Testing the Generalizability of a Career Commitment ... - Science Direct

32 downloads 100679 Views 1MB Size Report
Testing the Generalizability of a Career Commitment. Measure and Its Impact on Employee Turnover. GARY BLAU. Temple University. Using a longitudinally ...
Journal of Vocational

Behavior 35, 88-103 (1989)

Testing the Generalizability of a Career Commitment Measure and Its Impact on Employee Turnover GARY BLAU Temple

University

Using a longitudinally tracked sample of 133 full-time bank tellers from a large bank, this study tested the generahzability of a career commitment measure and its impact on emplyee turnover. The results indicated that career commitment could be reliably operationahzed and was distinct from job involvement and organizational commitment. Additional evidence for the convergent and discriminant validity of career commitment was also found. Career commitment was found to be significantly negatively related to turnover; however, this relationship was mediated by career withdrawal cognitions. Findings are discussed in terms of identifying the threshold level for operationalizing career commitment and the need for future research on career change. o 1989 Academic press, IIIC.

Individual career motivation has become an increasingly important topic to study in today’s changing workforce (e.g., Greenhaus, 1987; London, 1983). Part of understanding an individual’s career motivation involves examining that person’s level of career commitment. D. Hall (1971) and Morrow (1983) have conceptually distinguished career commitment from the related work attitudes of job involvement and organizational commitment. However, Morrow (1983) has also argued that additional empirical research is needed to further establish the discriminant validity of such work attitudes. In an exploratory study of career commitment using a sample of registered nurses, Blau (1985a) developed a reliable measure of career commitment which demonstrated discriminant validity relative to job involvement and organizational commitment measures. In this study, Blau (1985a) defined career commitment as “one’s attitude towards one’s profession or vocation” (p. 278). The anchors of “profession” and “vocation” were necessary so as not to make career commitment redundant with other related work attitude referents such as job involvement, work involvement, or organizational commitment. In a subsequent study of The author gratefully acknowledges the helpful manuscript comments of the two anonymous reviewers. Please send correspondence and reprint requests to Gary Blau, Human Resource Administration Department, Speakman Hall-Room 384, Temple University, Philadelphia, PA 19122. 88 OOOl-8791189 $3.00 Copyright All tights

Q 1989 by Academic Press. Inc. of reproduction in any form reserved.

CAREERCOMMITMENTANDEMPLOYEETURNOVER

89

career commitment, using field office employees from an insurance company and circulation supervisors from a newspaper company, Blau (1988) found further support for his career commitment measure. Career commitment could be reliably operationalized and was distinct from job involvement and organizational commitment in these two samples. However, in his discussion of career commitment, Blau (1988) also suggested revising the definition of career commitment to “one’s attitude toward one’s vocation, including a profession” (p. 295), since a profession is really a special type of vocation (Kerr, Von Glinow, & Schriesheim, 1977). Building upon Hall’s (1968) work, Kerr et al. (1977) identified six characteristics strongly present in an “ideal” profession: expertise, autonomy, commitment to work and the profession, identification with the profession, ethics, and collegial maintenance of standards. Other writers (e.g., Bartol, 1979; Benveniste, 1987) have agreed that these characteristics are important in defining a profession. Using this typology of a profession’s characteristics, nurses were classified as “highly” professional (Blau, 1985a), while insurance field office employees and newspaper circulation supervisors were described as “partially” professional (Blau, 1988). It was suggested by Blau (1988) that more appropriate item anchors in a career commitment scale for partially professional samples would be “field” (i.e., insurance field) or “industry” (i.e., newspaper industry) instead of “profession.” In concluding his discussion of career commitment, Blau (1988) raised the issue of the minimum level of professional characteristics required in a vocation, below which a measure of career commitment may no longer be distinctly operationalized or valid. Morrow and Goetz (1988) noted in their study of Hall’s (1968) multidimensional professionalism measure that workers may feel they have a job but not necessarily a profession or career. One purpose of this study is to test whether the career commitment scale can be reliably and distinctly operationalized using a “less” professional sample than previously studied (Blau, 1985a, 1988)-bank tellers. Following Kerr et al.‘s (1977) typology, bank tellers can be classified as below average on expertise, below average on autonomy, below average on commitment to work and profession, low on identification with the profession, average on ethics, and low on collegial maintenance of standards. If career commitment can be reliably and distinctly operationalized from job involvement and organizational commitment using a sample of bank tellers, this will further extend the generalizability of the career commitment scale. A second purpose of this study was to test for the impact of employee career commitment on subsequent turnover. Rhodes and Doering (1983) have proposed a model of career change which describes the process an individual moves through while changing careers. Building from Mobley’s turnover model (e.g., Mobley, 1977; Mobley, Griffeth, Hand, & Meglino,

90

GARY BLAU

1979), Rhodes and Doering (1983) posit that attitude toward one’s career (i.e., level of motivation, dissatisfaction) will lead to eventual career change (i.e., turnover) through intermediate career withdrawal cognitions (i.e., thoughts of changing careers, intention to search, intention to change). Little empirical research has tested the Rhodes and Doering (1983) model. London (1983) has argued that one’s level of career motivation depends upon three factors”areer identity, career insight, and career resilience. According to London (1983), career identity deals with the centrality of one’s career to one’s identity, career insight is the extent to which the person has a realistic view of him/herself, and career resilience examines a person’s resistance to career disruption. These factors affect subsequent individual career behaviors such as a decision to change careers. Career commitment or attitude toward one’s vocation is part of London’s (1983) career identity and career resilience components. Thus, combining Rhodes and Doering’s (1983) and London’s (1983) models logically implies that while career commitment should be significantly negatively related to subsequent employee turnover, career withdrawal cognitions should mediate this relationship. It is important to recognize that this proposed relationship will not exist if employee turnover is simply confined to individuals who leave their job/organization but stay in the same vocation or field. If turnover is limited to individuals changing jobs/organizations within the same field, then only job withdrawal cognitions and not career withdrawal cognitions would be predictive. Turnover must also include employees who leave their current vocation to go into a different field in order for the predicted career commitment-career withdrawal cognitions-turnover link to apply. Finally, the predicted turnover must be voluntary as opposed to involuntary (e.g., due to dismissal). Voluntary turnover is more heavily influenced by motivational factors (such as work attitudes) while involuntary turnover is more influenced by ability and performance-related factors (Cotton & Tuttle, 1986). To summarize, the purposes of this study were to further test the generalizability of a career commitment measure and to partially test the Rhodes and Doering (1983) model by investigating the link between career commitment and employee turnover. METHODOLOGY Subjects and Procedure Two surveys were constructed and administered to full-time bank tellers working for a large nonunionized bank based in a major northeastern city. Since tellers were located in various branches spread out over a considerable geographic area, the bank’s Human Resource Department coordinated the administration of the surveys. Subjects were told that

CAREER

COMMITMENT

AND

EMPLOYEE

TURNOVER

91

the purpose of the surveys was to explore their work-related attitudes and perceptions, and that participation in the study was voluntary. The bank was concerned about the high turnover rate of its teller workforce and thought that the surveys would provide some insight into the causes for this turnover. To ensure that subjects had at least some familiarity with the job and organization, only tellers with at least 3 months of job experience were eligible to participate. So that their survey responses could be matched over time, participating tellers were asked to give their name and the branch where they were employed. Subjects were assured that their individual survey responses were completely confidential. To help increase this confidentiality, subjects received a preaddressed business reply envelope showing the investigator’s University address with each survey. Thus, completed surveys were mailed directly back to the investigator. Of the 258 Time 1 (T,) surveys sent out (covering the total bank teller population with at least 3 months of experience), 174 (67%) were voluntarily completed and returned. A follow-up letter from the investigator and the bank’s Human Resource Vice President helped to increase the subject participation rate. These T, surveys asked participants for demographic and work attitude and perception information, including their feelings about their career, job, and organization. Approximately 6 months later at Time 2 (T2), a second shorter survey was sent to the 174 T, survey respondents. This second survey asked subjects for similar demographic information as the first survey, and for their thoughts about leaving their jobs and changing careers. Bank Human Resource personnel felt that 6 months was an adequate time period to examine how job involvement, organizational commitment, and career commitment could affect withdrawal cognitions. Of the 174 T, respondents, 133 (76%) completed and mailed back their T, surveys to the author. Follow-up on the 41 T2 nonrespondents indicated that 22 had left the bank between T, and T2, while the remaining 19 chose not to respond to the T2 survey. A demographic breakdown of the longitudinally tracked sample of 133 tellers showed that: (1) 90% were female; (2) 52% were married; (3) average age was 37 years; (4) average organizational tenure was 5.9 years; (5) 80% were Caucasian, 12% were Black, 4% were Hispanic, and 4% were Oriental; and (6) all tellers had a high school education, including 26% who had gone on to college. Conversations with Human Resource personnel indicated that these sample demographics were representative of the teller population within the bank. Measures Career commitment is defined as one’s attitude toward one’s vocation, including a profession (Blau, 1988). Seven items were used to measure career commitment on a five-point scale (1 = strongly disagree, 5 =

92

GARY

BLAU

strongly agree). As noted in Blau (198Sa), the original content for the career commitment items was partially based on items measuring occupational commitment (Downing, Dunlap, Hadley, & Ferrell, 1978) and career orientation (Liden & Green, 1980). Previous research has shown the general measure of career commitment used here to be reliable and valid across other samples. For example, using a longitudinal sample of nurses, Blau (1985a) found career commitment to have an internal consistency of .87 at Time 1 and .85 at Time 2 and a test-retest reliability of .67 over 7 months. In terms of discriminant validity, at Time 1 (T,) and Time 2 (T2) the correlations between career commitment and career withdrawal cognitions were significant (T, = - .38; T2 = - .41), while the correlations between career commitment and job withdrawal cognitions were not significant (T, = - .07; T2 = - .08). Additional supportive reliability and discriminant validity data were found by Blau (1988) using samples of newspaper company supervisors and insurance company field office employees. For the newspaper sample the internal consistency of the career commitment measure was .83, and it was .84 in the insurance sample. The discriminant validity of the career commitment measure was tested longitudinally in the insurance sample, and it was found that the relationship between career commitment and career withdrawal cognitions was significant (- .36) while the relationship of career commitment to job withdrawal cognitions was not significant (- .I 1). The seven items used in this study were as follows: (1) “If I could go into a different industry other than the banking industry which paid the same I would probably do so” (reverse-scored); (2) “I definitely want a career for myself in the banking industry”; (3) “If I could do it all over again, I would not choose to work in the banking field” (reversescored); (4) “If I had all the money I needed without working, I would probably still continue to work in the banking field”; (5) “I like this vocation (the banking field) too well to give it up”; (6) “This is the ideal vocation (the banking field) for a life work”; and (7) “I am disappointed that I ever entered the banking industry” (reverse-scored). Job involvement is defined as the degree to which the individual identifies with a job, that is, the importance of the job to one’s self-image (Dubin, 1956). For this study, a 9-item job involvement scale drawn from Kanungo’s (1982) W-item scale was used to measure job involvement. Many of Kanungo’s (1982) items (e.g., “I live, eat and breathe my job”; “the most important things that happen to me involve my job”) are based upon Lodahl and Kejner’s (1965) original job involvement measure. However, the Lodahl and Kejner (1965) measure was not used due to its factor stability problems (Blau, 1985b). Based upon his study of the dimensionality of job involvement, Biau (1985b) recommended that 9 of Kanungo’s (1982) 10 items be used to measure job involvement. Using

CAREER

COMMITMENT

AND

EMPLOYEE

TURNOVER

93

four distinct samples drawn from a midwestern university, Blau (1985b) found the following internal consistencies for Kanungo’s (1982) 9-item job involvement measure: administrators = .87, faculty = .85, clerical personnel = .83, and service/maintenance personnel = .86. A five-point scale (1 = strongly disagree to 5 = strongly agree) was used. Organizational commitment is defined in terms of the individual’s identification with the particular organization and its goals. For this study, the 9-item short-form version of Porter, Crampon, and Smith’s (1976) 15-item scale was used to measure organizational commitment. Item responses were recorded on a five-point scale (1 = strongly disagree to 5 = strongly agree). Evidence for the construct validity of the 15-item organizational commitment scale is provided by Mowday, Steers, and Porter (1979). The short-form scale was used to reduce survey length and to eliminate long-form items containing withdrawal intentions. Using a longitudinally tracked sample of nurses across three time periods, Blau (1987) found that the short-form organizational commitment scale demonstrated adequate internal consistency and test-retest reliability. Internal consistencies were T, = .87, T2 = .86, and T3 = .88. Test-retest reliabilities were T,-T2 (7 months) = .65, T,-T, (11 months) = .54, and T,-T, (18 months) = .43. Withdrawal cognitions were measured from two different perspectives, job and career, at T2. Job withdrawal cognitions were measured using three items theorized by Mobley (1977) and found (e.g., Miller, Katerberg, & Hulin, 1979) to be significant predictors of turnover: thinking of quitting, intention to search, and intention to quit. Answers to items were recorded on a five-point scale (1 = very unlikely to 5 = very likely). Consistent with previous research (e.g., Blau, 1985a; Michaels & Spector, 1982), a three-item job withdrawal cognitions scale was formed by linearly summing the responses to each item. Such a procedure allowed an internal consistency estimate to be calculated. Using a longitudinal sample of nurses tracked over 7 months, Blau (1985a) found a similar job withdrawal cognitions scale to show internal consistencies of .70 at Time 1 and .73 at Time 2. The test-retest reliability was .60. Career withdrawal cognitions were measured using a three-item scale parallel to the job withdrawal cognitions scale, but using the word “field” as a referent-instead of “job.” The three linearly summed items were: (1) thinking about leaving the banking field: (2) intention to look for a different field of employment; and (3) intention to leave the banking field. Item responses were recorded on a five-point scale (1 = very unlikely, 5 = very likely). Previous work by Blau (1985a, 1988) has shown such a career withdrawal cognitions scale to have adequate reliability and to demonstrate sufficient discriminant validity from a job withdrawal cognitions scale. For example, Blau (1985a) found that the career withdrawal cognitions scale had internal consistencies of .67 at Time 1 and .71 at

94

GARY

BLAU

Time 2, with a 57 test-retest reliability. The correlation between job withdrawal cognitions and career withdrawal cognitions was .40 at Time 1 and .45 at Time 2. Thus although related, job withdrawal cognitions and career withdrawal cognitions are distinguishable concepts. Social desirability is defined as the degree to which individuals describe themselves in favorable, socially desirable terms generally in order to achieve the approval of others (Crowne & Marlowe, 1964). Methodologically, organizational researchers (e.g., Arnold, Feldman, & Purbhoo, 1985; Ganster, Hennessey, & Luthans, 1983) have concluded that it is important to control for individual social desirability response bias in studies involving self-reports of work attitudes, motivation, performance, and withdrawal cognitions. For example, if such bias is present, individuals will underreport their negative work feelings and overreport their positive work feelings. Arnold et al. (1985) argue that general findings in the turnover literature may be weakened due to such bias, and that social desirability response bias should be controlled for when investigating work attitude-work outcome relationships. The 33-item measure of social desirability developed by Crowne and Marlowe (1964) was used. Sample items include: “I always try to practice what I preach,” “I am always willing to admit it when I make a mistake,” and “I am always courteous, even to people who are disagreeable.” Following Crowne and Marlowe (1964), answers to items were recorded either true or false, and one point was given to each response in the socially desirable direction. Previous organizational researchers (e.g., Arnold et al., 1985; Ganster et al., 1983) have found the Crowne and Marlowe (1964) scale to be reliable. For example, Arnold et al. (1985) found an internal consistency of .81, while for Ganster et al.‘s (1983) study it was .80. Turnover was measured by recording the number of bank tellers who left the bank for voluntary reasons (e.g., not including dismissal) within 18 months of the first survey administration. The l&month time period allowed for a sufficient subsample of leavers to accumulate. A total of 61 tellers voluntarily left the bank within the l&month time period (39 tellers left within the last 12 months, after the second survey administration). Stayers (N = 113) were coded 1 and leavers (N = 61) were coded 2. To identify whether leavers were simply changing jobs within the banking field or changing jobs and vocations, the bank made available to the author leaver home phone numbers so that postexit phone interviews could be attempted. Contacted leavers were asked why they left the bank, if anything could have been done to prevent their leaving, and to indicate what their new job was. These interviews were open-ended and content analyzed afterward. Of the 61 attempted phone interviews, the author was eventually able to get 50 leavers (82%) to participate. Repeated calls were made until contact with each leaver was established. Eleven leavers refused to

CAREER

COMMITMENT

AND

EMPLOYEE

TURNOVER

95

participate. Of the 50 phone interview participants, 29 (58%) indicated that they had changed jobs (e.g., to a new teller position at a different bank) within the banking field. These individuals will be called job leavers. The 21 (42%) remaining interviewees said that they had changed to jobs (e.g., secretarial, administrative positions) outside of the banking field. These individuals will be called career leavers. The primary reason given by job leavers for changing jobs was for better promotional opportunities (beyond the teller position) at other banks. For career leavers the main reasons for leaving were desire for a less stressful job and chance to make more money. Analysis Following Blau’s earlier work (1985a, 1988), factor analysis was used to determine whether the career commitment measure could be distinguished empirically from measures of job involvement and organizational commitment. Factor analysis has been previously used to test the discriminant validity of other measures, for example, the dimensions of job involvement (Blau, 1985b) and job characteristics versus job satisfaction (Ferratt, Dunham, & Pierce, 1981). To show discriminant validity, career commitment items should load on a separate factor versus the job involvement and organizational commitment items. A principal components analysis was initially performed on the 25 career commitment, job involvement, and organizational commitment items to determine the number of factors. To further test for convergent and discriminant validity, the relationship of career commitment to job and career withdrawal cognitions scales was compared against the relationships of job involvement and organizational commitment to job and career withdrawal cognitions scales. Job involvement and organizational commitment have previously shown a significant negative correlation to job withdrawal cognitions (e.g., Blau, 1985a; Miller et al., 1979). However, career commitment would not necessarily be significantly related to job withdrawal cognitions since career commitment has a broader referent (i.e., industry/vocation) than a specific job. It would be expected, though, that individuals with higher levels of career commitment should show less career withdrawal cognitions. In their model of career change, Rhodes and Doering (1983) distinguish job withdrawal cognitions from career withdrawal cognitions. Hierarchical regression analysis (Nie, Hull, Jenkins, Steinbrenner, & Bent, 1975) will be used to examine whether career withdrawal cognitions can predict a unique portion of employee turnover beyond job withdrawal cognitions. After examining the component relationship correlations, hierarchical regression analysis will also be used to test the predicted career commitment-career withdrawal cognitions-turnover relationship.

96

GARY BLAU

Using longitudinal data to examine the relationships between work attitudes and withdrawal cognitions removes some of the method variance inherent in a cross-sectional self-report methodology (Bateman & Strasser, 1984). In addition, the temporal ordering of the work attitude (T1) and withdrawal cognition (T2) variables is more representative of the presumed causal direction (Mobley et al., 1979). Finally, to control for the influence of social desirability on work attitude-outcome relationships, partial correlations were calculated. RESULTS A scree test (Cattell, 1966) performed on the principal components analysis indicated that a three-factor solution was th most appropriate and accounted for 54% of the total variance. Varim x and oblique rotations of the data were tested to see which woul 1 yield the clearest factor structure. The results indicated that a three-factor solution with varimax rotation was the “cleanest” and most interpretable. Table 1 presents the factor analysis. A minimum factor loading of .30 (Nunnally, 1978) was used as a guideline for considering an item to be part of a factor. An inspection of Table 1 shows that there were minimal double loading or weak loading complications for items. These results provide evidence that career commitment is operationally distinguishable from job involvement and organizational commitment using the present sample. Having a subjects to items ratio of at least 5:l helps to increase confidence in the stability of this factor solution (Nunnally, 1978). Based upon this factor analysis, a seven-item measure of career commitment was constructed and uSed in subsequent data analysis. Descriptive statistics, reliabilities, and intercorrelations among study variables are shown in Table 2. The descriptive statistics indicate that restriction of range is not a problem. All of the multi-item variables have at least adequate (above .70) internal consistency reliabilities (Nunnally, 1978). Correlation results with Time 1 variables show that social desirability is significantly positively related to organizational commitment, and that job involvement, organizational commitment, and career commitment are significantly positively related to each other. These results agree with previous research (e.g., Arnold et al., 1985; Blau, 1985a, 1988). Although career commitment is significantly related (p < .Ol) to job involvement and organizational commitment, the amount of shared variance is approximately 10% with job involvement and 8% with organizational commitment, indicating that career commitment is minimally redundant with these work attitudes. Morrow (1983) has suggested that work referent attitudes such as career commitment, job involvement, and organizational commitment be regarded as independent constructs,

CAREER COMMITMENT

AND EMPLOYEE

97

TURNOVER

TABLE 1 Factor Loadings of Career Commitment, Job Involvement, Commitment Items

and Organizational Factor

Item content Career Commitment I. Would go into a different industry if paid the sameY 2. Want career in this industry 3. If could do it all over would not choose this field” 4. If had all the money needed, still work in this field 5. Like vocation too well to give it up 6. Ideal vocation for a life work 7. Disappointed that ever entered this industry” Organizational commitment 1. Talk up organization as a great place to work 2. Would accept almost any assignment to stay with organization 3. Find that personal values are similar to organization’s values 4. Proud to tell others that part of organization 5. Organization inspires one’s best performance 6. Glad chose this organization over others to work for 7. Really care about fate of organization 8. This is best organization to work for 9. Willing to put in extra effort for organization Job involvement I. Most important things that happen involve job 2. Job small part of self 3. Very personally involved in job 4. Live, eat, and breathe job 5. Most interests centered around job 6. Have very strong ties to job 7. Most of life goals are job-orientd 8. Consider job as central to existence 9. Like to be absorbed in job Eigenvalues Percentage variance Note. (N = 174). L?Reverse-scored item.

I

2

3

- .Ol .11

.02 .20

.25

- .03

.12 .29 .28

.02 .20 .21 .09 .Ol .27

.17 .I0 .26 .09 .zI! .26 .lO -.02 .12 8.4 34%

.09

.18

.Ol

.I4

.20

-30

.29 .03 -32

.Ol .14 .08

.I3

.I6

A

- .02 .I3 -.06 - .05 .24 .22 .25 .24 2.7 11%

( 2.1 9%

7.2 6.5 1.4

5. Job withdrawal cognitions (T2) 6. Career withdrawal cognitions (Tz) 7. Turnover

3.3 2.8 .6

5.8 5.7 6.4 4.9

SD

3-15 3-12 l-2

lo-27 11-38 IO-43 9-31

Range (.76Jh .08 .16* .05 -.08 .02 .04

1

-.24** - .lO -.15

(.80) .34** .32**

2

TABLE 2 and Intercorrelations

y At Time 1 (T,), N = 174; at Time 2 (Tz) N = 133. ’ Internal consistency estimates are given in parentheses; NA, not available. * p < .05. ** p < .Ol.

19.1 22.1 29.0 21.5

M

1. Social desirability (T,) 2. Job involvement (T,) 3. Organizational commitment (T,) 4. Career commitment (T,)

Variable

Descriptive Statistics, Reliabilities,

-.36** -.ll -.22*

(.W) .28**

3

-.14 -.33** -.18*

(.82)

4

among Study Variables

t.911 .42** .36**

5

C.84) .27**

6

WA)

7

F s

cl %

Suggest Documents