OC age n tenure

  • Upload
    mudita

  • View
    217

  • Download
    0

Embed Size (px)

Citation preview

  • 8/6/2019 OC age n tenure

    1/17

    BASIC A ND A PPLIED SOCIAL PSYCHOLOGY , 1993, 14(2), 143-159Copyright 1993, Lawrence Erlbaum Associates, Inc.

    Age an d Tenure in Relation toOrganizational Commitment:A Meta-AnalysisAaron CohenDepartment of Political ScienceUniversity of Haifa

    Age and tenure are considered important correlates of organizational com-mitment (OC). However, the relations between these variables and OC werefound to be relatively weak. This meta-analysis examines the relationsbetween age and tenure and OC across different time frames of employmentstages. The numbers of samples dealing with the relations between OC andage and between OC and tenure were 84and 80, respectively. Age and tenurewere divided into time frames of employment stages and a separate meta-analysis was conducted for each of these subgroups. The findings indicatedifferent patterns of relations across employment stages. The relation betweenOC and age was strongest for the youngest subgroup. The relation betweenOC and tenure was strongest for the oldest tenure subgroup. These findingsare discussed in terms of the practical and conceptual implications of thedifferences between age and tenure in their relations with OC and with a viewto future research.

    Th e issue of o rganizational commitment (OC) continues to receive attentionboth from scientists and practitioners. This interest is evidenced by nu-mero us studies examining th e relations between O C an d its antecedents an doutcomes (for reviews, see Griffin & Bateman, 1986; M atthieu & Zajac,1990). On e of th e m ain reasons fo r the interest in O C is its relation t oimportant organizational outcomes such as turnover (Porter, Steers,Mowday, & Boulian, 1974), turnove r intentions (Shore an d M artin , 1989),performance (Meyer, Paunonen, Gellatly, Goffin, & Jackson, 1989), andabsenteeism (Farrell & Peterson, 1984).

    Requests for reprints should be sent to Aaron Cohen, Department of Political Science.University of Haifa, Mount Camel, Haifa 31 999, Israel.

  • 8/6/2019 OC age n tenure

    2/17

    144 COHENTwo of the important antecedents of OC are age and tenure, primarilybecause they are considered as the main indicators of side-bets (Becker,

    1%0), a term that has been used to refer to the accumulation of investmentsvalued by the individual which would be lost if he or she were to leave theorganization. However, recent literature, based on quantitative summariesof findings (Cohen & Lowenberg, 1990; Matthieu & Zajac, 1990), arguesthat, in general, the relations of OC with and with tenure have producedfew large correlations. Because the overall magnitude of the relationsbetween OC and age and OC and tenure was found to be relatively small,researchers began to question the importance of age and tenure as deter-minants of OC (Cohen & Lowenberg, 1990;Meyer & Men, 1984).

    A limitation of studies on these relati~nss that they have relied on linearcorrelation analysis. Another limitation is that both age and tenure areconsidered in theOC literature as time-reIated variables representing similareffects and processes in their relations with OC. This article attempts toovercome these limitations by examining whether age and tenure, whencontrolled by the employment stage, differ in their relations with OC.Answering this question may increase our understanding of OC develop-ment. If the relations of OC with apte and tenure differ across employmentstages, then the importance of incrwing levels of OC should be treateddifferently dependiagon he specific employment stage. This understandingshould enable the &sjdy of more effective ar#anizational cmr-planningprograms as well as provide mamgms with more accurate explanations andpredictions about their subordinates' and their own behavior on the job.

    THEORETICAL FRAMEWORKAge and Tenure as Determinants of OCTwo views of OC have dominated the literature (Griffin& Bateman, 1986).The first sees commitment as affective or attitudinal and hias been called theorganizational behavior or psyckoiogy approach. This view sees the indi-vidual as identifying with the organization and, therefore, committed tomaintaining membrswp to pursue his or her goals. This treatment ofcommitment perhaps has its most frequently cited origins in the work ofPorter and his associates (e.g., Mowday, Porter, & Steers, 1982; Porter etal., 1974) and has been termed clffectitive commitment (Meyer&Allen, 1984)and value commitment (Angle & Perry, 1981).The second view has its basis in Becker's ( 1 M ) side-bet theory (Griffin&Bateman, 1986). The essential element of this approach is the argument thatprevious events have certain cost or forfeiture implications for the presentand therefore place restraints on a person's options. Becker argued that,

  • 8/6/2019 OC age n tenure

    3/17

    over a period of time, certain costs accrue that m ake it more difficult for theperson t o disengage from a consistent line of activity, namely, maintainingmembership in the organization. Working in the organization increases anindividual's investments; hence, the costs of leaving can lead t o higher levelsof OC.Ritzer a nd Trice (1969) reasoned tha t side-bets will accumulate over time.For instance, through increased tenure the employee gains seniority andconnections within the organization (Becker, 1960). The argument is that anemployee's investment of time and effort, for example, makes it moredifficult to leave the organization; the more time with the o rganization, themore side-bets o r investments. Therefore, age and tenure were suggested bythem and others (Alutto, Hrebiniak, & Alonso, 1973; Meyer & Allen, 1984;Sheldon, 1971) as the best indicators of actions that build up one's stake inthe organization.Research findings, however, have demonstrated weak relations betweenOC and age and tenure. An explanation of these weak relations wasproposed by Meyer and Allen (1984). They argued tha t younger employeesmay be more committed because of their awareness that, with less workexperience, they often have fewer job opportunities. As they gain experi-ence alternate employment opportunities may increase; this decreases themagnitude of one important cost of leaving, that of having no job. Meyerand Allen's explanation is supported by Mowday et al. (1982) and Rusbultand Farrell (1983) who emphasized the importance of the availability ofattractive alternative job opportunities. Unavailability should prevent somenegative attitudes toward the job and organization tha t would be possible ifmany opportunities were available. The Meyer and Allen explanationsuggests that levels of OC vary across different age groups as a result offactors such as alternative job opportunities, and therefore age and tenuremay affect OC differently across different age groups. Accordingly, thereason fo r the low correlations of OC with age and tenu re may be tha t theserelations are not simple or d irect, but moderated by employment stages.It appears that the degree of O C attributable to age and tenure cannot beinterpreted, to any large extent, through the side-bet model. The reasoningof the investment model may lead to the argument that the relationsbetween age and tenure and OC should not be examined with simplecorrelations. This article argues that for those employees who have been ina job for only 1 month, 1 year, or 2 years, a meaningful test of age andtenure as determinants of OC cannot be conducted by charting a simplelinear relation. Therefore, if these demographic variables are to be used asdeterminants of OC, then the appropriate way to test their effect is tocontrol empirically those subjects for whom the t rue investment character-istics associated with the side-bet model have had a chance to operate. Weneed to exclude from consideration those individuals who are so early in

  • 8/6/2019 OC age n tenure

    4/17

    146 COHENtheir employment history that sigaificant investments have not yet beenmade. The dotemiltants of OC and the procccss through which commitmentdevelops and is maintained can vary at different employment stages in acareer. This aspect U s o be taken into account in examining the OC-ageand 06-tenure relations.Another pertinent question is whether age and tenure, when control4a-i byemployment stagas, have different patterns in their effect on OC. Thisquestion is apgpropriate because of the findings of M a n , Ferris, andKacmar (1992) who claimed that age and tenure affect job satisfactiondifferently. Bedian et al. argued that although age and tenure are bothtime-related variables which covary with one another, there may be twodifferent models that exphin their effects on job satisfaction. The first isthe career-staipe model which was suggabd as an appropriate model forexplaining the effect of age. The second is the experiencemodel which waspresented as a better repr~sentative f the effect of tenure. A limitation ofthe OC literature is that it employs both models without any attempt toseparate them conceptually as canslructs that represent different prmeses.Both age and tenure were shown as representing time-related variables thatshould have the same effect on OC. Foilowing Eedian et al., it can beargued that age represents the career-stwe model whereas tenure representsthe experience model.The OC-Age Relation Across Career StagesAccording to career developnzent models (Levinson, Darrow, Klein,Levinson, & MeKee, 1978; Super, 1957) the development and maintenanceof OC will differ across career stages as represented by age. In the earlycareer stage, lev& of OC vary dp oa an individtd's opportunitiesand the availability of attractive alternatives (Mowday et al., 1982; Rusbult& Farrel, 1983). Those in the early career stage face the contradictory tasksof making commitments and k m n g options open. An individual in theearly career stage attempts to becomeestablbhed in a job that interests heror him, but if this job proves inappropriate, she or he does not hesitate tochoose another. Employees at this stage express greater inteation to leavetheir organization and more willimcss to relocate than those in other agegroups (Orstein, Cron, & fiiocum, 19g8; Omnstcin& Isabelta, 1990). Basedon this, age is expected to be an mportant determinant of OC n the earlycareer stage. The experienccrs and oppwtunities of wnployges at this s t aewill vary signifiaantfy and will result indifferentOC levels depending on thenumber of alternative empfoyment opportunities. Such variations, repre-sented by the employee' a$e, will have a strong effect on OC of employeesat this stage.The career models of Super (1957) and Levinson et al. (1978) hold that

  • 8/6/2019 OC age n tenure

    5/17

    the propensity to leave an em ployer and chosen field of work decreases asone moves into the middle and later stages of employment. People in themidcareer stage are m ore interested in developing stable work and personallives and in making strong commitments to work, family, and the commu-nity. People in the late stage of their career are in a stage of relativetranquility. They are more oriented to "settling down" and are less willingto relocate or leave the organization for purposes of promotion. Mowday etal. (1982) argued that levels of commitment developed during the earlyemployment period appear to remain stable. Moreover, increased invest-ments in the form of time and energy make it increasingly difficult foremployees to leave their jobs voluntarily. The relative stability of commit-ment levels along with increased difficulties in leaving the organizationweaken the relation between age and OC in the mid and late career stages.Thus, on e would predict a weaker relation between age and O C in the midand late career stages than in the early career stage because very littlevariation in levels of O C a re expected in th e later career stages.OC-Tenure Relations Across Employment StagesIn addition t o the common side-bet model, the experience model, as relatedto the effect of tenure on OC, seems to be well represented by Reicher's(1986) OC development theory. Reicher's model provides a good explana-tion of the effect of tenure on OC across employment stages by arguing thatthe antecedents of commitment can be roughly classified into three catego-ries: (a) psychological (expectations, challenge, conflict), (b) behavioral(volitional, irrevocable acts), an d (c) structural (tenure in the organization,accumulated investmentdsunk costs). She suggested that each class ofantecedent variables may be primarily associated with early, mid, or lateorganizational commitments. At the early career stage, psychologicalattachments to the organization may be the primary antecedents ofcommitment. Over time, the individual engages in various acts tha t bind heror him t o the organization and lead to commitment.In later career stages, not only do psychological and behavioral linkagesoperate to produce comm itment on the part of the individual, but structuralvariables such as investments and lack of opportunity elsewhere maycombine to cement the individual's attachment t o the organization. Reichers(1986) concluded tha t the management o f comm itment among newcomersin organizations may be most problematic because behavioral and struc-tural bonds had not yet had a chance to develop fully. Hence, in the earlystage, personal psychological variables and not tenure explain a largeproportion of variation in OC. During the mid and late career stages,structural and personal considerations increase the effect of tenure on OC .All this evidence points to the need to examine whether the OC-age and

  • 8/6/2019 OC age n tenure

    6/17

    OC-tenure relations differ as a function of empbyment stage. The purposeof this meta-analysis is to reexmine these relations while controlw fordifferent employment stages. The expectations are that the OC-age relationwill be stronger in earlier career stages than in later stages and theOC-tenure relation will be stronger in the later stagps than in the earlierstages. Recent developaents in meta-analysis (Hunter & Schmidt, 1%Hunter, Schmidt, & Jackson, 1982) have rn& it possible to reexamineexisting studies using qusrntitrttive review methods. Such methods permit thestatistical aggregation of research findings and the systematic asswimant ofinterstudy moderators. Quantitative effects and samples can be cumulatedand, consequently, commonalities beyond the scope of narrative reviewscan be brought to light. Given the lack of empirical research examining theOC-age and OC-tenure relations while controlling for emplo;ynmonts ~ w ,a meta-analytic review appears to be concepttidy and praeticdy apprupri-ate. A meta-analytic review can increase our inskht into the researchquestion raised by Griffin and Bateman (1986) of whether differenttheoretical mechanisms and models become activated are are more valid atdifferent stages of the commitment process.

    METHODThe Hunter et al. (1982) and Hunter and Schmidt (1990) meta-andyticprocedure was used in this study because it aggregates correlation cocffi-cients across studies, corrects for the presence of stadstical artifacts, andprovides unbiased estimates of the thearetical population relations. Fol-lowing this method, this study consists of three basic steps: the e s th t i onof population mean correlation and variance, the correction for stadsticalartifacts, and the analysis of moderating effects.

    Published studies were identified bath by a2ulual and computer-mistedsearches of the social science, psychology, and miana-ial literature (from1967 to 1991). The total numbers of sannples that included correl~btionaldata dealing with the OC-qp and OC-tenure relations were 84 and 80,respectively. (The complete Iist of studks used in this research and for eachage and tenure subgroup is available from the author.)Meta-AnalysisSeveral additional points need to be emphasized to further clarify themeta-analysis that was conducted. This analysis controlied for the statisticalartifacts of sampling error and of predictor and criteria unreliability, butbecause of insufficient data made no corrections for range restrictions. Inaccordance with recent arguments in the literature (e.g., McDaniel, Hirsh,

  • 8/6/2019 OC age n tenure

    7/17

    ORGANIZATIONAL COMMITMENT 149Schmidt, Raju, & Hunter, 1986) regarding the rules for rejecting thesituational specificity hypothesis, it was decided not only to follow thePearlman, Schmidt, and Hunter (1980) rule of 7970, but also to adoptMcDaniel et al.3 (1986) recommendation that the actual amount ofvariance remaining after accounting for sample size should be consideredwhen determining generalizability. The chi-square test suggested by Hunteret al. (1982) was used only as supporting evidence. Additionally, it wasdecided to adopt Schmidt, Hunter, and Raju's (1988) suggestion thatconfidence intervals should be used to interpret validity generalizationresults.Based on Hunter et al.'s (1982) finding that the average correlation doesnot violate the independence assumption, it was also decided that, if morethan one correlation was reported for a single sample, the average of thesecorrelations would be used.

    Moderator effect. In the moderator analysis, a moderating effectwould be indicated if: (a) the average correlation varied from subgroup tosubgroup, and (b) the corrected variance average was lower in the sub-groups than for the data set as a whole (Hunter et al., 1982, p. 105).

    Definition of age and tenure time-frames. Employee age was di-vided into four time-frame subgroups: (a) up to 30 years, (b) 31 to 35 years,(c) 36 to 40 years, and (d) older than 40 years. Similar to previousoperationalizations (Gould, 1979; Hall & Mansfield, 1975; Rush, Peacock,&Milkovich, 1980; Slocum& Cron, 1985), the first subgroup represents thetrial or the exploration career stage, the second represents the establishmentor settling-down stage, the third represents the advancement stage, and thefourth represents the maintenance stage. Organizational tenure was dividedinto three subgroups: (a) 1 to 4 years, (b) 5 to 8 years, and (c) 9 years ormore. Again, similar to previous operationalizations (Gould & Hawkins,1978; Mount, 1984; Sturnpf & Rabinowitz, 1981), the first tenure subgrouprepresents the exploration or trial stage; the second, the establishment stage;and the third, the maintenance stage. Samples where age and tenure of therespondents were not reported were treated as missing values in themoderator analysis.

    One may argue that the findings could be affected by the number of yearswithin each of the age or tenure subgroups-specifically, that higheraverage corrected correlation might be expected in the age or tenuresubgroups with a greater number of years (e.g., older-than-40 category forage or 9-or-more year category for tenure). Two arguments can be providedin response to that contention. First, the rationale for the number of yearswithin each category is based on career developments theories (Gould &Hawkins, 1978; Levinson et al., 1978; Ornstein, et al., 1989; Stumpf &

  • 8/6/2019 OC age n tenure

    8/17

    150 COHENRabinowitz, 1981; Super, 1957). These theories argue, with ernpiri4support, that changes in behavior across age or tenure are not due tochronological changes but that different behaviors or attitudes can beexpected in different age or tenure subgsoups. In all these works, thenumber of years within each age or tenure category was determined sohlyon conceptual arguments and as a result age or tenure categories includemore years than others. These studies do not mention any effect of thenumber of years in age or tenure categories on the results.

    Second, the number of years cannot have any effect on the size of thecorrelation in meta-analytic research. The decision about which career stageeach of the samples of the analysis would belong in was based on theaverage age of the total simple was reported in each of the studies. Usingaverage age and/or tenure as a criterion for partitioningsamples is the mostcommon criterion in meta-analyses examining age or tenure (e.g., Cohn,1991; Raz & Raz, 1990, Schuerger, Zarrella, & Hotz, 1989). Samples weregrouped to this or another career-stage category based on the average age ofthe total sample. The correlations used for the meta-analysis are the totalsample correlations and not the correlations in one or another age or tenurecategory. Therefore, the size of the correlation in any srunple included herecould not be affected by the number of years in the age or tenure categorybecause the correlations are not b e d on category correlations. In sum-mary, there is no conceptual or mathematical reason to expect that thenumber of years in any age or tenure category will affect the size of thecorrelations.

    Measures of commitment. A methodological variable controlled forin this studywas the measure of OC.ControUing for the moderating effwtsof different OC measures enables compreh@risive ontrol of aEImethodoi~g-ical aspects of the OC-turnover relation and glSO prmits compaisons withthe two previous rneta-andyses (Cohen & Lowenberg, 1990; Nathieu &Zajac, 1990). Three types of measures were controlled for in this study.Following Randall (1990), it was decided to differentiate between thecommon Organizational Cor nmi ant Q u e s t i o d e (OCQ;orter et al.,1974) and other attitudiial measures. in addition, following M&thieua dZajac, it was decided to control for side-bet calculative measures b a d anRitzer and Trice's (1969) scale.

    RESULTSResults of the meta-analysis for age and tenure for the total sample,controlling for employment stages, are prasentd in Tables 1 and 2.Results demonstrate that the OC-age relation is higher in the early age

  • 8/6/2019 OC age n tenure

    9/17

    TABLE 1Meta-Analytic Results for the OC-Age Relation Controlled by Career Stages

    Observed ResidualAge k N r , CI VAR VAR % x2-All samples 84 36,482 .17 .I9 -.03 to .40 .0127 .0105 17 493.43.Up to 30 years 7 1,207 .22 .24 -.20 to .67 .0471 .0418 11 62.75*31 to 35 years 18 9,224 .14 .I5 -.08 to .39 .0147 .0128 13 141.56*36 to 40 years 21 11,655 .I7 .18 .00 to .36 .0086 .0068 20 105.63.Older than

    40 years 12 7,509 .19 .20 .06 to .35 .0060 .0043 27 47.53,Note. k the number of samples in each analysis; N = the total number of individuals

    in the k amples; r = the mean weighted uncorrected correlation; r, = the mean weightedcorrelation corrected for attenuation; CI = 95% confidence interval for r,; observed VAR =variance of the uncorrected correlations; residual VAR = observed VAR corrected forstatistical artifacts; qo = the percentage of variance across samples attributed to statisticalartifacts; and 2 = a chi-square test for variance remaining unaccounted for.

    *p < .001.TABLE 2Meta-Analytic Results for the OC-Tenure Relation Controlled by CareerStages

    Observed ResidualTenure k N r , CI VAR VAR % X2All samples 80 36,877 .09 .10 -.I0 to .29 .0105 .0084 20 393.43**Up to 4 years 8 5,325 .00 .00 - 10 to .10 .0038 .0023 39 20.26*5 to 8 years 23 10,256 .09 .09 - 10 to .28 .0102 .008O 22 106.32**9 years or more 15 9,231 .13 .14 -.07 to .34 .0109 .0093 15 103.96**

    Note. k the number of samples in each analysis; N = the total number of individualsin the k amples; r = the mean weighted uncorrected correlation; r, = the mean weightedcorrelation corrected for attenuation; CI = 95% confidence interval for r,; observed VAR =variance of the uncorrected correlations; residual VAR = observed VAR corrected forstatistical artifacts; % = the percentage of variance across samples attributed to statisticalartifacts; and X2 = a chi-square test for variance remaining unaccounted for.

    *p < .01. **p < .001.

    group (up to 30 years, r = .24) than for the total and for the other careerstages. The pattern of relations across career stages is not consistent.Whereas the correlation is highest in the exploration or trial career stage, itis lowest in the settling-down or establishment career stage (31 to 35 years,r = .15) and rises again in each of subsequent stage.

    The pattern of OC-tenure relations across career stages is consistent butopposite to that for age. There is a consistency in the effects in that there isno relation at the exploration or trial career stage (up to 4 years, r = .00),but rises gradually for each of the subsequent stages (r = .09, advancementstage; r = .14, establishment stage).

  • 8/6/2019 OC age n tenure

    10/17

    152 COHENTABLE 3Meta-Analytlc Results for the OC-Age and OC-Tenure Rslattons Controtled bythe Type of OC Measure

    Observed ResidualVariabie k N r r, CI VAR VAR % x2AgeAll OC measures 84 36,482 .17 .19 - .03 o .40Porter et al. (1974)OCQ measure 53 26,402 .18 .20 .O1 o .39

    Ritzer & Trice (1969)side-bet scalemeasure 9 1,803 .16 .18 -.08 to .43Other OC measures 22 7.910 .13 .15 -.I5 to .44TenureA11 OC measures 80 36,877 .09 .I0 - .10 o .29Porter et al. (1974)OCQ measure 48 26,175 .09 .10 - .08 o .27Ritzer & Trice (1969)side-bet scale

    measure 8 1.435 .18 .19 -Other OC measures 24 8,759 .07 .07 - .17 o .32Note. k = the number of smwks in each analyak N = the totat number of individualsin the k samples; r = the mean weighted uncorrkted correlation; r, = the mean wei@tedcorrelation corrected for attenuation; C1 = 95% confidence interval for r,; obsmred VAR =variance of the uncorrected correlations; residual VA R = observed VAR corrected forstatistical artifacts; Oh = the pcrcentw of variance across samples attributed to statisticalartifacts; and xf = a chi-square test for variance remain@ unaccounted for.* p < .001.Table 3 presents theOC-we and OC-tenure r-ow controlling for thetype of OC measurement. For Me, no corrshleralde dfferences were foundbetween the totald he subgroups of mwaires, For tenure, however,

    there was an nterest& difference between the t d nd the subwoups ofmeasures. The OC-tenure regation was much higher using the Kitzer andTrice (1969) side-bet measure ( r = .19) than for the total sample (r = .lo),for the Porter et al. (1974) attitudinal measure (r = .lo), or for dhermeasures (r = .07).Mareover, the muks obtdnad for the Eetzer and Tricemeasure were very consistent, as d em a ns tf d by the fact that the entirevariance across the sample correlations (i.e., I-) was attributed tostatistical artifacts.

    DISCUSSION"The goal of research in any area is the production of an integratedstatement of the findings of the many pieces of research done in that area"

  • 8/6/2019 OC age n tenure

    11/17

    ORGANIZATIONAL COMMITMENT 153(Hunter et al., 1982, p. 162). In the case of the OC-age and OC-tenurerelations, the need today is not necessarily for additional empirical data, butfor some means of making sense of the vast amounts of data that havealready been gathered. The main purpose of this study was to examine theserelations controlling for time frames of employment stages. In general, theresults have demonstrated that the relations of age and tenure to OC varyacross employment stages. The OC-age relation was strongest for theearliest employment stage, whereas the OC-tenure relation was strongestfor the latest stage.

    Based on these findings, it can be suggested that age and tenure asdeterminants of OC offer two different processes regarding the develop-ment of OC. Whereas the OC-tenure relation represents the job experiencemodel and is affected more by job and organizational factors, the OC-agerelation represents the career model and is also affected by external factors.The main argument is that age as an OC determinant is exposed to factorsexternal to the organization and the work itself (i.e., having to do with one'slife and attitudes beyond those in a very specific work situation such as theorganization). Tenure, on the other hand, is a variable focused more onevents and factors in the specific job and organization. Tenure cannot haveany effect in the early employment stages because investments in theorganization have not been accumulated. In the late employment stages,investments have been accumulated and evaluated by employees. Thoseemployees that perceive them as sufficient will stay, and those that perceivethem as insufficient will leave. That is, OC-tenure relations can developonly after the employee has spent some years in the organization and,hence, develops investments, evaluates them, and decides, based on theexchange relation, whether or not to commit himself or herself to theorganization.

    This explanation may also clarify the strong relation between tenure andthe Ritzer and Trice (1%9) calculative measure of OC. The calculativemeasure is more organizationally focused. It questions the respondents onthe likelihood of their leaving the organization given various levels ofinducements in pay, status, responsibility, job freedom, and opportunityfor promotion. That is, it concentrates only on pure organizationalrewards.

    It is worth noting the difference between my findings and those of Bedianet al. (1992). Their conclusion was that tenure is a more stable predictor ofjob satisfaction than chronological age. In this study, however, both ageand tenure were found to have a stronger effect on the relations of age andtenure to OCwhen controlling for employment stages. The findings of thetwo studies showed that OC and job satisfaction are related differently toage and tenure and demonstrated the differences between OC and jobsatisfaction. Conceptually, commitment is considered to be a different

  • 8/6/2019 OC age n tenure

    12/17

    construct than job satisfaction. "Commitment as a construct is more global,reflecting a general affective response to the organization as a whole. Jobsatisfaction, o n the other hand, reflects one's response either to one's job orto certain aspects of one's job" (Mowday et al., 1982, p. 28). Therefore,whereas job satisfaction can be rdated to variables associated with theimm ediate work environment, such as organizational tenure, OCappears tobe related to variables outside the immediate work situation as well. Jobsatisfaction is affected mainly by the job experience model as dem onstratedby tenure, whereas OC, as a broader concept, is affected by both the jobexperience and career models. Jo b satisfaction and OC hould be perceivedas separate concepts, each having its unique relation to different anttxed-ents. Fu ture research needs to reexamine the relations between OC and jobsatisfaction with age and tenure based on this study's finding and those ofBedian et al.Results of this study lead to several conclusions having implications forfuture research. The first implication has to do with the effects of themeasurement of commitment on the results. Controlling for commitmentmeasure shows that with respect to age, there was no meaninsful andconsistent difference between the results based on Porter et al.3 (1974)OCQ measure and those based o n R i t ~ e r nd Trice's (1969) side-bet measureor on the other affective measures. With respect to tenure, however,differences emerged between the OC attitudinal measure and the side-betmeasure in which a higher correlation was found with the side-bet mertsure.This finding partly supports Mathieu and Zajac's (1!2%) argument thatcalculative commitment is more strongly rda ted to age and tenure th an isattitudinid com mitment because it reflects in its measurement the nature ofaccumulation of investments as demonstrated by variables such as ageandtenure. In tha t regard, a question can be raised concerning the usefuIn.e6~ fcom bining studies using different measures of com mitment. More researchis needed to examine the effect of type of measurement o n results. It seems,however, that future meta-analyses on OC antecedents and outcomesshould control for the type of OC measurement to interpret the findingsmore accurately.The second recommendation is that more research is needed to examineage and tenure simultaneously in their relations to valued work a ttitudesand outcomes. Although most of the research so far hiis examined eitherage or tenure separately in their relation to work outcome vari&b1es, thefindings of this study as well as those of Bedian et al. (1992) havedemonstrated the importance of examining the effects of both thesevariables on work outcomes. Biedian et al. argued that a s nd tenure aretheoretically interesting variables for which, unfortunately, questions ofhow and why we might expect them to be related to impartant outcomeshave been neglected. They suggest& that such theoretical and m ethodolog-

  • 8/6/2019 OC age n tenure

    13/17

    ORGAN IZATIONAL COMMITMENT 155ical advances concerning these two variables should prove fruitful bothscientifically and practically.The third recommendation relates to the fact that a large percentage ofvariance across sam ples attributed to statistical artifacts in the main-effectsfindings and in the moderator-analyses findings remains unexplained. Inmeta-analytic terms, this means that the remaining unexplained variance inthe correlations of age and tenure with O C could be due to a rtifacts, suchas variations in the quality of OC measurement, or it could indicate thatthere are other moderators, some of which may also have strong effects(Hunter & Schmidt, 1990). There are probably other moderators affectingthese relations. Research should continue, but with some shift in focus.More theoretical developments should direct empirical research and quan -titative summaries of previous findings to assist in detecting other moder-ators that affect OC linkages with age and tenure. Much of the meta-analytic work in the industrial psychology literature has concentrated onmain-effect analyses rather than moderator effects. As Guzzo, Jackson,and Katzell (1987) pointed out, "one of the most exciting promises ofmeta-analysis is that it allows the reviewer to determine the effects ofmoderators th at have never been examined in a n original empirical study"(p. 414). Only through further examination of moderator effects willmeta-analytic research achieve its potential fo r stimulating the developmentof new theories and testing the adequacy of existing theories. Shoemaker,Snizek and Brayant (1977) suggested, for example, that "a fruitfu l avenuefo r further research had best take in to account differences in the types andlevels of employee studied" (pp. 602-603).Another set of recommendations deals with the conceptual implicationsof this study for fu ture research on OC-age and O C-tenure relations. Thefindings of this study showed tha t, even when controlling for em ploymentstages and the type of OC measurement, the magnitude of the correlationsof age and tenure with OC was not large. This finding suggests severalalternative directions for futu re research.The first would be to accept one of the arguments of Meyer and Allen(1984) who stated that "the instrument used in tests of the side-bets theorymay not be measuring commitment as Becker conceptualized it" (p. 377).The implication of this conclusion is that "in order to test the validity of theside-bet theory, however, a commitment measure must be used that iscongruent with Becker's conceptualization" (Meyer & Allen, 1984, p. 377).Meyer and Allen constructed a Continuous Commitment Scale which theythink measures more accurately what Becker had intended than do theaffective measures that other researchers have used to test his theory.McGee and Ford (1987) reexamined the scales of Meyer and Allen andconcluded that one of the subscales of the Continuance CommitmentMeasure "appears to more closely parallel the side-bet view of commitment,

  • 8/6/2019 OC age n tenure

    14/17

    156 COHENas described originally by Becker" (p. 640).To wee with the Meyer andAllen and the McGee and Ford conclusions would be to agree that futureresearch needs to examine the side-bet theory using the same strategy asbefore (i.e., evaluating the relations of age and tenure to an OC measure),but with a more appropriate "Wker-type" m a u r e of continuance06.A second direction would be to accept the other EAeyer and Allen (1984)argument that perhaps the strategy used to examine the side-bet theory wasinappropriate: "Using age and tenure as indexes s ~ s l sess appropriate thandirectly obtaining employees' perceptions of the size and importance ofinvestments they have made. This type of strategy would be consistent withBecker's theory" (pp. 337-338). Along this line of thought, one could gofurther and conclude that the strategy of examining Beoker's theory bmedon the relations of age and tenure with OC measures was inappropriate nomatter which OC measures were used. It is not sido-bet indices such as ageand tenure which are meaningful but rather the "individuals' perceptionsregarding the number and the magnitude of the side bets that they made"(Meyer& Allen, 1984, p. 378). Based on this argument, we need a differentstrategy for examining the thwry. Meyer and Allen thmselves sthe works of Farreil and Rusbult (1981) and Rusbult and Fwrell(lM3) ascontaining a more appropriate strategy for examining the side-bet theory inthe future. The practical implication of either of the precwhg conclusionsis that we need a new body of empirical research before we can reach anyconclusions regarding the side-bet theory. Because the two conclusionsrequire a new body of empirical research, only future studies wiU provetheir validity.

    A third direction is to follow the W a n et al. (1992) argument that it isquite common in behavioral sciences to overentphwka the proportion ofvariance explained to the exclusion of other issua. In the "softer," wilderareas of psychology (including organizational), the practical valw of aresult should not be based on the magnitude of the associated effect.Establishing the shape or functional ridation between key variables shouldbe emphasized more because the measures of exp1tlined variance may bemisleading or inappropriate indicators of a finding's importance. Becausethe intent is usually to explain or understand, maximizing variance ac-counted for is not a primary concern.On a final note, despite some of the limitations of this study, itdemonstrates that more attention should be paid to moderating effms inthe relations between OC and age and tenure. Future rasearch shouldfurther explore the explanations offered in this study. Another avenue forresearch could be a longitudinal study of OC-qe and W-tenure relationsacross employment stages, cornwing their effects at different st-.Although this study has shown that these relations vary with empbymentstage, future research should try to provide a more solid conceptual

  • 8/6/2019 OC age n tenure

    15/17

    ORGANIZATIONAL COMMITMENT 157framework for this moderating effect. This study dem onstrates the need todevelop such a conceptual framework and apply some of the careerdevelopm ent (Levinson et al., 1978; Super, 1957) and OC models (Mowdayet al., 1982; Reichers, 1986). The arguments proposed in this study may wellbe a good starting point for such research.

    ACKNOWLEDGMENTSI thank two anonymous reviewers for their helpful comments and sugges-tions. I also thank Karen Dustan and Pat Elemants for their editorialassistance.

    REFERENCESAlluto, J. A., Hrebiniak, L. G., & Alonso, R. C. (1973). On operationalizing the concept of

    commitment. Social Forces, 51. 448-454.Angle, H. L., & Perry, J. L. (1981). Organizational commitment and organizational

    effectiveness: An empirical assessment. Administrative Science Quarterly, 26, 1-14.Becker, H. S. (1%0). Notes on the concept of commitment. American Journal of Sociology,

    66, 32-40.Bedian, A. G., Ferris, G. R., & Kacmar, K. M. (1992). Age, tenure, and job satisfaction: A

    tale of two perspectives. Journal of Vocational Behavior, 40, 33-48.Cohen, A., & Lowenberg, G. (1990). A reexamination of the side-bet theory as applied to

    organizational commitment: A meta-analysis. Human Relations, 43, 1015-1050.Cohn, L. D. (1991). Sex differences in the course of personality development: A meta-analysis.Psychological Bulletin, 109, 252-266.Farrell, D., & Petersen, J. C. (1984). Commitment, absenteeism and turnover of new

    employees: A longitudinal study. Human R elations, 37, 681-692.Farrell, D., & Rusbult, C. E. (1981). Exchange variables as predictors of job satisfaction, job

    commitment and turnover: The impact of rewards, costs, alternatives and investments.Organizational Behavior and Human Performance, 27, 78-95.Gould. S. (1979). An equity-exchange model of organizational involvement. Academy ofManagement Review, 4, 53-62.Gould, S., & Hawkins, B. (1978). Organizational career stage as a moderator of the

    satisfaction-performance relationship. Academy of Management Journal, 21. 434-450.Griffin. R . W ., & Bateman, T. S. (1986). Job satisfaction and organizational commitment. In

    C. L. Cooper & I. Robertson (Eds.), International review o f industrial and organizationalpsychology @p. 157-188). New York: Wiley.Guzzo, R., Jackson, S. E., & Katzell, R. A. (1987). Meta-analysis analysis. In L.L. Cummings& B. M. Staw (Eds.), Research in Organizational Behavior (Vol. 9, pp. 407-442).Greenwich, CT: JAI.

    Hall, D. T., & Mansfield, R. (1975). Relationships of age and seriority with career variablesof engineers and scientists. Journal of Applied Psychology, 60, 201-210.

    Hunter, J. E., & Schmidt, E. L. (1990). Me thods of meta-analysis: Correcting error and biasin research find ings. Beverly Hills, CA: Sage.Hunter. J. E., Schmidt, F. L., & Jackson, G. B. (1982). Meta-analysb: Cum ulating researchfindings across studies. Beverly Hills, CA: Sage.

  • 8/6/2019 OC age n tenure

    16/17

    Levinson, D. J., Darrow, C. N., Klein, E. B., Levinson, M. H., & McKee, B. (1978). Theseasons of a man's life. New York: Knopf.Mathieu, J. E., & Zajac, D. M. (1990). A review and meta-analysis of the astafedonts,correlates and consequences of organizational commitment. Psy&lrlogical BdlllrSln, l&?,171-194.McDaniel, M. A., Hirsh , H. R., Schmidt, F. L., Raju, N. S., & Hunter, J. E. (1986).Interpreting the results of meta-analytic research: A comment on Schrnitt, Gooding, Noeand Kirsh (1984). Personnel Psychology, 39, 141-148.McGee, G. W., & Ford, R. C. (1987). Two (or more) dimensions of organizationalcommitment: Reexamination of the affective and continuance commitment scales. Journalof Applied Psychology, 72, 638-642.Meyer, J. P., &Allen, J . N. (1984). Testing the side ba heory of orgmim tiana l commitment:Some methodological considerations. JourMII of Applied psVcholo~,69, 372-378.Meyer, J . P., Paunonen, S. V., Gellatly, I. R., Goffin, R. D., & Jackson, D. N. (1W).Organizational commitment and job performance: It's the nature of the commitment thatcounts. Journal of Applied Psychology, 74, 152-1 56.Mount, M. K. (1984). Managerial career stage and facets of job satisfaction. Journal ofVocational Behavior, 24, 348-354.Mowday, R . T., Porter, L. M., &Steers, R. M . (1982). Employee-organizationallinkage.NewYork: Academic.Omstein, S., Cron, W . L., & Slocum, J . W. (1989). Life stage versus career stage: Acomparative test of the theories of Levinson and Super. Journal of OrganizationalBehavior, 10, 117-133.Omstein, S., & Isabella, L. (1990). Age vs. stage models of career attitudes of women: Apartial replication and extension. Journal of Vocational Behavior, 36, 1-19.Pearlman, K., Schmidt, F. L., & Hunter, J. E. (1980). Validity generaWation results for testsused to predict training success and job proficiency in claical occupations. Journal ofApplied Psychology, 65, 373-406.Porter, L. W., Steers, R. M., Mowday, R. T., & Boulian, P. V. (1974). Organizatianalcomm itment, job satisfaction and turnover among psychiatric technicians. Journal ofApplied Psychology, 59, 603-609.Randall, D. M. (1990). The consequences of organizational commitment: Methodologicalinvestigation. Journal of Organizational Ilehavior, 11, 361 37%.Raz, S., & Raz, N. (1990). Stmctural brain abnormalities in the ma jor psyohoscs: Aquantitative review of the evidence from computorizd imaging. P $ ~ W o g t m IBulfMin,108, 93-108.Reichers, A. E. (1986). Conflict and organizational commitment. Journal of AppliedP S Y C ~ O ~ O ~ V .1, 508-514.Ritzer, G., & Trice, H. M. (1969). An empirical study of Howard Becker's side-bet theory.Social Forces, 47 , 475-479.Rusbult, C. E., & Farrell, D. (1983). A longitudinal test of the invtstm mt model: The impacton job satisfaction, job commitment and turnover of vatktions in rewards, costs,alternatives and investments. Journal of A ~ q ~ l k iSychOlo#, 68,429438.Rush, J. C., Peacock, A. C., & Miikovich, G . T. (19W). Carm stages: A partial test ofLevinson's model of life/career stages. Journal of Vocational Behavior, 16, 347-359.Schmidt, F. L., Hunter, J . E., & Raju, N. S. (1988). Validity g mc rdim tion and situationalspecificity: A second look at the 75% mle and the Fisher's z t r ans fomt ion . Journal ofApplied Psychology, 73, 665-672.Schuerger, M. J., Zarrella, K. L., & Hotz, A. S. (1989). Factors that influence the temporalstablity of personality by questionnaire. Journal of P& rs o ~ i i tynd Soda!PsychoIogv, 56,777-783.

  • 8/6/2019 OC age n tenure

    17/17

    ORGANIZATIONAL COMMITMENT 159Sheldon, M. E. (1971). lnvestment and involvement as mechanisms producing commitment to

    the organization. Administrative Science Quarterly, 16, 143-150.Shoemaker,D. J., Snizek,W. E., & Brayant, C. D. (1977). Towards a further clarification ofBecker's side-bet hypothesis as applied to organizational and occupational commitment.Social Forces, 56, 598-603.Shore, L. M., & Martin, H. J. (1989). Job satisfaction and organizational commitment inrelation to work performance and turnover intentions. Human Relations, 42, 625-638.Slocum, J. W., & Cron, W. L. (1985). Job attitudes and performance during three careerstages. Journal of Vocational Behavior, 26, 126-145.Stumpf, S. A., & Rabinowitz, S. (1981). Career stage as a moderator of performancerelationships with facets of job satisfaction and role perceptions. Journal of VocationalBehavior, 18, 202-2 18.Super, D. (1957). The psychology of careers. New York: Harper.