further exploring the relationship between job search and voluntary individual turnover

18
PERSONNEL ESYCHOWCil 1993.46 FURTHER EXPLORING THE RELATIONSHIP BETWEEN JOB SEARCH AND VOLUNTARY INDIVIDUAL TURNOVER GARY BLAU Human Resource Administration Department Temple University This study tested the usefulness of a new job search behavior measure to account for voluntary turnover beyond more frequently tested work attitude and withdrawal cognition variables. Using two samples, 339 registered nurses and 234 insurance company employees, three mea- sures of job search were distinguished: preparatory job search behav- ior, active job search behavior, and general effort job search. Active job search behavior had a stronger relationship to voluntary turnover than preparatoryjob search behavior or general effort job search, and it accounted for significant additional turnover variance beyond work at- titude and withdrawal cognition variables. Stronger results were found when unavoidable leavers were deleted from the turnover subsample. At any point in time, many people are searching for potential job op- portunities in different organizations (Schwab, Rynes, & Aldag, 1987). These individuals come from both the ranks of the currently employed (Rosenfeld, 1977) and the unemployed (Devens, 1988). The “search for job alternatives” variable has been included in models of voluntary turnover (e.g., Mobley, 1977; Mobley, Griffeth, Hand, & Meglino, 1979; Steers & Mowday, 1981). However, much more research attention has been given to the role of individual work attitudes (e.g., job satisfaction, organizational commitment) and withdrawal cognitions (e.g., thoughts of leaving, intention to quit) in the turnover process than to study- ing job search (e.g., Cotton & Tuttle, 1986; Gerhart, 1990; Kopelman, Rovenpor, & Millsap, 1992; Lee & Mowday, 1987; Mowday, Koberg, & McArthur, 1984). Even with statistical corrections, much variance in tumover remains unexplained by such work attitude and withdrawal cognition variables (Steel & Ovalle, 1984). The purposes of this study are to: (a) distinguish among types of job search measures, and (b) initially test the usefulness of a new job search behavior measure for explaining voluntary turnover The author thanks Rosaline Gorin for her help on the LISREL analyses,and the editor Correspondence should be sent to Gary Blau, Human Resource Administration De- and three anonymous reviewers for their helpful comments on the manuscript. partment, Speakman Hall, Room 384, Temple University, Philadelphia, PA 19122. COPYRlGM 0 1993 PERSONNELPSYCHOLoCiY, INC. 313

Upload: gary-blau

Post on 21-Jul-2016

237 views

Category:

Documents


7 download

TRANSCRIPT

Page 1: FURTHER EXPLORING THE RELATIONSHIP BETWEEN JOB SEARCH AND VOLUNTARY INDIVIDUAL TURNOVER

PERSONNEL ESYCHOWCil 1993.46

FURTHER EXPLORING THE RELATIONSHIP BETWEEN JOB SEARCH AND VOLUNTARY INDIVIDUAL TURNOVER

GARY BLAU Human Resource Administration Department

Temple University

This study tested the usefulness of a new job search behavior measure to account for voluntary turnover beyond more frequently tested work attitude and withdrawal cognition variables. Using two samples, 339 registered nurses and 234 insurance company employees, three mea- sures of job search were distinguished: preparatory job search behav- ior, active job search behavior, and general effort job search. Active job search behavior had a stronger relationship to voluntary turnover than preparatory job search behavior or general effort job search, and it accounted for significant additional turnover variance beyond work at- titude and withdrawal cognition variables. Stronger results were found when unavoidable leavers were deleted from the turnover subsample.

At any point in time, many people are searching for potential job op- portunities in different organizations (Schwab, Rynes, & Aldag, 1987). These individuals come from both the ranks of the currently employed (Rosenfeld, 1977) and the unemployed (Devens, 1988). The “search for job alternatives” variable has been included in models of voluntary turnover (e.g., Mobley, 1977; Mobley, Griffeth, Hand, & Meglino, 1979; Steers & Mowday, 1981). However, much more research attention has been given to the role of individual work attitudes (e.g., job satisfaction, organizational commitment) and withdrawal cognitions (e.g., thoughts of leaving, intention to quit) in the turnover process than to study- ing job search (e.g., Cotton & Tuttle, 1986; Gerhart, 1990; Kopelman, Rovenpor, & Millsap, 1992; Lee & Mowday, 1987; Mowday, Koberg, & McArthur, 1984).

Even with statistical corrections, much variance in tumover remains unexplained by such work attitude and withdrawal cognition variables (Steel & Ovalle, 1984). The purposes of this study are to: (a) distinguish among types of job search measures, and (b) initially test the usefulness of a new job search behavior measure for explaining voluntary turnover

The author thanks Rosaline Gorin for her help on the LISREL analyses, and the editor

Correspondence should be sent to Gary Blau, Human Resource Administration De- and three anonymous reviewers for their helpful comments on the manuscript.

partment, Speakman Hall, Room 384, Temple University, Philadelphia, PA 19122.

COPYRlGM 0 1993 PERSONNELPSYCHOLoCiY, INC.

313

Page 2: FURTHER EXPLORING THE RELATIONSHIP BETWEEN JOB SEARCH AND VOLUNTARY INDIVIDUAL TURNOVER

314 PERSONNEL PSYCHOLOGY

beyond more frequently tested work attitude and withdrawal cognition variables. Before discussing the job search construct it is important to consider how job search fits into the voluntary turnover process accord- ing to existing turnover models.

The Predicted Role of Job Search in the Turnover Process

One reason for empirical research “neglecting” the job search vari- able is that turnover models either predict job search to be a less prox- imal or direct antecedent of turnover than other variables (i.e., March & Simon, 1958; Mobley, 1977; Mobley et al., 1979; Steers & Mowday, 1981), or the turnover models do not specifically include job search (i.e., Price, 1977; Price & Mueller, 1981). March and Simon (1958) hypoth- esized turnover to be a function of perceived desirability of movement and perceived ease of movement. Propensity to search was expected to be a distal or indirect antecedent of perceived ease of movement. Mob- ley (1977) predicted that intention to quit was the most direct cause of turnover. Job search was expected to follow intent to search and to pre- cede intention to quit. In a latter schematic representation (Mobley et al., 1979), job search was not separately shown, instead intent to search and intent to quit were the most immediate predictors of turnover.

Steers and Mowday (1981) proposed that the closest antecedent of turnover was the interaction of intention to leave and alternative job op- portunities (ease of movement). Job search was expected to help ex- plain alternative job opportunities. Price (1977) proposed that the in- teraction between job satisfaction and ease of movement directly pre- ceded turnover, while in a latter turnover model for nurses Price and Mueller (1981) indicated that intent to stay was the immediate turnover antecedent.

Based upon the above mentioned turnover models, job search would not be expected to make a unique contribution to explaining volun- tary turnover beyond these hypothesized more immediate antecedents. However, in their path analysis of the Mobley (1977) model, Hom, Grif- feth, and Sellaro (1984) found that job search and expected utility of searchhvithdrawal had direct effects on voluntary turnover. In a more recent longitudinal study of Mobley’s model, Hom and Griffeth (1991) found that withdrawal cognitions (thoughts of leaving, intent to search, intent to quit) and job search had direct effects on turnover. Hom and Griffeth (1991) surveyed registered nurses at three points in time, and found that the impact of withdrawal cognitions and job search on turnover increased over time.

Page 3: FURTHER EXPLORING THE RELATIONSHIP BETWEEN JOB SEARCH AND VOLUNTARY INDIVIDUAL TURNOVER

GARY BLAU 3 15

Thus, there is some empirical support indicating that job search may be a more direct determinant of individual turnover than previously con- ceptualized. Such results have potential implications for rethinking job search’s role in the turnover process. Prior to testing job search’s contri- bution for explaining voluntary turnover, it is important to discuss how job search is measured.

Measuring Job Search

In one set of studies (e.g., Feather & O’Brien, 1986; Hom & Griffeth, 1991; Hom et al., 1984) investigators used measures of job search em- phasizing general effort, while in a second set of studies (e.g., Dyer, 1973; Kanfer & H u h , 1985; Kopelman et al., 1992) specific job search behav- iors were used. For example, Hom et al. (1984) asked individuals such questions as: how much effort they expended in their job search, active- ness (never defined) of search, and how much time they spent looking for a job. General effort job search scales, containing items measuring effort and time, may not be as effective in explaining subsequent turnover be- havior because a general effort job search measure does not test how an individual searches (i.e., what that person specifically does or does not do).

Ideally, a job search scale should emphasize behaviors, because it is specific behavior(s) that best translate withdrawal cognitions into turn- over behavior (Kopelman et al., 1992). Furthermore, a “behavior only” job search scale would allow for investigating whether some job search behaviors explain subsequent turnover (and re-employment) better than other job search behaviors. Job search behaviors represent the specific activities through which effort and time are spent on job search (i.e., they are the “translation” of an individual’s choices; Tubbs & Ekeberg, 1991). There has been some theoretical work arguing that job search can be broken down into two distinct behavioral phases: preparatory and active (Bowen, 1982; Soelberg, 1967).

In his job search and choice model, Soelberg (1967) conceptually broke job search into two phases: planning job search, and job search and choice. Power and Aldag (1985) noted that additional research was needed to operationalize Soelberg’s (1967) model. During the plan- ning job search phase, Soelberg argued that individuals would allo- cate resources (i.e., time, effort, money) to help produce initial job al- ternatives to consider. Behaviorally, this suggests a “preparatory” job search phase in which individuals gather potential job leads through var- ious sources (e.g., relatives, newspaper, previous employers, current col- leagues). Preparatory job search behaviorally measures the effort one makes to gather job search information (Stem, Shaw, & Noe, 1989).

Page 4: FURTHER EXPLORING THE RELATIONSHIP BETWEEN JOB SEARCH AND VOLUNTARY INDIVIDUAL TURNOVER

316 PERSONNEL PSYCHOLOGY

The job search and choice phase began (Soelberg, 1967) when indi- viduals formally activated their search. Behaviorally, active job search finds the job seeker publicly and unambiguously communicating one’s availability through such behaviors as sending out resumes, telephoning, and interviewing with prospective employers. Active job search behav- iorally measures one’s “commitment” to hisher job search.

More recently, Bowen (1982, p. 207) has proposed two cycles of job search by the employed which supports the preparatory-active distinc- tion. During the first (preparatory) cycle of job search the employee determines the availability of “greener pastures” (i.e., are there other desirable jobs out there?). In the second (active) cycle the employee de- termines the “accessibility of those greener pastures” (i.e., applying to get another job).

Current job search behavior measures do not distinguish between preparatory and active search components. Many previous studies have used one-item job search measures. For example, number of employ- ers contacted (Dyer, 1973), number of applications filed (Ellis & Thy- lor, 1983), number of times an individual looked for another job (Blue- dom, 1982), and “activeness” (never defined) of job search (Lee & Mow- day, 1987; Spencer, Steers, & Mowday, 1983). Other studies have used multiple-item job search behavior measures (e.g., Becker, 1980; Black, 1981; Kanfer & H u h , 1985; Kopelman et al., 1992; Schmit, Amel, & Ryan, 1993; Sheppard & Belitsky, 1966; Vinokur & Caplan, 1987), but none make the distinction between preparatory and active search.

Testing the Relationship Between Job Search and Voluntary Turnover

If active job search measures individuals’ behavioral commitment to their job search, then active job search would be expected to show a stronger positive relationship to individual turnover, versus preparatory or general effort job search. Following the logic of Soelberg’s (1967) model, there is a closer linkage between a person’s active job search and leaving versus planning one’s job search and leaving. This connection is not tautological because despite high levels of active search an individual may not get another job offer allowing himher to leave. A recent meta- analytic study by Hom, Caranikas-Walker, Prussia, and Griffeth (1992) indicated that moderators beyond an individual’s control (e.g., an occu- pation’s labor market) will affect the turnover process.

Based upon this discussion of job search measures, it is expected that active job search behavior, but not preparatory job search behavior or general effort job search, will account for significant turnover beyond controlled-for antecedents. An integration of representative empirical

Page 5: FURTHER EXPLORING THE RELATIONSHIP BETWEEN JOB SEARCH AND VOLUNTARY INDIVIDUAL TURNOVER

GARY BLAU 317

findings (i.e., Arnold & Feldman, 1982; Bluedorn, 1982; Cotton & Tut- tle, 1986; Gerhart, 1990; Hom et al., 1984; Hom & Griffeth, 1991; Lee & Mowday, 1987; Michaels & Spector, 1982; Miller, Katerberg, & Hulin, 1979; Mobley, Homer, & Hollingsworth, 1978; Mowday et al., 1984) testing the voluntary turnover models mentioned earlier (i.e., March, & Simon, 1958; Mobley, 1977; Mobley et al., 1979; Price, 1977; Price & Mueller, 1981; Steers & Mowday, 1981) suggests that it would be methodologically appropriate to control for eight relevant turnover an- tecedents prior to testing the job search measures: (a) demographics (i.e., gender, age, tenure, education, marital status), (b) job satisfaction, (c) organizational commitment, (d) withdrawal cognitions, (e) ease of movement, ( f ) expected utility of withdrawal, (g) job satisfaction by ease of movement, and (h) intent to leave by ease of movement. From this literature review, three hypotheses will be tested:

Hypothesis I : Three types of job search are distinguishable: general effort job search, preparatory job search behavior, and active job search behav- ior.

Hypothesis 2: Active job search behavior will have a stronger positive relationship to voluntary turnover than preparatory job search behavior or general effort job search.

Hypothesis 3 Active job search behavior, but not preparatory job search behavior or general effort job search, will account for significant turnover variance beyond demographics, job satisfaction, organizational commit- ment, withdrawal cognitions, ease of movement, expected utility of with- drawal, job satisfaction by ease of movement, and intent to leave by ease of movement.

Method

To provide a stronger test for these hypotheses, data from two dif- ferent samples were collected. The first sample, consisting of registered nurses, was more occupationally homogeneous then the second sample which consisted of insurance company employees.

Nursing Sample and Procedure

Of the 746 surveys administered across three hospitals in a large east- em city covering the hospitals’ full-time registered nurse population, 339 (45%) were voluntarily completed and returned. For each hospital the subject participation rates were approximately the same (i.e., 42%, 44%, and 51%). Participants were told that the goal of the survey was to ex- plore work-related attitudes and perceptions, and that the results would

Page 6: FURTHER EXPLORING THE RELATIONSHIP BETWEEN JOB SEARCH AND VOLUNTARY INDIVIDUAL TURNOVER

318 PERSONNEL PSYCHOLOGY

be used only for general human resource research purposes. Complete confidentiality of participant responses was maintained and respondents mailed their surveys directly back to the investigator (using preaddressed business reply envelopes).

A demographic breakdown of the 339 participants showed that: (a) 94% were female; (b) 86% held the baccalaureate degree in nursing; (c) 62% were married; (d) 83% were Caucasian; (e) mean tenure as a registered nurse was 10.8 years, while mean length of service at each hospital was 5.6 years; ( f ) average age was 35 years; and (g) 48% worked on a day shift, 34% worked on a rotating shift, and the remaining 18% worked on either an evening or a night shift.

A demographic comparison of the 339 respondents to the 407 non- respondents (746-339) across the three hospitals showed that the re- spondent sample was not significantly different from the non-respondents concerning gender, education, marital status, race, hospital tenure, and shift. However, respondents were significantly older (35 vs. 32 years) and had higher (10.8 vs. 8.9 years) nurse tenure. The last four digits of the re- spondent’s social security number was provided on the survey, allowing later subject turnover to be matched to prior survey responses.

Insurance Company Sample and Procedure

%o hundred thirty-four of the 407 surveys administered to insurance company employees were voluntarily completed and returned (57% re- sponse rate). The insurance company was non-unionized and headquar- tered in a large eastern city. A survey administration process identical to the process used on the registered nurses was followed (i.e., same sur- vey rationale, use of business reply envelopes, four digit social security number on survey to collect later turnover data).

A demographic breakdown of the 234 participants indicated that: (a) 51% were male; (b) 47% had at least a baccalaureate degree; (c) 76% were married; (d) 79% were Caucasian; (e) mean tenure in the insurance company was 6 years; ( f ) average age was 32 years; and (g) a functional area breakdown indicated that 23% were clerical, 40% ,were field office, 18% were administrative, 11% were information systems, and 8% were maintenance personnel. A demographic comparison of the 234 respondents to the 173 non-respondents (407-234) indicated that the respondent sample was significantly different on two variables, company tenure and age. Respondents had less company tenure (6 vs. 9 years), and were younger (32 vs. 37 years).

Page 7: FURTHER EXPLORING THE RELATIONSHIP BETWEEN JOB SEARCH AND VOLUNTARY INDIVIDUAL TURNOVER

GARY BLAU 319

Measures

Job search was measured using 12 behavioral items and 4 general effort items. The 12 behavioral items were based on previous measures of job search behavior (e.g., Dyer, 1972; Kanfer & Hulin, 1985; Sheppard & Belitsky, 1966; Vinokur & Caplan, 1987). The 4 general effort items were based on Hom and Griffeth (1991) and Hom et al. (1984). Of these 12 behavioral items, 6 items were expected to represent preparatory job search behaviors, and the remaining 6 items would represent active job search behaviors (Soelberg, 1967). Confirmatory factor analysis (CFA) was used to test the dimensionality of job search since three dimensions were hypothesized a priori (Nunnally, 1978): preparatory job search behavior, active job search behavior, and general effort job search. The CFA procedure in LISREL (Joreskog & Sorbom, 1986) was used.

Subjects were asked to indicate the frequency with which they car- ried out each behavioral item within the last 6 months on a 5-point scale, where 1 = “Never” (0 times), 2 = “Rarely” (1 or 2 times), 3 = “Occa- sionally” (3 to 5 times), 4 = “Frequently” (6 to 9 times), and 5 = “Very frequently” (at least 10 times). A 6-month interval was used to increase variance in job search behaviors between subjects, while minimizing ret- rospective bias in recounting job search behaviors. A shorter time inter- val (e.g., 3 months) would have reduced the variance in job search be- haviors, while a longer time interval (e.g., 1 year) would have increased retrospective bias. Using the same 6-month time frame for both samples allows for a stronger comparison of findings between samples.

A frequency range was given within each verbal anchor under the assumption that subjects are more likely to correctly recall an approxi- mate (versus exact) number of times they carried out a particular behav- ior (e.g., Vinokur & Caplan, 1987). This 5-point response scale allowed the frequency of each search behavior to be more precisely measured, as opposed to previous search measures using a yes/no response scale (e.g., Black, 1981; Kanfer & H u h , 1985; Kopelman et al., 1992; Sheppard & Belitsky, 1966). The same 6-month time frame was used in asking sub- jects to respond to the general effort job search items on a 5-point scale (1 = “Strongly disagree,” 5 = “Strongly agree”).

Controlled-for antecedents of turnover. Job satisfaction was mea- sured using the 20-item short-form version of the Minnesota Satisfac- tion Questionnaire (Weiss, Dawis, England, & Lofquist, 1967). Orga- nizational commitment was measured using the 9-item short-form ver- sion of Porter, Crampon, and Smith’s (1976) 15-item scale. The short- form was used to reduce survey length and to eliminate items containing withdrawal intentions. Withdrawal cognitions were measured by linearly

Page 8: FURTHER EXPLORING THE RELATIONSHIP BETWEEN JOB SEARCH AND VOLUNTARY INDIVIDUAL TURNOVER

320 PERSONNEL PSYCHOLOGY

summing three items: thinking of quitting, intention to search, and in- tention to quit (Hom & Griffeth, 1991). Ease of movement was mea- sured using a 3-item scale previously tested by Peters, Jackofsky, and Salter (1981). Item responses for these variables were recorded on a 5-point scale (1 = “Strongly disagree,” 5 = “Strongly agree”).

Expected utility of withdrawal was measured by combining two 3- item measures of attitude toward job search and attitude toward quit- ting (Hom & Griffeth, 1991; Hom et al., 1984). Attitude toward job search was measured by asking subjects to evaluate the statement “For me, searching for an alternative to my present job is . . . .” on three 5-point semantic differential scales (awful-nice, pleasant-unpleasant, bad-good). Attitude toward quitting one’s job was measured with the same 5-point semantic differential scales. In addition, a methodological check variable, the 12-item subscale “Claim Good Qualities” (Robinson & Shaver, 1973) from Crowne and Marlowe’s (1964) 33-item social de- sirability measure was used. A sample item is: “I am always willing to admit it when I make a mistake” (1 = “Strongly disagree”; 6 = “Strongly agree”).

Turnover data was collected on participating subjects 12 months uf- ter the survey. Participants who had left their job and organization were coded 2, while stayers were coded 1. One year after the survey, 95 out of the 339 nurses (28%) had voluntarily left their jobs and hospitals. This overall voluntary turnover percentage was broken down across hospitals as 27%, 24%, and 35%. Of these 95 leavers, exit interview data indicated that 24 leavers said that they left for an “unavoidable ” (Abelson, 1987) non-job related reason (e.g., family-related, spouse-imposed relocation, medical-related). Of course such interview data may contain classifica- tion errors regarding the “real reason” for turnover (Campion, 1991). Leaving for an “unavoidable” reason may be a socially polite response so that such leavers can get a good future job recommendation. Follow- ing Abelson, turnover data will be analyzed with (i.e., All lkrnover) and without (i.e., Avoidable Turnover) these 24 leavers.

Sixty-three out of the 234 insurance company employees (27%) had voluntarily left their jobs and the company 1 year after the survey. Of these 63 leavers, exit interview data indicated that 13 had left for an “unavoidable” (Abelson, 1987) non-job related reason. lbrnover data with (i.e., All Turnover) and without (i.e., Avoidable Turnover) these 13 leavers will be analyzed.

Results

The confirmatory factor analysis (CFA) results (oblique solution) for the a priori hypothesized job search dimensions, using the registered

Page 9: FURTHER EXPLORING THE RELATIONSHIP BETWEEN JOB SEARCH AND VOLUNTARY INDIVIDUAL TURNOVER

GARY BLAU 321

TABLE 1 Factor Loadings of Job Search Items

Registered Insurance nurses employees

1 2 3 1 2 3 Item (item means in parentheses)

1. Read the help wantedlclassified ads in a news aper, journal, or professional association. (4.5 & 4.f)

2. Listed ourself as a job applicant in a newspa journal or professional association. (3.1 & 2 . v

3. Preparedrevised your resume. (3.0 & 2.8) 4. Sent out resumes to potential employers (4.0 & 3.9) 5. Filled out a job application. (3.9 & 3.7) 6. Read a book or article about getting a job or

changing jobs. (2.3 & 2.1)

.72 .63 - -

.64

.89

.80

- - .49 - .75 - - - .81 - - .83 -

.59 .51 - - - -.

7. Had a job interhew witha prospective employer. (2.6 & 2.4) .83 - .73 -

8. W e d with friends or relatives about possible job leads. (4.3 & 4.3)

9. Contacted an employment agency, executive search firm, or state employment service. (1.5 & 1.8)

10. Spoke with previous employers or business acquain- tances about their knowing of potential job leads.

11. Telephoned a prospective employer. (2.8 & 2.2) 12. Used current within company resources (e.g., col-

13. S ent a lot of time looking for a job alternative.

14. Devoted much effort to looking for other jobs.

15. Focused my time and effort on job search activities.

16. Gave best effort to find a new job. (3.0 & 2.7)

(3.7 & 3.3)

leagues) to generate potential job leads. (2.5 & 2.4)

($4 & 3.3)

(3.5 & 3.2)

(3.1 & 2.9)

.68 .77 - -

.39 - .42 -

.61 -

.57 - - - 5 2 -

- .75

S O .43 - -

.54

.59

- - .58

- - .56

.62

.60 - - .51 - - .64

Note: (-) denotes parameter item loading fixed at zero. All items significant, p<.O1. Registered Nurses - chi-square/df (201.76/104) = 1.94; GFI = .93; AGFI = .90; RMSR = .08. Insurance Employees - chi-square/@ (167.43/104) = 1.61; GFI = .92; AGFI = .89; RMSR = .09. Factor 1 = Preparatory Job Search Behavior (PJSB); Factor 2 = Active Job Search Behavior (AJSB); Factor 3 = General Effort Job Search (GETS).

nurse and insurance employee samples, are shown in Thble 1. As Ta- ble 1 indicates, the pattern of factor loadings for the 6-item measures of preparatory and active job search behaviors, and 4-item measure of general effort job search represent the corresponding conceptual dimen- sions of preparatory job search behavior, active job search behavior, and general effort job search across both samples. Item means are reported in parentheses after each item, first for the registered nurses and then for the insurance employees. The item means show a consistent pat- tern across samples. The most frequently enacted preparatory behaviors are reading the want ads and talking with friends or relatives. Sending

Page 10: FURTHER EXPLORING THE RELATIONSHIP BETWEEN JOB SEARCH AND VOLUNTARY INDIVIDUAL TURNOVER

324 PERSONNEL PSYCHOLOGY

TABLE 3 Means, Standard Devia@ns Correlations, and Reliabilities

Among Key Study Kznabtes for Insurance Employees

Variable M S D 1 2 3 4 5 6 7 8 9

1. Preparatory job search 19.3 3.8 (.80)a .~

2. Active job search 16.8 3.4 .44** (33) . ,

3. GCneral eff jobsearch 12.2 2.5 .53** .37** (.74)

4.Job satisfac. 61.6 10.9 -.25** -.20** -.17** (37) 5. Organiza.

6. Withdrgyal

7. Ease of

8. Expected util.

commit. 28.7 5.8 -28.. -.23** -26’. .39** (34)

cognitions 8.9 2.3 .29** .25** .31** -.41** -.39** (.SO)

movement 9.7 2.2 .26** 20** 23.. -.18* -.13* .23** (.72)

of withdr. 17.4 3.0 .31** 24** .32** -.33** -.35** .40** .21**(.71)

9.Allturnover 1.3 0.3 .18** .41** .20**-.15* -.17** .30** .14* .13* (NA)

turnoverb 1.2 0.4 .21** .45** .22** -.18** -.19** .33** .16* .17* (NA) 10. Avoidable

Note: N=234 %ternal consistency estimate in parentheses; NA=not applicable. bAvoidable turnover N=221, because 13 unavoidable leaven deleted. *p<.05; **p<.Ol

& Cohen, 1983) was used to test this hypothesis. Since turnover is a dichotomous variable (leavehay), Steel and Griffeth (1989) noted that as the dichotomous split deviates from an optimal range between .25 and .75, ordinary least-squares (OLS) regression results become distorted. The categorical splits for each sample (28/72, nurses; 27/73, insurance employees) are close to these range limits. Therefore, logistic regression will be used to check the replicability of the initial OLS regression results (Huselid & Day, 1991).

Hierarchical regression analyses for registered nurses and insurance employees are presented in Tables 4 and 5. Avoidable turnover was used since this is a “purer” measure of subjects who left for another job (Abelson, 1987). Consistent results are found across both tables. In Step 1, the demographic variables (gender, age, tenure, education, and marital status) were entered together as a predictor set. They accounted for a significant amount of variance for insurance employees but not for registered nurses. One contributing factor for these different sample results is that there was less variability on gender and education for the nursing sample.

In Step 2, job satisfaction, organizational commitment, withdrawal cognitions, ease of movement, and expected utility of withdrawal were

Page 11: FURTHER EXPLORING THE RELATIONSHIP BETWEEN JOB SEARCH AND VOLUNTARY INDIVIDUAL TURNOVER

GARY BLAU 323

TABLE 2 Means Standard Deviationq Correlatiotp, and Reliabilities

Among Key Study knables for Regrrtered Nurses

Variable M S D 1 2 3 4 5 6 7 8 9

1. Preparatory

2. Active job

3. General eff.

4.Job satisfac. 58.2 10.7 -.23** -.18** -.21** ( 3 5 ) 5. Organiza.

6. Withdriiyal

7. Ease of

8. Expected util.

jobsearch 20.3 4.1 (.81)a

search 17.9 3.6 .49** (80)

jobsearch 13.1 2.5 .57** .38** (.76)

commit. 26.3 6.0-.26** -.20** -.27** .40** (.83)

cognitions 9.7 2.8 .31** .27** .33** -.43** -.41** (.&I)

movement 10.4 2.3 .28** .24** .25** -.17** -.15* .21** (.77)

of withdr. 18.1 3.4 .34** .26** .37** -.39** -.38** .44** . 20** (.74)

9.All turnover 1.3 0.4 .20** .43** .23** -.16** -.14* .34** .13* .15* (NA)

turnover' 1.2 0.3 .23** .48** .25** -.18** -.17* .37** .16* .18** (NA) 10. Avoidable

Note: N=339 %ternal consistency estimate in parentheses; NA=not applicable. 'Avoidable turnover N=315, since 24 unavoidable leaven deleted. *p<.05; **p<.Ol

behavior, and general effort job search are three related but distinguish- able job search measures. Tmble 3 shows the variable means, standard de- viations, correlations, and reliabilities using insurance employees. These results are similar to those reported in Tmble 2 for registered nurses, and also provide support for Hypothesis 1.

According to Hypothesis 2, active job search behavior will have a stronger positive relationship to turnover than preparatory job search behavior or general effort job search. The results in Tmble 2 (registered nurses) and Tmble 3 (insurance employees) support Hypothesis 2. The active job search (all and avoidable turnover correlations) are signifi- cantly higher (p < .05) than the preparatory job search (all and avoid- able turnover) and general effort job search (all and avoidable turnover correlations) (McNemar, 1969).

According to Hypothesis 3, active job search behavior, but not preparatory job search behavior or general effort job search, will ac- count for significant turnover variance beyond demographics (gender, age, tenure, education, marital status), job satisfaction, organizational commitment, withdrawal cognitions, ease of movement, expected util- ity of withdrawal, job satisfaction by ease of movement, and intent to leave by ease of movement variables. Hierarchical regression (Cohen

Page 12: FURTHER EXPLORING THE RELATIONSHIP BETWEEN JOB SEARCH AND VOLUNTARY INDIVIDUAL TURNOVER

324 PERSONNEL PSYCHOLOGY

TABLE 3 Means, Standard Deviatiqns Correlations, and Reliabilities

Among Key Study Kz’anabtes for Insurance Employees

Variable M S D 1 2 3 4 5 6 7 8 9

1. Preparatory

2. Active job

3. Gen ral eff

4.Job satisfac. 61.6 10.9 -25’. -.20** -.17** (.87) 5. Organiza.

6. Withdrpyd

7. Ease of

8. Expected util.

9.Allturnover 1.3 0.3 .18** .41** .20**-.15* -.17** .30** .14* .13* (NA)

turnover* 1.2 0.4 .21** .45** .22** -.18** -.19** .33** .16* .17* (NA)

jobsearch 19.3 3.8 (.80)a

search 16.8 3.4 .44** (.83)

108 search 12.2 2.5 .53** .37** (.74)

commit. 28.7 5.8 -.28** -.23** -.26** .39** (.84)

cognitions 8.9 2.3 .29** .25** .31** -.41** -.39** (.80)

movement 9.7 2.2 .26** .20** .23** -.18* -.13* .23**(.72)

of withdr. 17.4 3.0 .31** .24** .32** -.33** -.35** .40** .21** (.71)

10. Avoidable

Note: N=234 a h t e n d consistency estimate in parentheses; NA=not applicable. bAvoidable turnover N =221, because 13 unavoidable leaven deleted. *p<.o5; **p<.o1

& Cohen, 1983) was used to test this hypothesis. Since turnover is a dichotomous variable (leavehtay), Steel and Griffeth (1989) noted that as the dichotomous split deviates from an optimal range between .25 and .75, ordinary least-squares (OLS) regression results become distorted. The categorical splits for each sample (28172, nurses; 27/73, insurance employees) are close to these range limits. Therefore, logistic regression will be used to check the replicability of the initial OLS regression results (Huselid & Day, 1991).

Hierarchical regression analyses for registered nurses and insurance employees are presented in Thbles 4 and 5. Avoidable turnover was used since this is a “purer” measure of subjects who left for another job (Abelson, 1987). Consistent results are found across both tables. In Step 1, the demographic variables (gender, age, tenure, education, and marital status) were entered together as a predictor set. They accounted for a significant amount of variance for insurance employees but not for registered nurses. One contributing factor for these different sample results is that there was less variability on gender and education for the nursing sample.

In Step 2, job satisfaction, organizational commitment, withdrawal cognitions, ease of movement, and expected utility of withdrawal were

Page 13: FURTHER EXPLORING THE RELATIONSHIP BETWEEN JOB SEARCH AND VOLUNTARY INDIVIDUAL TURNOVER

GARY BLAU 325

TABLE 4 Hierarchical Regressions Testing the Avoidable Turnover Variance Accounted for

by Search Measures Beyond Controlled-for Antecedents for Registered Nurses

(Step) Predictor set Fa R R' AR'

(1) Gender+Age+Tenure+Education+Marital status 1.86 .20 .04 (2) Job satisfaction+Organizational commitment+

Withdrawal cognitions+Ease of movement+ Expected utility of withdrawal 5.07** .48 .23 .19

(3) Job satisfaction * Ease of movement* .66 .49 .24 .01 (3) Intent to leave * Ease of movement* .93 .SO .25 .02 (4) Preparatory job search' 1.01 .52 .27 .02 (4) General effort job search' 1.45 .53 .28 .03 (4) Active job search' 3.21* .57 .33 .08

Note: (N=315) aSignificance of incremental variance. *Entered separately in different regression equations after Steps 1 and 2. 'Entered separately in different regression equations after Steps 1,2, and 3. *p<.05; **p<.Ol

TABLE 5 Hierarchical Regressions Testing the Avoidable Turnover Variance Accounted for by Search Measures Beyond Controlled-for Antecedents for Insurance Employees

(Step) Predictor set F~ R R= A R ~

(1) Gender+Age+Tenure+Education+Marital status 2.58* .24 .06 (2) Job satisfaction+Organizational commitment +

Withdrawal cognitions+Ease of movement+ Expected utility of withdrawal 4.65* .49 .24 .18

(3) Job satisfaction * Ease of movement* .27 .49 .24 .OO (3) Intent to leave * Ease of movement* .64 .SO .25 .01 (4) Preparatory job search' 1.15 .52 .27 .02 (4) General effort job search' 1.50 .53 .28 .03 (4) Active job search' 3.43. .56 .32 .07

Note: ( ~ = 2 2 1 ) =Significance of incremental variance. *Entered separately in different regression equations after Steps 1 and 2. 'Entered separately in different regression equations after Steps 1,2, and 3. *p<.05; **p<.01

entered together as a predictor set, and they accounted for a significant amount of turnover variance in both samples. In Step 3, two interaction terms were tested separately. As noted earlier, Price (1977) proposed that the interaction between job satisfaction and ease of movement was

Page 14: FURTHER EXPLORING THE RELATIONSHIP BETWEEN JOB SEARCH AND VOLUNTARY INDIVIDUAL TURNOVER

326 PERSONNEL PSYCHOLOGY

the immediate antecedent of employee turnover, while Steers and Mow- day (1981) predicted that the immediate antecedent was the interaction between intent to leave and ease of movement. As ’Iiibles 4 and 5 show, neither interaction term accounted for significant additional avoidable turnover variance. These results are consistent with previous research (e.g., Lee & Mowday, 1987; Price & Mueller, 1981).

In Step 4, the incremental avoidable turnover variance accounted for by preparatory job search, general effort job search, and active job search were separately tested. The results in Tables 4 and 5 support Hypothe- sis 3. Only active job search behavior accounted for additional significant avoidable turnover. These OLS regression results were replicated by lo- gistic regression (Huselid & Day, 1991). Corrections for shrinkage in R2 (Herzberg, 1969) indicated that 29% of avoidable turnover variance was accounted for using registered nurses and 26% using insurance employ- ees. By comparison, the adjusted R2 for the all turnover variable was 19% for registered nurses and 17% for insurance employees.

The results of this study provide additional empirical support for a two-dimensional measure of job search behavior (Blau, in press). The conceptual linkages in Soelberg’s (1967) model help to explain why the relationship between active job search behavior and voluntary turnover is stronger than the relationship between preparatory job search behav- ior and subsequent turnover. Although general effort job search has been found to be a significant determinant of turnover (Hom & Griffeth, 1991; Hom et al., 1984), general effort job search does not indicate how an individual searches. The results here suggest that general effort job search may not be as strong a predictor of turnover as active job search behavior. By operationalizing the “search for job alternatives” variable in turnover models (e.g., Mobley, 1977; Steers & Mowday, 1981) as spe- cific behaviors, the job search variable may play a more powerful role in understanding subsequent turnover.

Empirically showing that job search consists of two related but dis- tinct stages, preparatory and active job search, suggests a potential lengthening in the causal process of employee turnover not thus far shown in turnover models. However, longitudinal data, including the collection of preparatory job search behaviorsprior to active job search behaviors, is needed for a stronger causal analysis of job search’s role in the turnover process (Hom & Griffeth, 1991). Also, breaking the job search measures down into smaller time frames (e.g., a 3-month period) would allow investigators to more closely analyze the effectiveness of job

Page 15: FURTHER EXPLORING THE RELATIONSHIP BETWEEN JOB SEARCH AND VOLUNTARY INDIVIDUAL TURNOVER

GARY BLAU 327

seeking behavior over time. Thus, the present analyses constitute only an initial test for the impact of job search behavior on turnover.

What are some implications of a longer employee job search process for an organization? A longer search process may give the organiza- tion more “time” to find ways of retaining an employee it does not want to lose (i.e., dysfunctional turnover; Campion, 1991; Dalton, Todor, & Krackhardt, 1982). For example, finding out that an employee is gath- ering potential job leads (preparatory job search) may “push” that em- ployee’s superior to initiate serious discussion with hisher employee be- fore he or she goes past a “point of no return” for leaving (active job search resulting in another job offer).

Beyond the need for longitudinal data to more strongly test for job search stages, there are other limitations of this study. A 12-month time frame was used for gathering turnover data. This interval was desirable to create larger subsamples (stayers vs. leavers) for hypothesis testing. However, such a longer time frame may give a conservative estimate of the impact of job search measures on turnover. At the time of filling out the survey, one may have had little intention to leave their organization. Yet if by 9 months later that subject decided to leave, searched for a new job, found one, and resigned, then the survey measures would not predict such turnover. However, consistent results for explaining turnover were found in this study between the more mobile sample of registered nurses versus the less mobile insurance employees (Hom et al., 1992).

Future research may indicate that changes are needed in the prepara- tory and active job search behavior scales. Although the item content for both scales is based on previous job search research (e.g., Dyer, 1972; Kanfer & H u h , 1985; Sheppard & Belitsky, 1966), additional items may be needed to fully represent both stages of the job search construct (e.g., going to job fairs or information-related interviews, Vinokur & Caplan, 1987; seeking an internal transfer, Kopelman et al., 1992). Also, per- ceived ease of carrying out a behavior (e.g., reading a book or article about changing jobs vs. reading the daily want ads) may affect the fre- quency of performing that behavior. Wider discrepancies in search be- haviors may necessitate the use of z-scores (Laker, 1991). In it3 present form, the active job search scale cannot prescribe to the job seeker the degree to which different levels of active job search behaviors (e.g., send- ing out resumes, telephoning, interviewing) should be done to secure a new job.

However, the present study empirically supports the idea of break- ing job search behavior into distinct dimensions (Bowen, 1982; Soelberg, 1967). Identifying behavioral& what an individual does when seeking al- ternative employment seems to help explain turnover beyond its more traditional antecedents. Increasing attention is being given to the role of

Page 16: FURTHER EXPLORING THE RELATIONSHIP BETWEEN JOB SEARCH AND VOLUNTARY INDIVIDUAL TURNOVER

328 PERSONNEL PSYCHOLOGY

moderators in the turnover process (Hom et al., 1992). Blau (in press) found that task-specific self-esteem moderated the relationship between preparatory and active job search behavior such that there was a stronger relationship for high versus low self-esteem individuals. Finally, exclud- ing unavoidable leavers from the turnover analysis improved the correla- tion and regression results. This supports Abelson's (1987) study that the distinction between avoidable and unavoidable turnover is important.

REFERENCES

Abelson M. (1987). Examination of avoidable and unavoidable turnover. Journal ofAp-

Arnold H, Feldman D. (1982). A multivariate analysis of the determinants of job turnover. Journal ofApplied Psychology, 67,350-360.

Becker H. (1980). The assertive job-hunting survey. Measurement and Evaluaiion in Guid- ance, 13,4348.

Bentler P, Bonnet D. (1980). Significance tests and goodness of fit in the analysis of covariance structures. Psychological Bulletin, 88,588-606.

Black M. (1981). An empirical test of the theory of on-the-job search. Journal of Human Resources, 16,129-140.

Blau G. (in press). Testing a two dimensional measure of job search behavior. Orgnnita- tional Behavior and Human Decision Processes.

Bluedorn A. (1982). A unified model of turnover from organizations. Human Relaiions,

Bowen D. (1982). Some unintended consequences of intention to quit. Academy of Man- agement Review, 7,205-211.

Campion M. (1991). Meaning and measurement of turnover: Comparison of alternative measures and recommendations for research. Journal of Applied Psychology, 76, 199-212.

&hen J, Cohen P (1983). Applied regressionlcorrelation analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Erlbaum.

Cotton J, lbttle J. (1986). Employee turnover: A meta-analysis and review with implica- tions for research. Academy of Management Review, II,55-70.

Crowne D, Marlowe D. (1964). The approval motive: Studies evaluative dependence. New York Wiley.

Dalton D, Todor W, Krackhardt D. (1982). 'lbrnover overstated: The functional taxonomy. Academy of Management Review, 7,117-123.

Devens R. (1988). Employment and unemployment in the first half of 1988. Monthly Labor

Diasgow F, Kanfer R. (1985). Equivalence in psychological measurement in heteroge-

Dyer L. (1972). Managerial job seeking: Methods and techniques. Monthly Labor Review,

Dyer L. (1973). Job search success of middle-aged managers and engineers. Industrial and Labor Relations Review, 26,969-979.

Ellis R, Thylor S. (1983). Role of self-esteem within the job search process. Journal of Applied Psychology, 68,632-640.

Feather N, O'Brien G. (1986). A longitudinal study of the effects of employment and unemployment on school leavers. Journal of Occupational Psychology, 59,121-144.

plied PsyChOlOgy, 72,382-386.

35,135-153.

Review, 100,15-19.

neous populations. Journal of Applied Psychology, 70,662-680.

95,29-30.

Page 17: FURTHER EXPLORING THE RELATIONSHIP BETWEEN JOB SEARCH AND VOLUNTARY INDIVIDUAL TURNOVER

GARY BLAU 329

Gerhart B. (1990). Voluntary turnover and alternative job opportunity. Journal ofApplied

Hanisch K, H u h C. (1991). General attitudes and organizational withdrawal: An evalua-

Herting J. (1985). Multiple indicator models using LISREL. In Blalock H (Ed.), Causal

Henberg F! (1969). The parameters of cross-validation. Psychome&, Monograph Sup-

Horn P, Caranikas-Walker F, Prussia G, GrifTeth R. (1992). A meta-analytical structural

Psychology, 75,467476.

tion of a causal model. Journal of Vocational Behavior, 39,110-128.

models in the social sciences (2nd ed.). New York Aldine.

plement, No. 16. . .

equations analysis of a model of employee turnover. Journal ofApplied Psychology, 77,8%909.

Hom P, Griffeth, R. (1991). Structural equations modeling test of a turnover theory: Cross- sectional and longitudinal analyses. Journal ofApplied Psychology, 76,35&366.

Hom P, Griffeth R, Sellaro C. (1984). The validity of Mobley’s (1977) model of employee turnover. Organizational Behavior and Human Performance, 34,141-174.

Huselid M, Day L. (1991). Organizational commitment, job involvement, and turnover: A substantive and methodological analysis. Journal ofApplied Psychology, 76,383- 391.

Joreskog K. (1971). Simultaneous factor analysis in several populations. Psychome&, 36,409-426.

Joreskog K, Sorbom D. (1986). LISREL: Analysis of linear structural relationships by mari- mum likelihood. Chicago: National Education Resources.

Kanfer R, H u h C. (1985). Individual differences in successful job searches following lay-

Kopelman R, Rovenpor J, Millsap R. (1992). Rationale and construct validity evidence for the job search behavior index: Because intentions (and New Year’s resolutions) often come to naught. Journal of Vocational Behavior, 40,269-287.

Laker D. (1991). Job search, perceptions of alternative employment, and turnover. Journal ofApplied Research, 7,616.

Lee T, Mowday R. (1987). Voluntarily leaving an organization: An empirical investigation of Steers and Mowday’s model of turnover. Academy of Management Journal, 30,

Off. PERSONNEL F’SYCHOWGY, 38,8354347.

721-743. March J, Simon H. (1958). Organizations. New York Wiiey. McNemar Q. (1969). Psychological statistics (4th ed.). New York Wiley. Michaels C, Spector P. (1982). Causei of employee turnover: A test of the Mobley, Grif-

feth, Hand, and Meglino model. Journal ofApplied Psychology, 67,53-59. Miller H, Katerberg R, Hulin C. (1979). Evaluation of the Mobley, Homer, and Hollings-

worth model of employee turnover. Journal ofApplied Psychology, 64,509-517. Mobley W. (1977). Intermediate linkages between job satisfaction and employee turnover.

Journal of Applied Psychology, 63,237-240. Mobley W, Griffeth R, Hand H, Meglino B. (1979). Review and conceptual analysis of the

employee turnover process. Psychological Bulletin, 86,493-522. Mobley W, Homer S , Hollingsworth A. (1978). An evaluation of precursors of hospital

turnover. Journal ofApplied Psychology, 63,408-414. Mowday R, Koberg C, McArthur A. (1984). The psychology of the withdrawal process: A

cross-validational test of Mobley’s intermediate linkages model of turnover in two samples. Academy of Management Journal, 27,79-94.

Nunnally J. (1978). Psychometric theory. New York McGraw Hill. Peters L, Jackofsky E, Salter J. (1981). Predicting turnover: A comparison of part-time

and full-time employees. Journal of Occupational Behavior, 2,89-98.

Page 18: FURTHER EXPLORING THE RELATIONSHIP BETWEEN JOB SEARCH AND VOLUNTARY INDIVIDUAL TURNOVER

330 PERSONNEL PSYCHOLOGY

Porter L, Crampon W, Smith E (1976). Organizational commitment and managerial turnover: A longitudinal study. Organizational Behavior and Human h f o m n c e ,

Power D, Aldag R. (1985). Soelberg’s job search and choice model: A clarification, review

Price J. (1977). The study of turnover. Ames: Iowa State University Press. price J, Mueller C. (1981). PmfesssioMl turnover: The case of nurses. New York: Spectrum. Robinson L, Shaver R. (1973). Measures ofpsychological attitudes. Ann Arbor: Institute

for Social Research, University of Michigan. Rosenfeld C. (1977). The extent of job search by employed workers. Monthly LaborReview,

100,58-62. Schmit M, Amel E, Ryan A. (1993). Self-reported assertive job seeking behaviors of

minimally educated job hunters. PERSONNEL PSYCHOLOGY, 46,105-124. Schwab D, Rynes S, Aldag R. (1987). Theories and research on job search and choice. In

Rowland K, Fems G (Eds.), Research inpersonnel and human resource management (Vol. 5, pp. 129-166). Greenwich, CT: JAI Press.

15,87-98.

and critique. Academy of Management Review, 10,48-58.

Sheppard H, Belitsky A. (1966). Thejob hunt. Baltimore: Johns Hopkins Press. Soelberg P. (1967). A study of decision making: Job choice. Unpublished doctoral disserta-

tion. Carnegie-Mellon University, Pittsburgh, PA. Spencer D, Steers R, Mowday R. (1983). An empirical test of the inclusion of job search

linkages into Mobley’s model of the turnover decision process. Journal of Occupa- tional Psychology, 56,137-144.

Steel R, Griffeth R. (1989). The elusive relationship between perceived employment op- portunity and turnover behavior: A methodological or conceptual artifact? Journal of Applied Psychology, 74,846-854.

Steel R, Ovalle N. (1984). A review and meta-analysis of research on the relationship between intentions and employee turnover. Journal ofApplied Psychobgy, 69,673- 686.

Steers R, Mowday R. (1981). Employee turnover and post-decision accommodation pro- cesses. In Cummings LL, Staw BM (Eds.), Research in organizational behavior (Vol. 3, pp. 235-281). Greenwich, CT: JAI Press.

Ste@ B, Shaw K, Noe A. (1989). Antecedents and consequences of job search behaviors. Journal of Vocational Behavior, 35,254-269.

Timm N. (1975). Multivariate anabsis with applications in education and psychology. Mon- terey, C A Brooks Cole.

lhbbs M, Ekeberg S. (1991). The role of intentions in work motivation: Implications for goal-setting theory and research. Academy of Management Review, 16,180-199.

Vinokur A, Caplan R. (1987). Attitudes and social support: Determinants of job seeking and well-being for the unemployed. Journal ofApplied Social Psychology, 17,1007- 1024.

Weiss D, Dawis R, England G, Lofquist L. (1967). Manual for the Minnesota Sati$action Questionnaire. Minneapolis: Industrial Relations Center, University of Minnesota.