| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
a School of Public and Environmental Affairs, Indiana University–South Bend, South Bend, IN
b School of Business, Andrews University, Berrien Springs, MI
Correspondence: Douglas A. Singh, PhD, Associate Professor, School of Public and Environmental Affairs, Indiana University–South Bend, Wiekamp Hall–Room 2259, 1800 Mishawaka Avenue, P.O. Box 7111, South Bend, IN 46634-7111. E-mail: dsingh{at}iusb.edu.
Decision Editor: Vernon L. Greene, PhD
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionReferences |
|---|
Key Words: Recruitment • Licensure • Attitudes
One of the most serious and persistent problems confronting health care organizations today is that of selecting and retaining employees who will render a long period of service to their organizations (Ulschak and SnowAntle 1992
). Although this problem has received attention, particularly in hospital settings and among nursing and nurse aide staff, concern about turnover and retention among nursing home administrators (NHAs) has not been addressed adequately.
The administrator plays a key role in the operational success of a nursing facility. A model tested by Singh, Amidon, Shi, and Samuels 1996 provided some evidence that certain administrator characteristics and behavioral factors influence the quality of patient care delivered in a nursing facility. This finding underscores the critical role administrator selection can play in improving quality and maintaining consistency in high quality standards. While one would expect a prospective administrator's qualifications and skills—specifically the level of skill development in the key domains of practice established by the National Association of Boards of Examiners for Long Term Care Administrators—to be a primary consideration in the selection process, stability of the administrator is an equally important factor in promoting quality in nursing homes (Singh et al. 1996). Not only have administrator stability, retention, and turnover received scant attention in literature, but the extent of annual administrator turnover indicates that the industry has done little to address the problem. Because approximately 40% of facilities change administrators each year (Singh and Schwab 1998), it is little wonder that quality-of-care issues remain the predominant concern for consumers and the government alike. Nursing home governing bodies, which are ultimately accountable for facility outcomes, can improve quality in their facilities by taking steps to counteract the destabilizing influence of high administrative turnover. To accomplish this goal, greater efforts must be made to select administrators who are both competent to fill the position and likely to remain with the facility.
This study has two main objectives: First, to determine which factors, attitudes, and personal characteristics of NHAs are associated with tenure in the administrator position, and second, to construct a predictive model that can help decision makers avoid hiring administrators who are likely to leave their positions prematurely. We will also discuss implications of the findings for long-term care management and licensure policy.
Factors associated with low versus high tenure are evaluated by studying administrators who held a previous NHA position and subsequently moved to a similar position with another facility. NHAs who left the industry, got promoted to upper management positions within the industry, moved to another state, or retired could not be included in the study. Attitudinal factors related to the administrators' previous job environments, past patterns of job stability, and individual characteristics were evaluated to identify the variables that can help predict future employment stability. For a discussion of factors associated with length of employment in current positions, we refer the reader to an earlier study (Singh and Schwab 1998).
|
Methods |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
A total of 552 NHAs completed and returned the survey for an effective response rate of 53.3%. The initial mailing, which yielded a 41% response rate, clearly identified each respondent by name and facility. Before the mailing began, some industry representatives had cautioned us that administrators were likely to view as sensitive certain items on the survey, and therefore might not participate even though we had assured confidentiality in the cover letter accompanying the survey. Hence, we did not identify the respondents in the second mailing, which was sent only to the initial nonrespondents and was undertaken 4 weeks after the initial mailing. Another 12.3% of the total target population responded to the second mailing. Because the remaining nonrespondents were unidentifiable, a third mailing could not be pursued.
Returns in mail surveys generally do not follow a set pattern. For example, 4 weeks after the mailing, cumulative response rates for two commercial surveys sent to research firms and purchasing departments were found to be close to 25% (Zikmund 1997). A 2.5-page survey of NHAs in South Carolina produced a cumulative return of 52% after the first 4 weeks, an additional 9% after 8 weeks, and an additional 14% after a third mailing and telephone follow-up, for a cumulative response rate of 75% (Singh 1997b). In our study, return rates of 41% after 4 weeks and an additional 12% thereafter lead us to believe that participation was not hampered by the nature of questions on the survey. The four-page length of the survey perhaps had a bearing on people's willingness to take the time to complete the survey.
Research experts have argued that mail surveys may not be reliable unless they either achieve a minimum of 50% response or demonstrate with some form of verification that the nonrespondents are similar to the respondents (Erdos 1970). This study slightly exceeds the generally accepted minimum response rate, but may still leave some concern regarding inherent response biases. However, comparisons against available indicators for the targeted population provide some evidence that the sample is representative. Proportional responses from the two states were almost identical (52.8% from Michigan; 53.2% from Indiana). The annual job turnover among respondents is 38.5%, compared to 40.1% in the targeted population (see Singh and Schwab 1998). Facilities experiencing two NHA job changes in one year constitute 7.0% of all facilities in the sample, compared to 6.5% in the population. The proportions of men and women in the sample correspond to those in the population (men in the sample = 44.3%; population = 42.3%; test of proportions = .423, p = .20). However, not-for-profit facilities are overrepresented in the sample (33.1% vs 25.8% in the population; 
2 = 15.52, p = .000). But, information on past facility affiliation is unavailable in the data set, so any bias on account of affiliation remains unknown.
To study job factors that influence NHA tenure, the survey included 41 questions (rated on a 4-point numerical scale) pertaining to the administrator's job environment in the position held immediately preceding the current position. These attitudinal questions were adapted from literature in organizational theory which suggests that decisions to change employment are associated with a number of perceived individual, organizational, and environmental considerations. Limited by a four-page, letter-sized questionnaire, we included as many pertinent variables in the measurement scale as we possibly could. The survey questions were designed to measure NHAs' perceptions of job satisfaction and organizational commitment (Abelson 1996; Camp 1994; Mowday, Porter, and Steers 1982); job demands, job expectations, and skill compatibility (McEvoy and Cascio 1985; Mobley 1982); value congruence (Sims and Kroeck 1994); decision-making autonomy and discretion (Price and Mueller 1986; Weil and Kimball 1995); performance accomplishments, recognition, career progression, and remuneration (Cordero, DiTomaso, and Farris 1994; Cotton and Tuttle 1986; Kerr and Slocum 1987; Porter and Steers 1973); and family expectations, attachment to the community, and community involvement (Abelson 1996; Mobley 1982; Mowday et al. 1982).
A factor analysis reduced the 41 variables to seven principal dimensions that summarized how NHAs viewed their former job environments. The seven factors or perceptual dimensions extracted from the data were named Realized Expectations, Commitment, Skill Compatibility, Career Opportunities and Rewards, Personal Time, Performance Outcomes, and Community Attachment. Four of the 41 variables did not associate with any of the seven factors. These single, unfactored measures represent market competition, stress, individual effort, and involvement in local community organizations. Six of the 41 questions asked the respondents to evaluate how closely their own values and philosophies matched those of the company in the six domains of practice—resident care, financial management, personnel management, marketing and census development, physical resource management, and overall organizational management. Initial factor analysis results showed little variation among these measures. Therefore, only one of the measures (overall organizational management) was retained in the final factor analysis. The principal components method using varimax rotation provided the most satisfactory solution in view of the theoretical considerations discussed earlier. The dimension scales derived from the 36 questions (five of the 41 questions were deleted) along with their eigenvalues and reliability indicators are reported in Table 1 . Scores for some of the measures required reversal to maintain consistency with other questions. The questions requiring this transformation were precoded on the survey and are identified in Table 1 . Mean scores and standard deviations for the perceptual dimensions are also reported in Table 1 .
|
= .45) presents a concern. As expected, a proportionately larger number of married respondents answered the first question, "my spouse had a satisfying job" (Table 1 ); for the remaining two questions, proportional responses closely follow the proportion of married and unmarried NHAs in the sample. However, the problem of internal consistency in the measurement of this dimension remains even after the first question is deleted from the construct, or when the reliability computation includes only those who are married. The construct, however, was retained due to theoretical considerations (Abelson 1996; Mobley 1982; Mowday et al. 1982). Another issue relates to the validity of the self-reported measure, Performance Outcomes. The objectivity of the responses could not be externally verified. This construct has the highest mean score (see Table 1 ), which is significantly higher than the mean scores for any of the other constructs. This may indicate an upward bias in the reporting of performance outcomes. Another conceivable source of bias relates to problems with recall because roughly 16% of the respondents had been separated from their previous positions for more than 5 years. But, we do not think this presents a major concern because the questionnaire asks for perceptual and attitudinal impressions rather than specific recollection of quantitative information. Of those who responded, 306 administrators had held at least one NHA position prior to their current position. Elimination of those who were not permanently appointed employee administrators left 290 usable cases. The attitudinal data and personal characteristics of this group were analyzed using logistic regression to extract variables that would predict tenure. The binary dependent variable used in the logistic regression model was whether the NHA stayed in his/her previous position for less than 3 years or at least 3 years. Because our data include the actual length of employment, the derived model could be assessed for accuracy in its classification of NHAs as either "potential leavers" (occupied the position for less than 3 years) or "potential stayers" (stayed in the position for 3 years or longer). During the model- building process, the following variables were evaluated to get the best predictive model: age; gender; marital status; years of experience; education (two dummy variables with bachelor's degree as the reference category); field of academic concentration; number of NHA jobs previously held; nature of separation (voluntary or involuntary); nature of move (two dummy variables with lateral move as the reference category); and the perceptual dimensions obtained by factor analysis. For the "nature of separation" variable, voluntary separations include administrators who either left the organization voluntarily or asked for a transfer within a multifacility organization. Involuntary separations include NHAs who were discharged, forced to resign, or involuntarily transferred to a different facility within the chain and the transfer was viewed as a demotion. Only 1 of the 25 involuntary transfers resulted in a demotion. It is interesting to note that 21 of the 25 involuntary transfers were promotions. Because these involuntary transfers are career advancements that most NHAs would welcome, they are likely to reflect attitudes and motivations that are different from other types of involuntary transfers. Therefore, these promotional transfers were not included in involuntary separations.
The 3-year marker to separate potential leavers and potential stayers was based on actual job changes that occurred in the subject facilities during a 12-month period after the initial data had been collected. Facilities were monitored for a period of one year, and each facility in which change in administrators had occurred was identified using rosters obtained from the respective state health departments in Michigan and Indiana (regulations require nursing facilities to notify their respective state health departments whenever a change in administrator occurs). The dichotomous variable indicating a change or no change was cross-tabulated with the NHA's length of employment obtained on the original survey (the second mailing was blind, so the cross-tabulation includes 385 cases; 124 administrators could not be identified and their tenure could not be cross-tabulated). The odds of leaving versus staying pointed to a marked decrease in the odds of leaving after 3 years of continuous employment (Table 2 ). Administrators were found to be almost twice as likely to leave sometime during their first 3 years of employment. After 3 years of continuous employment at a facility, the odds of leaving are cut in half.
|
|
Results |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
|
The characteristics of potential leavers are of particular interest, and, within this group, there are some notable differences between voluntary and involuntary separations. Of all voluntary moves among potential leavers, 56.6% resulted in subsequent promotions, 40.7% resulted in lateral moves, and 2.7% led to demotions. When the separation is involuntary, the subsequent position is more likely to be lateral (in 69.4% of cases). A smaller number (19.4%) get promotional opportunities and 11.1% settle for a lesser position. Therefore, it seems that a significant number of voluntary separations are initiated to pursue career advancement. The profile of administrators who leave voluntarily within the first 3 years of employment in pursuit of promotional opportunities was found to be very similar to those who leave for other reasons, except that voluntary leavers who pursue promotional opportunities have less overall experience (number of jobs held and total years of experience). Approximately 14% of all administrators find advancement opportunities through transfers within the same organization.
Logistic Regression Model
Independent variables in the logistic regression model explain about 33% of the variability in the dependent variable (Table 4 ). In other words, average length of employment in each of the previous positions held, nature of past separation (voluntary or involuntary), commitment to the organization, facility performance outcomes, and community attachment together account for 33% of the variation in whether an administrator would stay for at least 3 years. Based on observed length of employment, the model makes accurate predictions about future tenure approximately 72% of the time. However, the model was more successful in predicting potential leavers than potential stayers. The model uses a threshold predictive value of .50. In the classification table (Table 5 ), the probability is at least .50 that 110 (72.4%) of the 152 cases predicted to be potential leavers (i.e., tenure of < 3 years) are classified correctly. From a plot of estimated probabilities (not shown), the probability of accurate prediction was found to be .75 or greater in 55 of the 152 cases (36%) classified as potential leavers. For the remaining 42 cases predicted as potential leavers, the probability is < .50 that they are in fact potential leavers (10 of the 42 cases [24%] had a predicted probability of < .25). In predicting potential stayers (i.e., tenure of 
3 years), 25 out of the 72 cases (35%) classified as potential stayers have a probability of .75 or greater for being correctly classified. Hence, it can be said that overall the model correctly predicts with a probability of .75 both potential leavers and stayers in 80 of the total 224 (
36%) cases. One of the reasons why the model does not have greater predictive power is extraneous variables that could not be included in this study but have a relationship to tenure.
|
|
, p. 5), average length of employment in previous positions is the strongest variable predicting the likelihood that an NHA will stay 3 years or longer (Table 6 ). In other words, a pattern of past job stability is likely to predict future stability. Community attachment and nature of separation (voluntary vs involuntary) are two other variables that are statistically significant at p 
.05.
|
, p. 17). Hence, we employed the likelihood ratio test comparing the full model with these two variables present to a reduced model with the two variables removed. The change in log likelihood was found to be statistically significant (2 = 7.665, df = 2, p = .02) and it also improved the predictability of the model providing evidence that performance outcomes and commitment should be retained in the model. The dichotomous variable representing formal education in nursing was also tried, but it was nonsignificant (p = .80). Other variables tried in the model (age, gender, marital status, years of experience, education, field of academic concentration, nature of move, number of NHA jobs previously held, and the remaining perceptual dimensions obtained by factor analysis) were also found to be nonsignificant. Also, specific inclusion of the nature of past separation (voluntary or involuntary) variable controls for any implicit assumptions that having an administrator stay for 3 years or longer is beneficial in all cases. Results of the model are specifically in reference to NHAs who separate voluntarily; that is, given that the administrator leaves a position voluntarily, tenure of 3 years is influenced by the predictors in the model. Table 7 gives a correlation matrix for the variables of interest.
|
|
Discussion |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
; Singh 1997b). During the first 3 years of employment, the probability that a newly hired administrator will separate, voluntarily or involuntarily, is 37% (70/187—Table 2 ). But, as discussed later, identifying potential leavers during the hiring process and use of retention strategies after hiring can improve overall administrative stability in nursing facilities. It should be pointed out that, in a few cases, turnover may actually be desirable. For instance, when an administrator is not performing, nursing home governing bodies may decide to initiate change. According to this study, management initiatives to remove an NHA occur 75% of the time during the first 3 years; only 25% of involuntary separations take place later. For involuntary separations occurring within the first 3 years, the NHA's average length of employment is 1.3 years. Hence, upper management's assessment of a poor fit between the administrator and the facility is acted upon relatively quickly in most cases. This is a positive indication because NHAs who do not perform competently are unlikely to help improve the operations or maintain facility performance at desired levels.
Most NHAs (81%) depart from their positions voluntarily. However, it is surprising to observe the high proportion of voluntary leavers who find promotional opportunities after an average length of employment of only 1.3 years in the previous position. There are no differences in the patterns of job stability between advancement seekers and those who depart voluntarily for other reasons. There are also no differences between the two groups on the various attitudinal dimensions. The only conclusion we can draw from this is that advancement seekers very likely get offers of promotional opportunities they find hard to resist. The extent of voluntary turnover in pursuit of promotions after relatively short tenures suggests a limited pool of adequately trained NHAs to select from in order to fill vacancies created by rapid turnover in the industry.
After controlling for voluntary and involuntary separations, logistic regression results show that NHAs who have evidenced a pattern of short tenures and frequent job changes in the past are also likely to be potential leavers in future positions. Hence, instability appears to be an administrative trait that can be easily assessed by reviewing the tenure history on a prospective administrator's resume and verifying this information through reference checks. Interestingly, this "tendency to turnover" as a personality characteristic has been observed in other studies (Casio 1992; Kettliz, Zbib, and Motwani 1998; Mowday and Lee 1987), but the strength of this variable in our logistic regression model was quite surprising. It was clearly the most significant independent variable predicting tenure.
A community environment that is compatible with personal lifestyle and overall satisfaction with the quality of life a community offers contributes to job stability. The correlation between performance outcomes and administrative tenure is not surprising. A relationship between performance outcomes and turnover has been reported in studies of general managers (Kerr and Slocum 1987; Porter and Steers 1973). Positive performance results, such as facility meeting its budget commitments, compliance with certification regulations and standards, and low turnover among department heads and licensed staff (see Table 1 ), contribute to the NHA's sense of accomplishment and job satisfaction. Performance is also closely associated with upper management's desire to retain a well-performing administrator. Hence, good facility performance fosters a win-win situation by minimizing both voluntary and involuntary separations. It should be noted that from the results of this study we cannot conclude that an administrator's past performance is an indicator of future performance in a different facility, although such a conclusion makes intuitive sense.
The NHA's commitment to the organization is also a significant variable. Commitment is indicated by a high degree of trust in the organization, enthusiasm and loyalty toward the company, willingness to make one's best contribution to the organization, and personal feelings of a good fit with the organization (see Table 1 ). Compared to the factors discussed earlier, commitment is less tangible and more difficult to assess. However, our results show that past patterns of instability also reflect lower levels of commitment. In addition, commitment is correlated with several other dimensions (see Table 7 ) and thus reflects these dimensions, notably realized expectations (r = .61) and career opportunities/rewards (r = .52). Realized expectations point to the leadership style of the NHA's supervisor, autonomy in decision making, degree of harmony between the NHA's and the organization's ethical/moral values and management philosophies, and reasonableness of corporate expectations (see Table 1 ). Assistance and support from upper management in dealing with problem areas in the facility can improve outcomes and also contribute to fulfillment of the administrator's own expectations, foster commitment, and eventually lead to greater retention. These factors underscore the responsibility of the nursing home corporation in creating a leadership environment that will help promote greater retention of its administrators. NHAs are more likely to leave an organization when their job expectations are not met and when opportunities for growth and advancement are not available.
Management Implications
Findings in this study have implications for screening and recruitment of administrators as well as for achieving higher retention once a new NHA has been appointed to fill a vacancy.
Policy Implications
The supply of well-trained nursing home NHAs may be inadequate to meet current demands. Licensure policies directly impact the supply of administrators, and have a strong influence on the level of training and skills NHAs possess in the relevant domains of practice (Singh 1997a). Current policies need to be reevaluated and refocused toward the objective of creating an adequate pool of well-trained administrators. Without appropriate policy interventions designed to rectify existing problems, the future outlook for any major improvements in administrator retention and quality of care in nursing facilities is likely to remain grim.
Limitations and Future Recommendations
Facility-related measures, such as independent versus chain ownership and nonprofit versus for-profit operation, are correlated with NHA turnover and retention (Singh and Schwab 1998), but this study did not obtain information on such variables for past positions. Perceptual measures obtained from rating scales sometimes have inherent shortcomings, as reflected in the relatively low reliability values obtained on two of the constructs (Performance Outcomes and Community Attachment) appearing in the logistic regression model. Even though these constructs are theoretically sound, their interpretation may be open to some question. For future studies, we recommend including additional relevant variables in the measurement of these dimensions. Finally, the NHA survey did not include administrators who left the nursing home industry altogether. If such career changers largely leave due to burnout or disenchantment with the NHA profession, their exclusion from the study slants the results in favor of those who either like to stay in the field or lack transferable skills needed for a career change.
For future studies of this kind, we recommend a longitudinal design in which attitudinal dimensions are measured at an initial stage, and then measured at yearly intervals to be correlated with turnover and retention. Above all, such a design will minimize recall bias. A longitudinal design will require scales that are more sensitive than the 4-point scales used in this study.
Conclusions
There is high instability among NHAs, especially during the first 3 years of employment. In the selection of new administrators, past patterns of job hopping with short tenures is a strong indicator that the administrator will be unlikely to stay for at least 3 years. Geographic compatibility in relocating a new administrator is also an important factor. However, the industry presents a dilemma because of an apparent shortage of skilled NHAs who, once recruited, quickly find advancement opportunities at other facilities. Hence, retention of well-performing administrators and ongoing training and development are just as critical as the initial screening and selection. In anticipation of loyalty and commitment from NHAs, upper management should simultaneously evaluate its own commitment to its well-performing administrators. Management styles that promote support, personal growth, and achievement of facility performance objectives can help improve retention. Finally, licensure policies need to be refocused to ensure that the industry has an adequate supply of qualified administrators.
|
Acknowledgments |
|---|
Received for publication December 9, 1998. Accepted for publication February 21, 2000.
|
References |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  P. M. Keith Nursing Home Administrators' Attribution of Antifacility Bias to Ombudsman Programs Journal of Applied Gerontology, April 1, 2006; 25(2): 120 - 136. [Abstract] [PDF]
|
|
|
|
 |
|
 N. G. Castle Measuring Staff Turnover in Nursing Homes Gerontologist, April 1, 2006; 46(2): 210 - 219. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N. G. Castle Turnover Begets Turnover Gerontologist, April 1, 2005; 45(2): 186 - 195. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N. G. Castle, T. J. Lowe, J. A. Lucas, J. P. Robinson, and S. Crystal Use of Resident Satisfaction Surveys in New Jersey Nursing Homes and Assisted Living Facilities Journal of Applied Gerontology, June 1, 2004; 23(2): 156 - 171. [Abstract] [PDF]
|
|
|
|
|
|
N. G. Castle Providing Outcomes Information to Nursing Homes: Can It Improve Quality of Care? Gerontologist, August 1, 2003; 43(4): 483 - 492. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N. G. Castle Administrator Turnover and Quality of Care in Nursing Homes Gerontologist, December 1, 2001; 41(6): 757 - 767. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
