the four participating EAPs. In future investigations, it would be advantageous for the research effort to be conducted internally by a qualified member of the EAP staff. This arrangement would eliminate problems created by an independent researcher imposing on EAP staff, whose time for matters associated with data collection is apt to be very limited. With appropriate attention to the rights of clients, this arrangement would also facilitate direct communication between the researcher and the research subjects. The opportunity for direct communication between the researcher and the research subjects would permit the methodology of the present study ix: be supplemented with qualitative research, specifically interviews with subjects who do not complete their mental health referral. The amount. of ‘variance remaining unexplained in the present study indicates that helpseeking factors that were not accounted for need to be identified and included in future research. Interviews with subjects from the dropout group would provide information about their reasons for not completing their referrals. Factors identified in this fashion could be 156 investigated further to determine whether they belong in the decision-making stage of the helpseeking model and whether their inclusion in a discriminant function analysis would increase the percentage of referral dropouts who can be identified. Follow-up interviewing also would help to account for subjects who might be misclassified as dropouts in the event they have sought professional assistance from a provider other than the one their EAP interviewer expected them to see. In addition, the interviews might reveal whether subjects have changed their minds about seeking professional help and have decided instead to make use of informal sources of psychological support. Information gleaned about reliance on nonprofessional sources of assistance would help to explain (a) why subjects turned to these sources instead of to professional help, (b) how well they coped without professional treatment, and (c) how their helpseeking actions should be accounted for in terms of the Krause prerequisite conditions and the helpseeking model by Fischer et al. (1983). It is also recommended that the relatively crude scales created by the researcher for the present study be refined before being used in future research investigations. Subject-perceived social pressure to seek mental health treatment might be assessed with greater sensitivity and thoroughness if the researcher’s scale were redesigned to measure each reported source of pressure. Similarly, intrinsic motivation to seek professional mental health care might be better assessed with a longer scale than the one devised for this study. Despite the moderately low reliability of that scale, the 157 classification results achieved with it were surprisingly good. Further empirical testing is needed, however, to evaluate the psychometric benefits of using multiple items to measure each motivational factor. Finally, longitudinal research is needed to trace and explain the development of pretreatment dropout. The administration of the research scales in the present study provided a single observation of the relationships between several helpseeking variables and referral behavior. This design