explaining nursing turnover intent: job satisfaction, pay satisfaction, or organizational...

16
Explaining nursing turnover intent: job satisfaction, pay satisfaction, or organizational commitment? * LILLIE LUM 1 , JOHN KERVIN 2 KATHLEEN CLARK 3 , FRANK REID 2 AND WENDY SIROLA 4 1 Health Services Innovation Group, Toronto, Canada 2 Centre for Industrial Relations, University of Toronto, Canada 3 Quality of Nursing Worklife Unit, McMaster University, Canada 4 Hospital for Sick Children, Toronto, Canada Summary A number of models have been developed to explain nurses’ turnover behavior. The common theme that emerges from these models is that turnover behavior is a multistage process that includes attitudinal, decisional, and behavioral components. The purpose of this study was to assess both the direct and indirect impact of certain pay policies upon the turnover intentions of paediatric nurses. The two major questions addressed were: What was the relative impact of job satisfaction, pay satisfaction, and organizational commitment upon the turnover intentions of paediatric nurses eligible for these pay policies? What model accurately portrays the relationship among these three independ- ent variables and turnover intentions? Exploration of the causal pathways among these variables and demographic factors revealed complex models of association. The results suggest that job satisfaction has only an indirect influence on the intention to quit, whereas organizational commitment has the strongest and most direct impact. A further finding that pay satisfaction had both direct and indirect eects on turnover intent was consistent with administrators’ assumptions underlying the pay policies. Control vari- ables such as having a degree, having children, and working 12-hour shifts were found to have both direct and indirect influences upon pay satisfaction and turnover intent. # 1998 John Wiley & Sons, Ltd. J. Organiz. Behav. 19: 305–320 (1998) Introduction In the early 1990s, the shortage of experienced registered nurses in teaching hospitals prompted renewed interest in nursing retention. The issue was not a generalized shortage of nurses but rather an acute and chronic maldistribution of nurses in certain geographic regions and in particular areas of specialized nursing practice. Bed closures and cancellation of surgeries were directly related to the lack of sucient numbers of nurses qualified to care for acutely ill patients. One very important question which continues to puzzle both administrators and quality of * This project was funded by the Ontario Ministry of Health, Hospital Incentive Fund (Ontario, Canada). CCC 0894–3796/98/030305–16$17.50 Received 1 September 1994 # 1998 John Wiley & Sons, Ltd. Accepted 23 July 1996 JOURNAL OF ORGANIZATIONAL BEHAVIOR, VOL. 19, 305–320 (1998)

Upload: lillie-lum

Post on 06-Jun-2016

214 views

Category:

Documents


1 download

TRANSCRIPT

Explaining nursing turnover intent:job satisfaction, pay satisfaction,or organizational commitment?*

LILLIE LUM1, JOHN KERVIN2 KATHLEEN CLARK3, FRANK REID2

AND WENDY SIROLA4

1Health Services Innovation Group, Toronto, Canada2Centre for Industrial Relations, University of Toronto, Canada3Quality of Nursing Worklife Unit, McMaster University, Canada4Hospital for Sick Children, Toronto, Canada

Summary A number of models have been developed to explain nurses' turnover behavior. Thecommon theme that emerges from these models is that turnover behavior is a multistageprocess that includes attitudinal, decisional, and behavioral components. The purpose ofthis study was to assess both the direct and indirect impact of certain pay policies uponthe turnover intentions of paediatric nurses. The two major questions addressed were:What was the relative impact of job satisfaction, pay satisfaction, and organizationalcommitment upon the turnover intentions of paediatric nurses eligible for these paypolicies? What model accurately portrays the relationship among these three independ-ent variables and turnover intentions? Exploration of the causal pathways among thesevariables and demographic factors revealed complex models of association. The resultssuggest that job satisfaction has only an indirect in¯uence on the intention to quit,whereas organizational commitment has the strongest and most direct impact. A further®nding that pay satisfaction had both direct and indirect e�ects on turnover intent wasconsistent with administrators' assumptions underlying the pay policies. Control vari-ables such as having a degree, having children, and working 12-hour shifts were foundto have both direct and indirect in¯uences upon pay satisfaction and turnover intent.# 1998 John Wiley & Sons, Ltd.

J. Organiz. Behav. 19: 305±320 (1998)

Introduction

In the early 1990s, the shortage of experienced registered nurses in teaching hospitals promptedrenewed interest in nursing retention. The issue was not a generalized shortage of nurses butrather an acute and chronic maldistribution of nurses in certain geographic regions and inparticular areas of specialized nursing practice. Bed closures and cancellation of surgeries weredirectly related to the lack of su�cient numbers of nurses quali®ed to care for acutely ill patients.One very important question which continues to puzzle both administrators and quality of

* This project was funded by the Ontario Ministry of Health, Hospital Incentive Fund (Ontario, Canada).

CCC 0894±3796/98/030305±16$17.50 Received 1 September 1994# 1998 John Wiley & Sons, Ltd. Accepted 23 July 1996

JOURNAL OF ORGANIZATIONAL BEHAVIOR, VOL. 19, 305±320 (1998)

worklife researchers is whether certain human resource strategies, such as increased pay, have anye�ect upon nurses' attitudes regarding turnover.

Pay is viewed as part of the sanction system used by the organization to motivate compliancewith its rules and regulations (Mueller and Price, 1990). For the individual employee, pay isviewed as an important reward or outcome. In human resources management, Meltz andMarzetti (1988) proposed the use of salary di�erentials as economic incentives to enhancenursing job satisfaction and retention for `di�cult to sta�' areas. A negative relationship betweenpay levels and turnover has been frequently reported in the literature. Yet in spite of this evidence,little is known about a�ective and cognitive variables which mediate this relationship(Motowidlo, 1983). There is little doubt that pay in¯uences turnover but the underlyingdynamics are in need of further investigation.

The purpose of this paper is to present the results of the ®rst phase of a program evaluationstudy designed to assess the impact of pay policies upon the turnover intentions of paediatric sta�nurses at a metropolitan teaching hospital. The issue is: Are nurses more likely to stay within anorganization if given a salary supplement? Under the pay policy examined here, two types oftenure-based payments were made: bonuses o�ered to intensive care sta� nurses only, and a 5 percent salary di�erential above union scale for all sta� nurses. The major objective of thesesupplements was to reduce the annual rate of turnover, which was 31.9 per cent in the intensivecare units and 26.9 per cent in other clinical areas. (These ®gures include three categories: quits,layo�s, and other reasons such as retirement). By comparison, Picot and Baldwin (1990b) haveshown that for a large sample of employees in the national labor force the overall turnover ratewas about 21 per cent per year, roughly equally split between the three categories: quits (7.7 percent), layo�s (6.9 per cent), and other reasons (6.8 per cent).

The anticipated success of these pay policies was based upon two major assumptions. First, thesupplements were assumed to be su�ciently high in value and structured in such a way as to havethe most favorable in¯uence upon the senior sta� nurses who were presumed to be the mostexperienced, and thus most valuable and committed employees. A second assumption was thatthese pay policies would `neutralize' the impact of attractive job opportunities in other hospitals.Indeed, during this period there were high vacancy rates in a variety of paediatric settings.Although the relationship between job alternatives and turnover has not received much attention,Motowidlo (1983) has suggested that pay satisfaction predicts turnover decisions only whenemployees perceive that they are likely to improve their pay through changing employers.

A number of models have been developed to explain employee turnover behavior (for example,Price and Mueller, 1981; Parasuraman, 1989; Weisman, Alexander and Chase, 1980). Thecommon theme that emerges from these models is that such turnover behavior is a multistageprocess that includes attitudinal, decisional, and behavioral components. Three major classes ofdeterminants include individual factors, economic opportunity, and work-related factors. Thelast category includes factors such as job satisfaction and organizational commitment. The desireto remain in an organization was ®rst discussed in the context of models of voluntary turnoverand was seen largely as a function of job satisfaction (March and Simon, 1958). Commitment, onthe other hand, is seen as a more global attitude serving as a stabilizing force that acts to maintainbehavioral direction when expectancy/equity conditions are not met (DeCoutis and Summers,1987). In other words, job satisfaction is seen to be an intervening variable between environ-mental and personal characteristics and organizational commitment. In an alternate model,Bateman and Strasser (1984) propose that organizational commitment is rationalized bysubsequent job satisfaction attitudes. In the research reported here, both the direct and indirectin¯uence of job satisfaction and organizational commitment, as well as pay satisfaction, wereassessed in terms of their impact upon intended turnover.

# 1998 John Wiley & Sons, Ltd. J. Organiz. Behav. 19: 305±320 (1998)

306 L. LUM ET AL.

In this study, we used `turnover intent' rather than `turnover' as the dependent variable. Thisapproach draws upon a number of recent studies that have assessed the role of intentions inpredicting and understanding turnover. Satisfaction is conceptualized as an a�ective oremotional response, whereas intentions are statements about speci®c behaviors of interest.Fishbein (1967) and Newman (1974) have demonstrated that behavioral intentions to stay orleave are consistently related to turnover and that this relationship is generally stronger andaccounts for more variance in turnover than does the satisfaction±turnover relationship. Mobley,Horner and Hollingsworth (1978) reported moderate to strong correlations between intention toquit, job search and thinking of quitting, and turnover, among hospital employees. Mobley,Gri�eth, Hand and Meglino (1979) suggest that intentions o�ers a better explanation since itcaptures the individual's perception and evaluation of job alternatives.

The two major questions addressed in this study are: What was the relative impact of jobsatisfaction, pay satisfaction, and organizational commitment upon the turnover intentions ofpaediatric nurses eligible for these pay policies? What causal model accurately portrays therelationship among these three independent variables and turnover intentions for this sample ofsta� nurses?

Related Studies

In practical terms, turnover represents a major problem, particularly in terms of recruitmentcosts and organizational continuity. For nursing and health care organizations, additional con-cerns include the ability to care for patients and the quality of care given. As a consequence of itsimportance, turnover has been the subject of an extensive number of studies in many disciplines.The resulting models of turnover hypothesize numerous causes and consequences. Only thosestudies which provide the basic background of these models and are relevant to nursing turnoverwill be presented below.

Pay satisfaction and turnover intent

Numerous causes of pay satisfaction have been proposed. These include personal and job inputs,monetary and nonmonetary outcomes, the comparison process, and pay policies and administra-tion. Current models of pay satisfaction continue to have their basis in concepts of equity theory(Adams, 1965). Equity theory emphasizes that pay satisfaction is caused by feelings regarding theequity of one's pay. These feelings result from perceptual and comparative processes. The theoryholds that an employee formulates a ratio of outcomes (including pay) to inputs. This ratio isthen compared to the outcome/input ratio of some referent source. If the ratios correspond, payequity and satisfaction result. If an employee's pay is perceived to be less than another's, feelingsof being inequitably underpaid may ensue. Lawler (1971) o�ers a similar approach suggestingthat pay satisfaction or dissatisfaction is a function of the discrepancy between what one feels oneshould receive and how much pay one does receive. Pay dissatisfaction is su�ciently commonthat a ®nding of a lower rate of satisfaction with pay than with other job components can beanticipated, and even predicted (Lawler, 1990).

Reported consequences of pay satisfaction include a variety of unwanted employee behaviorssuch as turnover, absenteeism, willingness to strike, and lowered job performance (Heneman,

# 1998 John Wiley & Sons, Ltd. J. Organiz. Behav. 19: 305±320 (1998)

EXPLAINING NURSING TURNOVER INTENT 307

1985a). Weiner (1980) assessed the pay satisfaction of blue- and white-collar employees in apublic sector organization. Pay satisfaction was found to be signi®cantly predictive of bothabsenteeism and turnover, though the strength of the relationships was not particularly large.Motowidlo (1983) used pay satisfaction to predict turnover intentions and actual turnover amongsales representatives. He concluded that pay has an impact on turnover only through its e�ects onpay satisfaction which, in turn, impacts turnover only through its e�ects on turnover intentions.Newman (1974) reported that satisfaction with pay was found to be a marginally signi®cantpredictor variable for intended turnover of nursing home sta�.

In nursing, a number of studies have included pay satisfaction as a component of a multi-dimensional measure of job satisfaction (Frisina, Murray and Aird, 1988). Typically, nurses havenot rated pay as a high priority in their job satisfaction (Frisina et al., 1988). However, in recentyears nurses have expressed increasing concern about their salary and bene®ts. In 1988, Murrayand Smith explored the career, job, and hospital satisfaction of sta� nurses in metropolitanhospitals of a large Canadian city. They found only 10 per cent of the nurses were satis®ed withtheir pay and 46 per cent were dissatis®ed. The signi®cance of this ®nding was not that so many ofthe nurses were dissatis®ed with their pay, but that this was the single largest expression ofdissatisfaction of all the satisfaction rating scales in their study (Murray and Smith, 1988, p. 68).Bateman and Strasser (1984) have reported that nurses' satisfaction with work, coworkers, andsupervision were strong predictors of commitment but not of pay satisfaction. As far as nursesare concerned, there are no studies which report the concurrent e�ects of pay supplements uponpay satisfaction and turnover intent.

Job satisfaction and turnover intent

Many studies conceptualize turnover to be a psychological response and rest on the belief thatturnover is an individual choice behavior. At the individual level, satisfaction with one's job is themost frequently studied psychological variable in the satisfaction±turnover relationship (Mobley,1979). Studies involving professional and nonprofessional employees indicate a negative relation-ship between overall satisfaction and turnover although the variance accounted for is less than14 per cent. When satisfaction is included in multiple regressions with variables such as intentionsand commitment, its e�ect on turnover may become nonsigni®cant (Marsh and Mannari, 1977;Mobley et al., 1978).

Job dissatisfaction has been repeatedly identi®ed as the single most important reason whynurses leave their jobs. Price and Mueller (1981) were able to conclude empirically that jobdissatisfaction had an indirect e�ect on turnover through its direct e�ect on formation of intentto leave. Irvine and Evans (1992) reported signi®cant causal relationships between intent toturnover, and both job satisfaction (beta � ÿ0:53) and organizational commitment (beta �0:34). Both intrinsic factors such as autonomy, respect, and recognition and extrinsic factors suchas pay, shiftwork, and workload have been cited as variables related to job satisfaction. Cottonand Tuttle (1986) found that overall job satisfaction, satisfaction with work itself, pay satis-faction, satisfaction with supervision, and organizational commitment were negatively related toturnover. Weisman et al., (1980) cited understa�ng and in¯exible work schedules as importantcontributors to job dissatisfaction.

In short, the relationship between satisfaction and turnover has been consistently found inmany turnover studies. However, it usually accounts for less than 16 per cent of the variance inturnover (Porter and Steers, 1973). It is apparent that models of the employee turnover processmust move beyond satisfaction as the primary explanatory variable.

# 1998 John Wiley & Sons, Ltd. J. Organiz. Behav. 19: 305±320 (1998)

308 L. LUM ET AL.

Job satisfaction, organizational commitment,and turnover intent

Commitment to the employing organization has been a topic of considerable research for the pasttwo decades (Blegen, 1993). Williams and Hazer (1986) make the distinction between commit-ment and job satisfaction in that the former is an a�ective response to the whole organization,whereas the latter represents an a�ective response to speci®c aspects of the job. Interest instudying commitment has continued since it has been shown to be consistently related to(1) employee behaviors such as turnover, absenteeism, and performance (Angle and Perry, 1981;Bluedorn, 1982; Porter and Steers, 1973); (2) attitudinal, a�ective, and cognitive constructs suchas job satisfaction, job involvement, and job tension (Hall and Schneider, 1972; Hrebiniak andAlutto, 1972; Porter, Steers, Mowday and Boulian, 1974); (3) characteristics of the employee'sjob and role, and task identity (Steers, 1977); and (4) personal characteristics of the employeesuch as age, gender, need for achievement, and job tenure (Angle and Perry, 1981; Hebriniak andAlutto, 1972; Steers, 1977). Many studies have reported a signi®cant association betweenorganizational commitment and turnover intentions (Ferris and Aranya, 1983; O'Reilly andCaldwell, 1980; Stumpf and Hartman, 1984; Weiner and Vardi, 1980). Satisfaction and commit-ment have invariably been reported to be negatively related to turnover and intent to leave(Arnold and Feldman, 1982; Bluedorn, 1982; Hollenbeck and Williams, 1986), and positivelycorrelated with one another (Bluedorn, 1982).

The major focus of these models and research ®ndings has been to identify antecedents ofcommitment from a variety of categories. These have included personal characteristics, workexperiences, job characteristics, organizational factors, and role-related factors. The conceptualturnover model that has received the most attention in the psychological literature was proposedby Mobley (1977). Although this model considers organizational commitment as an attitude tobe related to satisfaction, no clear causal relation is hypothesized. Studies have either failed toinclude both satisfaction and commitment or overlooked the causal relationship between thetwo variables. Williams and Hazer (1986), through the use of structural equation methodology,have concluded that commitment has a more important e�ect on intent to leave than doessatisfaction. Personal and organizational characteristics were found to have no direct impactupon turnover intentions. However, they in¯uenced satisfaction directly, and in¯uencedcommitment indirectly through their impact upon satisfaction and its subsequent e�ect oncommitment.

Bateman and Strasser (1984) have argued that many of the earlier studies have limitedinterpretability due to the tendency to use `static correlational analysis' of the relationshipbetween commitment and its presumed antecedents. The results of their longitudinal study ofnursing sta� showed that personal characteristics played a minor role in explaining variation incommitment. They also found that overall satisfaction is not a cause of commitment but rather aresult of it. They have suggested that commitment may be a construct that is neither simultaneouswith nor a consequence of job satisfaction. Sheridan and Abelson (1983) and Sheridan (1985)similarly have suggested that the relationship between variables such as organizational commit-ment and turnover is nonlinear, that is, termination represents a discontinuous change inwithdrawal behavior, occurring only after certain levels of tension and commitment are exceeded.Shore and Martin (1989) and Mueller and Price (1990) concluded that although both satisfactionand commitment are related to turnover, organizational commitment is more strongly related toturnover intentions. Yet important discrepancies exist concerning the relative contribution of jobsatisfaction and organizational commitment to the withdrawal process, suggesting the need forfurther study.

# 1998 John Wiley & Sons, Ltd. J. Organiz. Behav. 19: 305±320 (1998)

EXPLAINING NURSING TURNOVER INTENT 309

Models and hypotheses

Taken together, the ®ndings cited above suggest two di�erent models of the relationshipsinvolving work and pay satisfaction, organizational commitment, and turnover intent, as shownin Figure 1. The di�erence between the two models lies in the role of organizational commitment.Some researchers see it as a component of job satisfaction (model A), while others see it as havingan independent e�ect of its own on turnover intent (model B).

In this paper we hypothesize a slightly di�erent relationship (Figure 2). In our model, alljob satisfaction factors are mediated by organizational commitment, except for pay satisfactionwhich we postulate to have an additional direct e�ect on turnover intent. In short, wehypothesize:

H1: Job satisfaction e�ects on turnover intent are mediated by organizational commitment.H2: Pay satisfaction a�ects directly both job satisfaction and turnover intent.

The basis for the hypothesized mediating role of organizational commitment follows from thework of DeCoutis and Summers (1987), Shore and Martin (1989), and Mueller and Price (1990),who found the e�ect of job satisfaction less than that of organizational commitment. Our

Figure 1. Models derived from the literature

# 1998 John Wiley & Sons, Ltd. J. Organiz. Behav. 19: 305±320 (1998)

310 L. LUM ET AL.

expectation of a direct e�ect of pay satisfaction on turnover intent stems from our belief thateconomic factors have a separate impact on behavioral intentions, and is consistent with whatMueller and Price (1990) have proposed. Indeed, for the pay policy initiatives described above,hospital administrators implicitly assumed that pay bonuses would directly impact turnoverintentions, independently of any e�ect on job satisfaction.

Methodology

Sample

The target population included all full- and part-time registered sta� nurses at the selectedhospital who were eligible for one or both of the two pay policies (a 5 per cent salary di�erentialabove union scale for all sta� nurses, and an additional bonus for intensive care sta�). Anonproportional strati®ed sample was used with human resources sta� ®les as the samplingframe. All nurses in both the neonatal and paediatric intensive care units (NICU and PICU) weresurveyed. For the 32 remaining clinical units, one in four sta� nurses were randomly selected. Thetarget sample consisted of 466 sta� nurses; 222 from the general care areas and 244 from thecritical care units. Data were collected over a 4-month period. Each respondent received apersonalized envelope containing a 12-page questionnaire and a self-addressed, stamped returnenvelope. Two follow-up mailings produced a usable return rate of 77.5 per cent (N � 361).

Instrumentation

Scales to measure the three major independent variables (job satisfaction, pay satisfaction,and organizational commitment) were developed or adapted from established existing scales.A 5-point Likert scale was used for scoring each item.

Job satisfaction scaleA job satisfaction scale (JS) consisting of 25 items was derived from the original scale (37 items,Cronbach's alpha � 0:86) developed by Stamps, Piedmont, Slavitt and Haase (1978) for thepurpose of measuring job satisfaction among nurses. The modi®ed JS included questions aboutwork-related issues such as autonomy, task requirements, organizational policies, professionalinteractions, and perceived job status. (We were concerned with creating a short instrument tomaximize response rates given nurses' time pressures, but we retained many items in this scalebecause of its multidimensional nature).

Figure 2. Hypothesized model

# 1998 John Wiley & Sons, Ltd. J. Organiz. Behav. 19: 305±320 (1998)

EXPLAINING NURSING TURNOVER INTENT 311

Pay satisfactionPay satisfaction (PS) was measured by a 4-item scale developed for the purposes of this study,roughly modeled on the pay satisfaction items in the Minnesota Satisfaction Questionnaire. Ourscale contained general items seeking attitudinal information about pay level, adapted to thespeci®cs of our survey situation (i.e. speci®c references to nurses and the hospital). Additionalitems (not analyzed here) were developed to assess respondents' attitudes towards each of the paypolicies.

Organizational commitment scaleA 5-item organizational commitment scale (OC) was adapted from the original 15-iteminstrument originally developed by Mowday, Steers and Porter (1979), and primarily used inindustrial settings. They reported a reliability value (Cronbach's alpha) of 0.88 with a modi®ed9-item scale tested with hospital-based workers. The items selected focused upon the respond-ent's attitudes about the hospital as a desirable place in which to work (omitting those tappingintent to stay (Angle and Perry, 1981), which would have confounded this measure with themeasure of turnover intent).

Intentions toward turnoverIntentions toward turnover (TI), that is, leaving the hospital, were measured by a 3-item indexwhich sought information about respondents' intention to look for another job. The measure isan index in that each item represents a conceptually di�erent dimension of turnover. The items,and their dimensions, are as follows: (1) `In the last few months have you ever thought seriouslyabout looking for a nursing job at another hospital?' (consideration of organizational but notoccupational exit), (2) `In the last few months have you ever thought seriously about looking for anon-nursing job?' (consideration of occupational exit), and (3) `Taking everything into consider-ation, how likely is it that you will make a serious e�ort to ®nd a new job within the next year?'(consideration of other factors, such as the job market, potentially a�ecting exit).

As well as scales for the four principal variables, the survey questionnaire also incorporatedmeasures of a number of demographic and work-related control variables. The former includedage, gender, education, income, number of pre-school children living at home, other dependents,and marital status. The latter included shift pattern most frequently worked, years of experience,years of tenure on unit and at the hospital, type of unit (intensive care or other), and employmentstatus.

Results

Sample characteristics

The majority of the sample were female (96.9 per cent) with a mean age of 32.4 years (range 23±63 years). Slightly more than half (53.8 per cent) reported being married or cohabiting. A largeproportion of the sample (71.2 per cent) did not have children; of those with children (N � 91)the mean number was 1.7. Of the respondents, 14 per cent reported having one or more adultdependents. The most commonly reported range of annual nursing income was $30,000±$39,000.A small proportion of the sample (21 per cent) possessed a baccalaureate degree. No attempt was

# 1998 John Wiley & Sons, Ltd. J. Organiz. Behav. 19: 305±320 (1998)

312 L. LUM ET AL.

made in this phase of the study to ascertain the type of degree obtained. The remainder (79 percent) had a diploma from a hospital or college nursing program.

The majority of the sample (76.4 per cent) worked full-time, 19.7 per cent worked half-time ormore but not full-time, and the remaining 3.9 per cent worked less than half time. Most respon-dents reported self-scheduling although the sta� were generally required to work predominantly12-hour shifts. The mean years of paediatric experience was 7.8, with an actual range of 3 monthsto 35 years. The mode was 2.5 years experience. The mean years of tenure at this hospital was 6.6(range 51 to 40) years, although this may not represent one continuous period of work. Theaverage length of time spent working on the respondent's current clinical unit was 4.9 years witha range of 2 months to 25 years.

Psychometric properties of the scales(JS, PS, and OC)

The psychometric properties are reported in Table 1. Each scale item was scored from 1 to 5,higher values indicating greater satisfaction or commitment. For the turnover intent index, themean and standard deviation (N � 358) were 2.10 and 1.01 respectively; correlations among thethree index items ranged from 0.33 to 0.50.1

Table 1. Psychometric properties of the scales (JS, PS, and OC)

Number of items N Alpha Mean S.D.

PS 4 354 0.69 2.98 0.72JS 9 323 0.75 3.61 0.56OC 5 331 0.83 3.65 0.62

JS, job satisfaction scale; PS, pay satisfaction scale; OC, organizational commitment scale.

Relationships among turnover intentand JS, PS, and OC

Our analysis involves two steps. The ®rst requires developing a set of structural equationsincorporating the four principal variables to produce and compare path-analytic modelsrepresenting the three models (A, B, and C) presented above. In the second step, demographicand work-related variables are added to see whether they might substantially alter the ®ndings.(No prior hypotheses were generated involving these variables).

Intercorrelations, means, and standard deviations for the principal variables are shown inTable 2. (The means and standard deviations di�er slightly from those in Table 1 because oflistwise deletion of cases with missing values). These correlations are all moderate in size. Theirrelative values suggest that organizational commitment may be the most important determinantof turnover intent, as hypothesized above.

The 3-item index of turnover intent produced a non-normal distribution with a relatively highconcentration of cases at the lower values of the distribution (scores of 27 per cent of therespondents fell at the lowest value of the index, re¯ecting de®nite lack of interest in exit on all

1 The index item measuring occupational exit did not behave di�erently from the item measuring organizational exit.Their correlations with organizational commitment were ÿ0:28 and ÿ0:29 respectively.

# 1998 John Wiley & Sons, Ltd. J. Organiz. Behav. 19: 305±320 (1998)

EXPLAINING NURSING TURNOVER INTENT 313

three items). To correct for this non-normality, the variable was adjusted using `PRELIS' soft-ware (Joreskog and Sorbom, 1988), which produced a covariance matrix based on latent values ofturnover intent. The e�ect of this adjustment was to reduce slightly the correlations of turnoverintent with the other variables in the model (with pay satisfaction: from ÿ0:264 to ÿ0:261; jobsatisfaction: from ÿ0:292 to ÿ0:286; organizational commitment: from ÿ0:365 to ÿ0:349). Theadjusted version was used for deriving coe�cients and testing goodness-of-®t for all three models.

The results of the ®rst step, comparing the three models, are presented in Figure 3, and therelative goodness-of-®t of the three models is given in Table 3. The path coe�cients representstandardized maximum likelihood solutions to the sets of structural equations. Table 3 con®rmsthat model `C', generated from the above hypotheses, provides a superior ®t to the data com-pared to the two models (`A' and `B') derived from the literature. The values of chi-square and theroot mean square residual are both substantially smaller for model `C', while the adjustedgoodness-of-®t index is noticeably greater.

Figure 4 shows the same model with demographic and work-related characteristic variablesadded. Some of the original factors showed little or no impact on the principal variables, andwere dropped from the analysis (gender, income, marital status, type of unit, employment status).Because of the high intercorrelations among four of the variables (age, years of paediatricexperience, years on unit, and years at the hospital) we selected years of experience as a proxy forall four. Also, working predominantly 12-hour shifts and working predominantly night shiftswere correlated highly enough to prevent the iterations from converging, so the latter wasdropped from the analysis. Education was measured by whether the respondent possessed abachelor's degree. The number of pre-school children and number of other dependents were bothcollapsed to dummy variables (one or more � 1) because of excessive skewness. The result wasthe ®ve control variables shown in Figure 4.

Using the inferential `theory trimming' technique (beginning with a full model and removingweak paths) we investigated the relationships among control variables and the four principalfactors. As a consequence, the resulting coe�cients can only be regarded as descriptive of thenursing population in the hospital studied and, unlike the model discussed above, are technicallynot generalizable to other hospitals.

As the ®gure shows, the basic model of the four variables is highly stable. The path betweenpay satisfaction and turnover intent increases slightly in absolute magnitude (from ÿ0:18 toÿ0:21), likely as a result of the removal of some weak suppressor e�ect.

Table 2. Zero-order correlations, means, and standard deviations (N � 290)

PS JS OC TI Means S.D.

PS Ð Ð Ð Ð 2.99 0.71JS 0.39 Ð Ð Ð 3.61 0.56OC 0.25 0.40 Ð Ð 3.66 0.61TI ÿ0:26 ÿ0:29 ÿ0:37 Ð 2.09 1.00

JS, job satisfaction scale; PS, pay satisfaction scale; OC, organizational commitment scale; TI, turnover intent.

Table 3. Comparison of models A, B, and C

A B C

Chi-square (df ) 41.6 (5) 27.4 (5) 7.6 (3)Adjusted goodness-of-®t index 0.68 0.78 0.93Root mean square residual 0.054 0.069 0.026

# 1998 John Wiley & Sons, Ltd. J. Organiz. Behav. 19: 305±320 (1998)

314 L. LUM ET AL.

With respect to the control variables, an interesting pattern emerges. These factors exert directe�ects on either or both of pay satisfaction and turnover intent, but not on the interveningvariables of job satisfaction and organizational commitment. Further, all three of the factors thata�ect both pay satisfaction and turnover intent show an inconsistent relationship (suppressore�ect). Although pay satisfaction is negatively related to turnover intent, having a bachelor'sdegree increases both. On the other hand, having young children at home or working predomi-nantly 12-hour shifts decreases both pay satisfaction and turnover intent.

Discussion and Implications

The ®ndings of this study contribute to current knowledge of the antecedents of turnover intentof nurses, particularly the role of pay. Pay, as well as supply and demand issues in¯uenced by thelabor market and the employee's perception of the economy, have been previously identi®edas determinants of job satisfaction, organizational commitment, and behavioral intentions to

Figure 3. Standardized path coe�cients for the three models

# 1998 John Wiley & Sons, Ltd. J. Organiz. Behav. 19: 305±320 (1998)

EXPLAINING NURSING TURNOVER INTENT 315

turnover. While pay is considered to be an important aspect of job satisfaction, the relationshipbetween pay attitudes and turnover intent has not been clear. By deriving and testing causalinferences regarding pay satisfaction, job satisfaction, and organizational commitment, thisresearch has somewhat clari®ed the combined e�ects of these variables on turnover intent. Theresults reveal a model of association which is more complex than assumed by previousresearchers, and by the initiators of the pay policies at the study hospital.

More speci®cally, two ®ndings deserve mention. First, these results provide additional evidencethat job satisfaction does not appear to a�ect turnover intentions directly. The path model shownin Figure 2 is consistent with the assumed causal process of satisfaction a�ecting commitment,which in turn a�ects turnover intent. Those models, found in the literature, that ignore themediating role of organizational commitment fail to ®t the data as well as the model proposed andtested here.

A second important ®nding of this study is that pay satisfaction has both a direct and indirecte�ect upon nurses' turnover intent. Although pay satisfaction was signi®cantly associated withreducing intended turnover, its indirect e�ect upon turnover intent, mediated through jobsatisfaction and organizational commitment, was weaker. One implication of the two-path resultis that organizational policies designed to reduce turnover by increasing pay satisfaction need notattempt to increase job satisfaction and organizational commitment at the same time.

For the purposes of this study, pay satisfaction was treated as a unidimensional construct.However, in a complex setting such as a hospital, more complex reactions to pay changes arelikely to occur. Heneman and Swab (1979) have indicated that the concept `pay' can be dividedinto four distinct categories: level, structure, system, and form. In light of this, we recommendfuture studies include multidimensional measures of reactions towards the pay system. From apractical standpoint, putting pay satisfaction in multidimensional terms means that organiza-tions should recognize that a change in pay practices in one of the pay dimensions may have few,if any, e�ects in the others (Heneman, 1985b). In particular, the anecdotal data gathered as partof this study (respondents' comments in an open-ended survey question) indicated that many ofthe nurses felt less than satis®ed with the method with which the supplements were derived, thatis, intensive care nurses being rewarded more highly than the general sta� nurses2. Price and

2 For example, `It is unfair that the intensive care nurses get pay bonuses while we (general sta� nurses) only get salarydi�erentials', and, `The salary di�erentials were o�ered just to buy us (general sta� nurses) o� so we don't complainbecause the ICU nurses are the ones they want to keep'.

Figure 4. Standardized path coe�cients for model C with additional variables

# 1998 John Wiley & Sons, Ltd. J. Organiz. Behav. 19: 305±320 (1998)

316 L. LUM ET AL.

Mueller (1990) and Blegen (1993) have proposed that it is not only the actual salary o�ered to thenurses which a�ects their attitudes, but also their perceptions of whether they are being paidfairly for their work. In other words, a sense of distributive justice may be an important moder-ating variable to consider in future studies.

The satisfaction-to-commitment mediation process found in this study is consistent withPorter et al.'s basic model (1974) which proposes that commitment takes longer to develop and ismore stable than satisfaction. This model suggests that job satisfaction has only an indirectin¯uence on the intention to quit and highlights the need to study the mechanisms through whichsatis®ed nurses become committed to their organization.

As an attitude, organizational commitment di�ers from the concept of job satisfaction inseveral ways. Commitment can be viewed as being more global, thus re¯ecting a general a�ectiveresponse to the organization. Job satisfaction emphasizes the speci®c task environment in whichan employee performs his or her duties. Although day-to-day events in the workplace may a�ectan employee's level of work satisfaction, transitory events should not cause an employee to re-evaluate his or her attachment to the overall organization or clinical specialty.

Any retention strategy that is expected to have an e�ect upon turnover intentions mustsupersede the daily variations in routines and policies. For example, additional attention needs tobe directed towards the work environment. DeCoutis and Summers (1987) concluded thatindividuals do not become committed to an organization by virtue of some unique con®gurationof personal characteristics. Rather, individuals enter an organization with certain needs, desires,and skills, and expect to ®nd a work environment in which they can utilize their abilities andsatisfy many of their basic needs. Organizations which meet these requirements are characterizedby participation in the decision-making process; clear communications about organizationalintentions, activities and performance; autonomy with respect to work de®nition and conduct;and a sense of cohesion among organizational members.

Other factors also showed noteworthy e�ects. In particular, nurses with greater experiencewere more satis®ed with their pay and less likely to leave, which was the anticipated e�ect of thesalary supplements. The reason may lie in the fact that more experienced nurses received sub-stantially more money than junior nurses. The anecdotal data showed consistently that theyperceived the pay supplements to be an important recognition of their contribution to theorganization. Federico, Federico and Lundquist (1976) found that higher salary was associatedwith longer tenure whereas the di�erence between expected and actual salary was associated withshorter tenure. DeCoutis and Summers (1987) found that of the managerial employees theystudied, voluntary turnover occurred within the ®rst 8 months. Motowidlo (1983) has alsosuggested increasing rewards to employees during high periods of turnover lessens the attractive-ness of job alternatives at comparable organizations. The implication which emerges from thesedata is that retention of senior nursing sta� necessitates recognition of their skills throughe�ective pay strategies. As for the other control variables, having a degree both increases pay (thepositive relationship with pay satisfaction) and makes one more marketable (the positiverelationship with turnover intent). Young children at home increases demands for food, clothing,and shelter (the negative relationship with pay satisfaction) and makes the insecurity of jobchange more problematic (the negative relationship with turnover intent). Working 12-hour shiftsinvolves extended e�ort that may create perceived inequities that reduce pay satisfaction; it is notclear why these shifts would reduce turnover intent3.

3 We are indebted to an anonymous reviewer for suggesting some of the explanations for the ®ndings involving controlvariables.

# 1998 John Wiley & Sons, Ltd. J. Organiz. Behav. 19: 305±320 (1998)

EXPLAINING NURSING TURNOVER INTENT 317

A potential limitation of this study may be the lack of distinction between commitment to theorganization and commitment to a paediatric specialty. This di�erence might explain the presenceof individuals with high commitment scores who nevertheless have thoughts about turnovershould the opportunity arise. Despite the fact that the satisfaction scores were generally high, theresults showed that 39 per cent of the general care nursing sta� and 28 per cent of the critical carenurses were considering looking for a new job in the near future. Several other possible explana-tions exist for these ®ndings. Measures of overall job satisfaction by virtue of their greatervagueness may evoke a response bias known as the halo e�ect (Irvine and Evans, 1992). Murrayand Smith (1988) have suggested that career satisfaction might be a more appropriate measurethan job satisfaction. Perhaps by virtue of their professional socialization, nurse turnoverdecisions are not as strongly in¯uenced by conditions of the job as are nonprofessional employees'decisions. Several human resource management implications emerge. Based upon the multiplefacets of pay satisfaction proposed by Heneman (1985b), the use of pay supplements to in¯uenceturnover intent might have a more signi®cant and longer impact if they included additionalfeatures such as performance merit or recognition of outstanding contributions to the professionof nursing and not simply tenure.

To what extent might these ®ndings apply to other occupations? While the model testedabove has wide theoretical applicability, there are speci®cs of our testing situation that may limitgeneralizability. Nursing is a predominantly female profession. Further, these nurses wereinvolved in pediatric care, which suggests the possibility of higher-than-usual levels of nurturing.Also, many worked in the highly technical environment of intensive care. Testing the model withother populations will help to answer the question of generalizability.

In conclusion, the ®ndings of this study show that both personal and organizational factorshave an impact upon nurses' turnover intentions. Both simple and multivariate analyses indicatethat an association exists between turnover intent, as measured by job seeking behavior, and thevariables job satisfaction, pay satisfaction, and organizational commitment. This supports thebasic assumption underlying this study, that, when nurses are satis®ed with their jobs and pay,and feel committed to the organization, they are less likely to terminate employment voluntarily.Further, pay attitudes, unlike job satisfaction generally, a�ect turnover intent directly.

References

Adams, J. S. (1965). `Injustice in social exchange'. In: Berkowitz, L. (Ed.) Advances in Experimental SocialPsychology, Academic Press, New York, 267±299.

Angle H. L. and Perry, J. L. (1981). `An empirical assessment of organizational commitment andorganizational e�ectiveness', Administrative Science Quarterly, 26, 1±13.

Arnold, H. J. and Feldman, D. C. (1982). `A multivariate analysis of the determinants of job turnover',Journal of Applied Psychology, 67, 350±360.

Bateman, T. S. and Strasser, S. (1984). `A longitudinal analysis of the antecedents of organizationalcommitment'. Academy of Management Journal, 27(1), 95±112.

Becker, T. E. and Billings, R. S. (1993). `Pro®les of commitment: An empirical test', Journal ofOrganizational Behavior, 14, 177±190.

Blegen, M. A. (1993). `Nurses' job satisfaction: A meta-analysis of related variables', Nursing Research,42(1), 36±41.

Bluedorn, A. C. (1982). `A uni®ed model of turnover from organizations', Human Relations, 35, 135±153.Cavanaugh, S. J. (1989). `Nursing turnover: Literature review and methodological critique', Journal ofAdvanced Nursing, 14(1), 587±596.

# 1998 John Wiley & Sons, Ltd. J. Organiz. Behav. 19: 305±320 (1998)

318 L. LUM ET AL.

Cotton, J. and Tuttle, J. (1986). `Employee turnover: A meta-analysis and review with implications forresearch', Academy of Management Review, 11(1), 55±70.

Dear, M. R., Weisman, C. S., Alexander, C. S. and Chase, G. A. (1982). `The e�ect of intensive carenursing role on job satisfaction and turnover', Heart & Lung, 11(6), 560±565.

DeCoutis, T. A. and Summers, T. P. (1987). `A path analysis of a model of the antecedents andconsequences of organizational commitment', Human Relations, 40(7), 445±470.

Federico, J. M., Federico, P. and Lundquist, G. W. (1976). `Predicting women's turnover as a function ofextent of met salary expectations and biodemographic data', Personnel Psychology, 29, 559±566.

Ferris, K. and Aranya, N. (1983). `A comparison of two organizational commitment scales', PersonnelPsychology, 36, 87±98.

Fishbein, M. (1967). `Attitude and the prediction of behaviour'. In: Fishbein, M. (Ed.) Readings in AttitudeTheory and Measurement, Wiley, New York.

Frisina, A., Murray, M. and Aird, C. (1988). What do Nurses Want? A Review of Satisfaction and JobTurnover Literature, The Nursing Manpower Task Force of the Hospital Council of MetropolitanToronto Report, Toronto, Ontario.

Hall, D. T. and Schneider, B. (1972). `Correlates of organizational identi®cation as a function of careerpattern and organizational type', Administrative Science Quarterly, 17, 340±350.

Heise, D. R. (1970). `Problems in path analysis and causal inference'; In: Borgatta, B. F. (Ed.) SociologicalMethodology, Jossey-Bass, San Francisco.

Heneman, H. G. (1979). `Work and rewards theory'. In: Yoder, D. and Heneman, Jr., H. G. (Eds),ASPA Handbook of Personnel and Industrial Relations, Bureau of National A�airs, Washington.

Heneman, H. G. (1985a). `Pay satisfaction', In: Rowland, K. M. and Ferris, G. R. (Eds)Research in Personnel and Human Resources Management, Vol. 3, JAI Press, Greenwich, CT,pp. 115±139.

Heneman, H. G. (1985b). `Pay satisfaction', Research in Personnel and Human Resources Management, 3,115±139.

Hrebiniak, L. G. and Alutto, J. G. (1972). `Personal and role-related factors in the development oforganizational commitment', Administrative Science Quarterly, 17, 555±573.

Hollenbeck, J. R. and Williams, C. R. (1986). `Turnover functionality versus turnover frequency: A note onwork attitudes and organizational e�ectiveness', Journal of Applied Psychology, 71, 606±611.

Irvine, D. and Evans, M. (1992). Job Satisfaction and Turnover Among Nurses: A Review and Meta-Analysis, Quality of Nursing Worklife Research Unit Monograph, 10, Faculty of Nursing, University ofToronto, Toronto.

Joreskog, K. G. and Sorbom, D. (1988). PRELIS: A Preprocessor for LISREL, 2nd edn, Scienti®cSoftware, Inc., Mooresville, IN.

Lawler, E. E. (1971). Pay and Organizational E�ectiveness, McGraw-Hill, New York.Lawler, E. E. (1990). Strategic Pay: Aligning Organizational Strategies and Pay Systems, Jossey Bass,San Francisco.

March, J. G. and Simon, H. A. (1958). Organizations, Wiley, New York.Marsh, R. M. and Mannari, H. (1977). `Organizational commitment and turnover: A prediction study',Administrative Science Quarterly, 221, 57±75.

McCloskey, J. and McCain, B. (1987). `Satisfaction, commitment and professionalism of newly employednurses', Image, 19(1), 20±24.

Meltz, N. M. and Marzetti, J. (1988). The Shortage of Registered Nurses: An Analysis in a Labour MarketContext, Registered Nurses Association of Ontario, Toronto, Ontario.

Mobley, W. H., Horner, S. O. and Hollingsworth, A. T. (1978). `An evaluation of precursors of hospitalemployee turnover', Journal of Applied Psychology, 63, 408±414.

Mobley, W. H., Gri�eth, R. W., Hand, H. H. and Meglino, B. M. (1979). `Review and conceptual analysisof the employee turnover process', Psychological Bulletin, 36(3), 493±521.

Motowildo, S. J. (1983). `Predicting sales turnover from pay satisfaction and expectation', Journal ofApplied Psychology, 68, 484±489.

Mowday, R., Steers, R. and Porter, L. (1979). `The measurement of organizational commitment', Journalof Vocational Behaviour, 14, 224±247.

Mowday, R., Steers, R. and Porter, L. (1982). Employee±organizational Linkages: The Psychology ofCommitment, Absenteeism and Turnover, Academic Press, New York.

Mueller, C. W. and Price, J. L. (1990). `Economic, psychological and sociological determinants ofvoluntary turnover'. The Journal of Behavioural Economics, 19(3), 321±335.

# 1998 John Wiley & Sons, Ltd. J. Organiz. Behav. 19: 305±320 (1998)

EXPLAINING NURSING TURNOVER INTENT 319

Murray, M. and Smith, S. (1988). Nursing Morale in Toronto: An Analysis of Career, Job and HospitalSatisfaction Among Hospital Sta� Nurses, The Nursing Manpower Task Force of the Hospital Councilof Metropolitan Toronto, Toronto.

Newman, J. E. (1974). `Predicting absenteeism and turnover: A ®eld comparison of Fishbein's model andtraditional job attitude measures', Journal of Applied Psychology, 59(5), 610±615.

O'Reilly, C. A. and Caldwell, D. F. (1980). `Job choice: the impact of intrinsic and extrinsic factors onsubsequent satisfaction and commitment', Journal of Applied Psychology, 65, 559±565.

Parasuraman, S. (1989). ``Nursing turnover: An integrated model', Research in Nursing and Health, 12(1),267±277.

Picot, G. and Baldwin, J. (1990a). `Patterns of quits and layo�s in the Canadian economy, Part I', CanadianEconomic Observer, October, 5.1±5.21.

Picot, G. and Baldwin, J. (1990b). `Patterns of quits and layo�s in the Canadian economy, Part II',Canadian Economic Observer, December, 4.1±4.28.

Porter, L. W. and Steers, R. M. (1973). `Organizational work, and personal factors in employee turnoverand absenteeism', Psychological Bulletin, 80, 151±176.

Porter, L. W., Steers, R. M., Mowday, R. T. and Boulian, P. V. (1974). `Organizational commitment,job satisfaction, and turnover among psychiatric technicians', Journal of Applied Psychology, 59,603±609.

Price, J. and Mueller, C. (1981). `A causal model of turnover for nurses', Academy of Management Journal,24(3), 543±565.

Sheridan, J. E. (1985). `A catastrophe model of employee withdrawal leading to low job performance,high absenteeism, and job turnover during the ®rst year of employment', Academy of ManagementJournal, 28, 88±109.

Sheridan, J. E. and Abelson, M. A. (1983). `Cusp catastrophe model of employee turnover', Academy ofManagement Journal, 26, 418±436.

Shore, L. and Martin, H. (1989). `Job satisfaction and organizational commitment in relation to workperformance and turnover intentions', Human Relations, 42(7), 625±638.

Stamps, P., Piedmont, E., Slavitt, D. and Haase, A. (1978). `Measurement of work satisfaction amonghealth professionals', Medical Care, 16(4), 337±352.

Stumpf, S. A. and Hartman, K. (1984). `Individual exploration to organizational commitment orwithdrawal', Academy of Management Journal, 27, 308±329.

Weiner, N. (1980). `Determinants and behavioral consequences of pay satisfaction: A comparison of twomodels', Personnel Psychology, 33, 741±757.

Weiner, Y. and Vardi, Y. (1980). `Relationship between job, organization and career commitment to workoutcomes: An integrative approach', Organizational Behaviour and Human Performance, 26, 81±96.

Weisman, C. S., Alexander, C. S. and Chase, G. A. (1980). `Job satisfaction among hospital nurses: alongitudinal study', Health Services Review, 15, 341±364.

Weisman, C. S., Dear, M. R., Alexander, C. S. and Chase, G. A. (1981). `Employment patternsamong newly hired hospital sta� nurses: Comparison of nursing graduates and experienced nurses',Nursing Research, 30(3), 188±191.

Wetzel, K., Berg, S. and Gallagher, D. (1989). `Nursing education and organizational commitment:Degree versus diploma programs', Canadian Journal of Nursing Administration, Nov/Dec., 9±14.

Williams, L. J. and Hazer, J. T. (1986). `Antecedents and consequences of satisfaction and commitment inturnover models: A reanalysis using latent variable structural equation methods', Journal of AppliedPsychology, 71(2), 219±231.

# 1998 John Wiley & Sons, Ltd. J. Organiz. Behav. 19: 305±320 (1998)

320 L. LUM ET AL.