Recruiting and retaining direct-care staff has long been a challenge for administrators of programs that serve persons with disabilities. A literature review revealed that beginning in 1950, researchers set out to develop a paper-and-pencil test that would identify good employees. Regardless of how they defined good, employers were unable to identify tests that selected promising job applicants. Efforts to develop research-based selection tools decreased in the early 1970s with the transition from state institutions to community programs. However, the growth of community programs did not mitigate the turnover problem, and the need remains for a reliable hiring tool. Whereas recent researchers have identified practices that can reduce turnover, these practices should be more widely implemented and their efficacy examined.
Editor in charge: K. Charlie Lakin
The importance of hiring good direct-care staff members to work with persons who have developmental disabilities or mental illness has long been acknowledged. More than 60 years ago, Hadley and Proctor (1940) noted that the lack of proper personnel to do the basics leads to trouble, and Bonner (1949) stated, “One of the secrets of better mental hospitals in the future lies in the better selection and training of ward employees” (p. 669). Cleland and Peck (1959) understood why ward attendants are important: “Ward attendants might be on the lowest rung of the ladder, but to the patients on the ward the attendants assume an importance rarely matched in our society” (p. 877). Later researchers have not wavered in their estimation of the importance of direct-care staff (Cuadra & Reed, 1957; Garland, Oyabu, & Gipson, 1988; Schmidmayr & Weld, 1971). Larson, Hewitt, and Anderson (1999) put it succinctly when they wrote, “Direct-support workers are the backbone of the service provision to people with developmental disabilities” (p. 36).
Given the importance of direct-care staff, over the past 50 years considerable efforts have gone into determining how to hire and retain them. Unfortunately, there still are no well-researched tools for hiring applicants who will be good direct-care staff members, and most administrators deal with unacceptably high rates of turnover. We reviewed the literature in order to understand the past efforts to hire and retain direct-care staff. Our expectation was that an analysis of past efforts would provide guidance and set directions for future research.
Efforts to Develop a Paper–Pencil Test to Select Applicants
A comprehensive review of the journals devoted to research on persons with developmental disabilities revealed that beginning in 1950 and running through 1967, researchers sought a paper-and-pencil test that would discriminate between applicants who would be poor and those who would be good employees. Some researchers attempted to determine whether existing and commonly used assessment instruments, such as the Minnesota Multiphasic Personality Inventory (MMPI) or the Wechsler-Bellevue Intelligence Test, would be effective hiring tools. Other researchers sought to develop instruments that would identify good applicants. Four basic approaches were used.
One approach to developing a selection tool is to administer numerous paper-and-pencil test questions and then, post hoc, determine those items that were answered differently by applicants who proved to be poor and those who proved to be good employees. Those items, when validated in a follow-up study, would comprise a paper-and-pencil tool with good predictive validity. Kline (1950) made use of empirical keying when he administered the Personality Inventory Test to 108 newly hired attendants and 2 years later used supervisor ratings to divide the attendants into unsatisfactory and satisfactory employees. Post hoc, he determined that 28 items on the test were answered differently by the unsatisfactory compared to satisfactory employees. Kline referenced plans to cross-validate the predictive value of these 28 items. However, no follow-up study appeared in the literature. Cuadra and Reed (1957) administered the California Personality Inventory to 332 psychiatric aides hired over a period of 2.5 years. After 21 weeks on the job, the supervisors classified the recently hired aides as very poor, poor, fair, good, or excellent. The criteria for classification were length of service and ward performance. Of the 472 items on the inventory, 43 were found to differentiate between short- and long-term employees, and 47 items discriminated between employees with unsatisfactory and satisfactory supervisor ratings. However, upon cross-validation the initially discriminating items no longer predicted either tenure or job performance.
Mental ability assessments
In this approach the mental ability of each applicant is measured in order to test the hypothesis that mental ability is correlated with either length of stay or job performance. Horner (1954) studied whether the Wonderlic Personnel Test, a quick measure of intelligence, correlated with the length of employment or supervisors' ratings. After 3 years of research, he could not find any correlation between the scores on this instrument and length of service, supervisors' ratings, or employment status. As an incidental finding, he learned that attendants with above average mental ability tended to resign if they did not quickly progress to a higher position. Barron and Donohue (1951), using the Otis Quick Scoring Mental Ability Test, found that the best attendants, according to supervisor ratings were those who had a 10th-grade education and whose mental ability was within the dull–normal classification. Yerbury, Holzberg, and Alessi (1951) administered the Revised Beta Examination, an intelligence test, to a group of recently hired psychiatric aides. After 6 months on the job, the employees were rated by their supervisors and classified as either poor or good. They found that the Revised Beta Examination did not discriminate between poor and good employees. Similarly, Steinberg, Wittman, Wenig, and Prentice (as cited in Shotwell, Dingman, & Tarjan, 1960) found the Wechsler–Bellevue valueless for selection purposes.
Researchers also studied whether identifiable personality profiles were characteristic of applicants who would turn out to be poor employees. Barron and Donohue (1951) determined that the MMPI was effective in eliminating psychopaths who applied to become attendants but was not useful in predicting the job rating of the psychiatric aides who were hired. Butterfield and Warren (1962) classified 2,700 direct-care staff members into three groups: discharged, resigned, and satisfactorily employed. They found that the MMPI failed to differentiate among the three groups of attendants. Moreover, they found that scores on this instrument of the not-hired group were not significantly different from those of attendants who were discharged, who resigned, or who were still employed. Dismayed that the highly used MMPI did not predict the groupings of the hired applicants, Butterfield and Warren (1963) did a follow-up study. They administered the MMPI to 109 new employees. Six months later they divided the employees into three groups: discharged, stayers with poor evaluations, and stayers with good evaluations. They found that many applicants in the discharged group scored in the “clinically significant” range on four of the MMPI scales. Specifically, the discharged applicants scored high on the K Scale and the L Scale, meaning that they endorsed statements on the MMPI that falsely indicated that they had rare, sterling qualities, and they denied other statements in order to suggest that they did not have the slightest hint of psychopathology. The discharged applicants also scored high on the Paranoia Scale, indicating that they were by nature suspicious of the motives and intents of other people; and they scored high on the Mania Scale, indicating they had excessive energy, which they were apt to manifest impulsively. In short, the MMPI screened out applicants who were deceitful or overtly disturbed; but it did not discriminate between applicants who proved to be poor versus good attendants.
Yerbury et al. (1951) administered the Multiple Choice Rorschach to newly hired psychiatric aides. Using supervisor ratings at 6 months postemployment, they found that this measure did not differentiate between poor and good aides. Tarjan, Shotwell, and Dingman (1955) developed the Work Assignment Aid, which tapped such traits as ego strength, aggressiveness, radicalism, neuroticism, and intelligence. After 5 years of research, they concluded that none of the subscales on the Work Assignment Aid discriminated between unsatisfactory and satisfactory employees. They determined that their research proved the inability of this instrument to predict the job success of new attendants (Tarjan, Shotwell, & Dingman, 1956). Similarly, Cuadra and Reed (1957) found no correlation between the scores obtained by psychiatric aides on the California Personality Inventory and length of employment or supervisors' ratings. Cattell and Shotwell (1954) found only modest success using the Sixteen Personality-Factor Questionnaire to predict supervisors' ratings.
Researchers also sought to determine whether applicants with particular backgrounds were more likely to stay at their institutional job. The investigators consistently found that there was a distinct stayer profile for attendants at large, state institutions. Stayers had a patriarchal family orientation, came from a rural background, and showed greater acceptance of the authority of their parents (Cleland & Peck, 1959; Kimbrell & Blanchard, 1964). Stayers also tended to be young and single (Kline, 1950). If hiring for the night shift, birth order was a factor. Later-born applicants were less likely to quit the night shift than were firstborn applicants (Cleland, Seitz, & Patton, 1967).
The search for paper-and-pencil tests that would identify good direct-care staff was discontinued in the late 1960s, probably for two reasons. First, the deinstitutionalization movement began in the 1970s, and attention shifted away from the problems deemed characteristic of institutions. Second, the efforts to find a paper–pencil selection tool were probably discontinued because they had failed. As Butterfield (1967) concluded at this time, the preponderance of the research indicates that there is no paper-and-pencil test that predicts either length of tenure or supervisors' ratings. Since then, the situation has not changed.
Analysis of the Failure
The search for a tool to select good direct-care staff was driven by two concerns: high turnover and the impact of staff members on ward functioning. Even though no selection tool was discovered, these needs still existed. There is no well-researched tool for hiring direct-care staff, and turnover among direct-care staff in community programs is excessively high (Mitchell & Braddock, 1994). High turnover is a problem because it has a deleterious effect on services to persons with developmental disabilities (Hatton et al., 1999) and it is costly (Larson, Lakin, & Bruininks, 1998; Pillemer, 1996). Perhaps the first step toward mitigating these problems is to understand why the early attempts failed.
Poor Predictor Variables
One reason for the failure to identify a selection tool was that early researchers did not adequately define their predictor variable: good staff. Good staff meant those who stayed; and it also meant staff members who got high job-performance ratings. The reasons that the search for a hiring tool failed were due, in part, to their definition of good.
One criterion of good staff members was their length of tenure. Facing excessive turnover, the early researchers decided that a good attendant was one who stayed. So they sought to develop an interview tool that would identify “stayers.” Indeed, in one study, Yerbury et al. (1951) went so far as to remove nonstayers from a group that supervisors had rated as good and place the former attendants in the poor group. These staff members were switched to the poor group because “turnover of ward personnel was [such] a serious administrative problem [that] by virtue of not holding their jobs for at least six months they could be considered undesirable employees” (p. 92).
This search for stayers was based on a false assumption. The assumption was that that an attendant's decision to leave a job was an inherent part of his or her personality and, by implication, employees who left were flawed (Melbin, 1961). A close examination of the findings did not support the assumption that leavers were flawed. Researchers often found that the job performance of voluntary leavers was comparable to the performance of stayers (e.g., Cope, Grossnickle, Covington, Durham, & Zaharia, 1987; Kline, 1950). Horner (1954) even noted that a greater percentage of the desirable rather than the undesirable attendants resigned within the first 6 months.
Unreliability of supervisor ratings
The other criterion used to define good staff was supervisor ratings. However, supervisor ratings were not reliable. A supervisor's ratings of the same attendants did not remain constant over even a short period of time, and their ratings were not reliable from setting to setting (Tarjan et al., 1956). Clearly, the measurement of the predictor variable must be reliable over time and across settings.
Questionable validity of supervisor ratings
The supervisors at institutions were generally the nurses in charge of the wards. Head nurses valued direct-care staff members who were neat and clean in personal appearance, who arrived for work on time, who kept the unit clean, and who understood who was in charge (Butterfield, 1967; Shotwell et al., 1960). For example, Cliff, Newman, and Howell (1959) referenced the criteria supervisors used to assess direct-care staff. Of the 18 criteria, 7 measured the attendants' compliance with institutional regulations and procedures (e.g., attendance and promptness; accepting changes in assignment). Only one criterion dealt with relating to the patients (e.g., awareness of patients' needs.) In an article that captured the Zeitgeist of the institutional days, Edgett (1951) stated that the ideal attendant was neat, compliant, and punctual.
Shotwell et al. (1960) were among the first to question whether attendants who were neat, compliant, and punctual were really good staff members. There was a sense among the researchers that “good” should have less to do with maintaining the ward and more to do with patient care. As Butterfield and Warren (1963) suggested, perhaps the ultimate criterion for determining that staff members are good is whether they bring out the best in the persons for whom they care.
Confounded predictor variables
The final and perhaps most important reason that the early researchers failed to develop a tool for selecting direct-care staff members is that they regarded staying and high ratings by supervisors to be synonyms. Indeed, many of the researchers simultaneously used staying and supervisor ratings as predictor variables. In other words, they sought an assessment tool that would identify stayers and would also identify direct-care staff members who were highly rated by their ward supervisor. In retrospect, there was no compelling reason to think that staying and job performance are related. In fact, they are not. Turnover and hiring staff who perform well at their job are unrelated issues. Each issue needs to be examined separately.
Turnover in community programs
Changing the site of the caregiving in the 1970s from institutions to the community did not mitigate the turnover problem. A search of the primary journals in the field from 1970 to 1990 coupled with a search of the psychINFO database from 1990 to present using the search terms employee turnover followed by mental retardation revealed that there is a high rate of turnover among direct-care staff members in community programs (Baumeister & Zaharia, 1987; O'Connor & Sitkei, 1975; Zaharia & Baumeister, 1978). Lakin, Bruininks, Hill, and Hauber (1982) sampled turnover at 236 community residential facilities. The average rate of direct-care staff turnover was 80% per year. Some agencies reported turnover as high as 400%. In a nationwide survey, administrators of 2,191 community facilities reported that hiring and retention were their biggest problems (Bruininks, Kudla, Wieck, & Hauber, 1980).
The turnover problem continued into the past decade. In a nationwide study of publicly and privately operated community facilities, Mitchell and Braddock (1994) found that the average annual rates of turnover were 34% and 70%, respectively. In a study of 110 small group homes in Minnesota, Larson and Lakin (1999) found turnover at 46%. The average small group home in Minnesota had 6% of their direct-care positions unfilled at any one time (Larson, Hewitt, & Anderson, 1999).
Turnover in nursing homes
Turnover is not just endemic to programs serving persons with developmental disabilities. A search of the CINAHL database from 1985 to present using the word personnel combined with turnover coupled with a search of Medline using the words personnel turnover combined with nurses aides identified 37 studies. Among other conclusions, the researchers reported that nursing homes also experience high turnover of direct-care staff members, ranging from 50% to 400% annually, with the average turnover being between 55% and 70% (Almquist & Bates, 1980; Cohen-Mansfield, 1997).
There is a “leaver” profile among direct-care staff members in community programs. Leavers are more likely to be younger (Caudill & Patrick, 1991; Hatton et al, in press; Wagnild, 1988), better educated (Larson & Lakin, 1992), and have shorter tenure (Lakin, Bruininks, Hill, & Hauber, 1982; Razza, 1993). One suspects that these young, better educated people who leave community programs are more mobile and more marketable than are the stayers.
Low pay for direct-care human service providers is consistently correlated with turnover (Lakin, 1988; Larson & Lakin, 1999; Mitchell & Braddock, 1994) and job dissatisfaction (Kay, 1994; Rublee, 1986). Turnover is also inversely correlated with the local unemployment rate (Butterfield, Barnett, & Bensburg, 1966).
Researchers have determined that management practices also impact turnover (Butterfield, 1967; Cohen-Mansfield, 1997; Hughes & Flowers, 1987; Ziarnik, 1980). Indeed, the impact of management practices on turnover is so strong that George and Baumeister (1981) concluded that “turnover …is a symptom of underlying dysfunction in the organizational structure of community residential services” (p. 645). Chapin (1999) concurred.
Researchers have contended that management controls many of the factors that prompt good staff to resign (e.g., Asis, 1975, cited in Georgopoulous, 1973; Bachelder, 1993; Reagan, 1986). Bales (1975) found that direct-care staff members leave an agency because of insufficient orientation, no career ladder, little expressed appreciation, and inadequate training. Similarly, other researchers queried direct-care staff members and found that job satisfaction, intention to quit, and turnover are associated with little opportunity for professional growth (Alexander, Lichtenstein, Oh, & Ullman, 1998), minimal support from supervisors (House, 1990), little opportunity for input (Banaszak-Holl & Hines, 1996; Helmer, Olson, & Heim, 1993), a poor understanding of the organization's mission (Hatton & Emerson, 1993; Razza, 1993; Sharrard, 1992), and lack of clarity about job duties (Jamal, 1984).
Hiring Good Staff
Effects of Staff on Clients
More than 30 years ago, Butterfield (1967) commented that those familiar with institutions for people who have mental retardation have seen the wonders and the horrors that employees can work upon their residents. Indeed, there is a large body of literature supporting the position that staff members impact both clients' behavior and their quality of life.
By way of overview, Tizard (1953) was among the first to point out that staff members' behavior affects the behavior of the people in their charge. He examined the effects of supervisory style on work productivity and overall adjustment of adolescents who had mental retardation. Tizard found that strict supervision worked well for 7 of the 11 boys; but for the other 5 boys strict supervision was associated with depression, attempts to run away, less work accomplished, and psychosomatic illness. Laissez faire supervision was associated with a loud, unproductive work unit. Friendly supervision resulted in a happy, productive work climate, with cohesion and support among the boys. Similarly, Mayhew, Enyart, and Anderson (1978) found that the residents' social behaviors were increased through staff reinforcement. Willer and Intagliata (1981) found that when residents' expressions or feelings were controlled via punishment, the residents were more likely to exhibit negative behavior. As a final example, Owen, McDonald, and Baine (1994) determined that direct-care staff members who used questioning-type speech initiated more conversation from adults with developmental disabilities than did those who used direct commands.
Other studies have shown that staff can enhance the well-being of clients. Specifically, Bogdan, Taylor, DeGrandpre, and Haynes (1974) determined that the success or failure of care programs in a mental hospital largely depended on the attitude of the attendants. Schalock, Harper, and Genung (1981) identified persons with disabilities who, in spite of maladaptive behaviors, were able to remain in a community setting because the direct-care staff members had patience, tolerance, and commitment.
Who are Good Staff?
Compliant, punctual, and neat might have defined good staff members when the implicit mission was to warehouse persons with developmental disabilities. However, the principle of normalization demands more of the staff (Perrin & Nirje, 1985). Normalization, as modified by Wolfensberger (1972), means the “utilization of means which are as culturally normative as possible, in order to establish and/or maintain personal behaviors and characteristics which are as culturally normative as possible” (p. 28). Among other things, normalization suggests that human service providers have the responsibility to promote socially appropriate behavior in people with developmental disabilities. Clearly, good staff members are expected to enhance the well-being of persons with developmental disabilities.
Toward Selecting Good Staff
A person's ability to be a helper depends heavily upon his or her interpersonal relationship skills (Carkhuff, 1969). Good interpersonal skills are so important for caregivers that Hastings, Remington, and Hatton (1995) included relationship skills in a model they proposed for understanding the job performance of staff serving persons with developmental disabilities. By implication, some direct-care staff members have a value system about human relations that augments their ability to enhance the clients' quality of life, and others do not.
Programs would certainly benefit if only applicants were hired whose inherent value system about human relations was consistent with enhancing the well-being of persons in their care. Appropriately, when the Community Support Skills Standards Project (Taylor, Bradley, & Warren, 1996) identified 12 competencies that human service providers should possess, 7 of the competencies dealt with relationship skills. Specifically, a human service provider should empower the individual he or she is working with, be a sensitive communicator, facilitate the participant's access to services, promote the development of the participant's life skills, be an advocate, support the participant's career and education development, and conduct crisis intervention in ways that defuse the situation and preserve the participant's dignity. However, we not aware of any valid and reliable hiring tool that enables administrators to identify the applicants who have these relationship skills.
Articulate the mission
Administrators can impact many of the factors associated with turnover. The foundation for impacting turnover is facilitating the development of the agency's mission, articulating the mission, and then aligning the organization so that it can implement that mission (Schwahn & Spady, 1998). The importance for community program administrators to clearly define their mission and to align both resources and worker expectations with their mission is well-documented (George & Baumeister, 1981; Hatton et al., 1999; Knowles & Landesman, 1986; Slater & Bunyard, 1983). Indeed, there is evidence that direct-care staff members who understand that their mission is to promote the development of people with developmental disabilities, who view their agency in a positive light, and who feel supported and recognized for their efforts are not leavers (Garland et al., 1988; Porter, Steers, Mowday, & Boulian, 1974).
Improve human resource practices
Konczak, Pelley, and Dossett (1990) and Hatton et al. (in press) found that lack of job training, lack of job clarity, and minimal support from supervisors are causally associated with turnover intention. Similarly, investigators have found that lack of role clarity for direct-care staff is associated with burnout (Blumenthal, Lavender, & Hewson, 1998). Stryker (1982) studied the effect of managerial interventions on personnel turnover in 19 nursing homes. He found that nursing homes with low turnover had good training for the supervisors, supervisors who gave feedback to attendants, and clear procedures for implementing expected direct-care duties. Investigators have also found that in-service education (Price & Mueller, 1981), development of a career ladder (Kohn & Brache, 1982; Myers & Myers, 1984), opportunity for forming relationships with the clients (Bersani & Heifetz, 1985), and involvement in care planning (Banaszak-Holl & Hines, 1996; Blumenthal et al., 1998) improve morale, reduce absenteeism, and decrease turnover.
Assess management skills
It would be useful to conduct a nationwide survey of administrators of community programs to assess their needs for training in strategic planning and human resource management. The high rate of turnover combined with its known causes suggests that a survey would find that many administrators lack skills in strategic planning and human resource management. Well-designed training to enhance administrators' skills in these areas probably would result in lower turnover.
Prepare for turnover
Because the money for the care of persons with special needs is limited, direct-care staff members are often paid only a little more than minimum wage (Caudill & Patrick, 1991). It is an entry-level position often taken by young people until they can get a higher paying job. In short, a certain amount of turnover can be expected. Administrators must prepare for it by expediently bringing new employees to a high level of job performance through orientation training that articulates the mission, clarifies the expectations, and builds needed skills (Knowles & Landesman, 1986).
Hiring Good Staff
Utilize situational questions
Questions designed to solicit responses to situations that reveal the applicants' inherent value system about human relationships are useful. For example, with regard to the Community Support Skills Standards competence about crisis intervention, it would be useful to give this question to an applicant.
Two 7th-grade boys are calling each other names and pushing each other around in the hall. Seeing this, a teacher walks up to them and asks, “Boys, where are you supposed to be right now?” What is your opinion of how the teacher handled this situation?
An applicant who said that the teacher mishandled the situation because she let the students get away with misbehaving is likely, if put into a similar situation, to escalate the problem. Alternatively, an applicant who said that the teacher handled it fine because she prevented a fight and got the boys redirected to class is likely, if faced with a similar situation, to defuse the situation and efficiently restore order. With regard to the Community Support Skills Standards competency of participant empowerment, it would be useful to know an applicant's response to this situation: “You are the supervisor of a group home. A participant will be moving from the group home into an independent living apartment in three months. What will you do to help the participant be successful in this new setting?” The applicant who says that he or she will determine the skills the participant needs to be successful in the apartment and then work with the participant to develop any missing skills would be a good hire.
The first author has developed situational questions for hiring good staff. The interview consists of four sets of questions, with 6 questions in each set. In order, the questions in each set assess the applicant's inherent value system with respect to not punishing people (Hall & Braun, 1988), ensuring success (Dreikurs & Soltz, 1964), allowing for and promoting independent decision-making (Perske, 1972), using logical consequences (Dreikurs & Grey, 1968), gently intervening when there is a significant behavior problem, and teaching for behavior change (Hall, 1989). The four sets are presented in increasing levels of difficulty. Good answers to the first two sets of questions, a total of 12, indicate that the applicant's inherent value system about human relations is consistent with enhancing the well-being of persons with developmental disabilities. Good answers to the next two sets of 6 questions indicate that the person has the skills to enhance the well-being of persons with special needs.
We have unpublished data indicating that the instrument has good interrater reliability. Three raters independently listened to audiotapes of 30 applicants, scoring each of the 24 questions on a 4-point scale in accordance with the established “listen fors.” The interrater reliability among the three raters was .86, .88, and .89 for each applicant's total score.
Situational questions have a second and equally important attribute. They reveal each applicant's unique pattern of strengths and weakness. This profile can be used to design a supervision plan and tailor the applicant's on-the-job training.
There are two salient findings that emerge from this review of the literature on hiring and retaining direct-care staff. First, hiring good staff and reducing turnover are two separate issues, each with its own solution. Second, the problems that initially prompted the extensive research on hiring and turnover still exist. This review suggests ways to apply these research findings to practice. With respect to hiring, there is also a need to develop a reliable tool for assessing the impact of direct-care staff on the well-being of persons with developmental disabilities, and there is a need to determine whether situational questions asked during the interview predict the applicant's caregiving skills. With respect to turnover, administrators control many of the factors that lead to turnover, which suggests that it would useful to develop systematic training to equip administrators with skills at strategic planning and human resource management and then determine whether this training reduces turnover.
Authors:Philip S. Hall, PhD ( email@example.com), Director, School Psychology, and Nancy D. Hall, EdD, Vice President of Academic Affairs, 500 University Ave. W., Minot State University, Minot, ND 58707