21
* Academy of Management Journal 1997, Vol. 40, No, S, 1208-1ZZ7. INTEGRATING JUSTICE CONSTRUCTS INTO THE TURNOVER PROCESS: A TEST OF A REFERENT COGNITIONS MODEL KARL AQUINO RODGER W. GRIFFETH DAVID G. ALLEN Georgia State University PETER W. HOM Arizona State University In the model tested, we posit that employees' outcome and supervisor satisfaction result froni referent outcomes, justifications, and the like- tihood of amelioration. These satisfaction facets are then related to turnover through withdrawal cognitions. Structural equation modeling was used to assess the fit ofthis model and alternalives. Although the original model had a good fit, a competing model provided a hetler one, with most hypothesized relationships supported. The implications of accepting the alternative model over the original are discussed. It has been suggested that turnover research is in a "fallow period" and in need of rejuvenation (O'Reilly, 1991). Indeed, several reviews and com- mentaries have indicated that most researchers have sought either to address methodological issues or to empirically validate existing theories of with- drawal that focus on affect-induced quitting (Lee & Mitchell, 1994; O'Reilly, 1991). For these reasons, the turnover literature would benefit greatly from the introduction of alternative theoretical perspectives that take into account contextual variables surrounding the occurrence of quitting (Lee, 1996). The present study offers such a perspective. We developed and tested a model clarifying the psychological processes by which felt deprivation in- stigates quitting. The model draws upon referent cognitions theory (RCT; Cropanzano & Folger, 1989: Folger, 1987; Folger, Rosenfield, & Rheaume! 1983; Folger, Rosenfield, & Robinson, 1983) to explain voluntary turnover. We greatly appreciate the excellent and helpful comments of the three anonymous review- ers. Particularly appreciated are the comments of one reviewer who helped to clarify the study's contributions. This study was funded in part by a grant to the first author from the Research Program Committee of the College of Business Administration of Georgia State University. 1208

Integrating Justice Constructs Into The

Embed Size (px)

DESCRIPTION

Paper

Citation preview

Page 1: Integrating Justice Constructs Into The

* Academy of Management Journal1997, Vol. 40, No, S, 1208-1ZZ7.

INTEGRATING JUSTICE CONSTRUCTS INTO THETURNOVER PROCESS: A TEST OF A REFERENT

COGNITIONS MODEL

KARL AQUINORODGER W. GRIFFETH

DAVID G. ALLENGeorgia State University

PETER W. HOMArizona State University

In the model tested, we posit that employees' outcome and supervisorsatisfaction result froni referent outcomes, justifications, and the like-tihood of amelioration. These satisfaction facets are then related toturnover through withdrawal cognitions. Structural equation modelingwas used to assess the fit ofthis model and alternalives. Although theoriginal model had a good fit, a competing model provided a hetler one,with most hypothesized relationships supported. The implications ofaccepting the alternative model over the original are discussed.

It has been suggested that turnover research is in a "fallow period" andin need of rejuvenation (O'Reilly, 1991). Indeed, several reviews and com-mentaries have indicated that most researchers have sought either to addressmethodological issues or to empirically validate existing theories of with-drawal that focus on affect-induced quitting (Lee & Mitchell, 1994; O'Reilly,1991). For these reasons, the turnover literature would benefit greatly fromthe introduction of alternative theoretical perspectives that take into accountcontextual variables surrounding the occurrence of quitting (Lee, 1996).

The present study offers such a perspective. We developed and tested amodel clarifying the psychological processes by which felt deprivation in-stigates quitting. The model draws upon referent cognitions theory (RCT;Cropanzano & Folger, 1989: Folger, 1987; Folger, Rosenfield, & Rheaume!1983; Folger, Rosenfield, & Robinson, 1983) to explain voluntary turnover.

We greatly appreciate the excellent and helpful comments of the three anonymous review-ers. Particularly appreciated are the comments of one reviewer who helped to clarify the study'scontributions. This study was funded in part by a grant to the first author from the ResearchProgram Committee of the College of Business Administration of Georgia State University.

1208

Page 2: Integrating Justice Constructs Into The

1997 Aquino. Griffeth. Allen, and Hom 1209

an employee's voluntary withdrawal from an organization. Extant turnoverresearch has not applied the full referent cognitions theory model. In con-trast, we draw from that theory to provide a comprehensive theoretical ac-count of how justice perceptions prompt employee withdrawal, thus unitingtwo research streams that have seldom been combined.

The foundation for our study is in part prior research examining rela-tionships between outcome fairness and intentions to quit (turnover inten-tions; Konovsky & Cropanzano. 1991). But this study is one of the few, to ourknowledge, that directly examine the relationship between justice percep-tions and actual quitting. The proposed model integrates referent cognitionstheory with the turnover literature by including predicted relationshipsamong outcome and supervisor satisfaction, withdrawal cognitions, and ac-tual turnover.

THEORETICAL BACKGROUND

Organizational Justice and Turnover

The concept of distributive justice is commonly invoked to explain theconditions under which people become dissatisfied with outcomes (Adams,1965; Walster, Walster, & Berscheid, 1978). According to current theories,fairness judgments are made when people compare what they have re-ceived—their outcomes—with those of a referent other (Martin. 1981]. Thiscomparison process underlies relative deprivation, or the feeling of discon-tent arising from a belief that one is getting less than one deserves relative toa comparison other (Croshy, 1984; Martin, 1981). Felt deprivation producesa range of psychological and behavioral effects in organizations, includingdissatisfaction, stress, and absenteeism [Martin, 1981). Despite the far-reaching significance of felt deprivation, surprisingly little research has ex-amined how it relates to turnover.

In general, turnover theorists have conceived "rudimentary" notions ofequity (cf. Mobley, 1977) by assuming that distributive injustice in rewardsunderlies dissatisfaction and, ultimately, organizational exits. More re-cently, scholars have recognized that unfair procedures (Greenberg, 1987;Lind & Tyler, 1988; Thihaut & Walker, 1975) can also stimulate exits (Dit-trich & Carrell, 1979; Price & Mueller, 1986).

Although acknowledging procedural fairness (Hom & Griffeth, 1995;Konovsky & Cropanzano, 1991; Price & Mueller, 1986), turnover perspec-tives remain relatively simplistic and incomplete in light of contemporarytheoretical developments in the justice literature. For example, in present-day turnover models, only undifferentiated effects of procedural justice onjob attitudes are envisioned (cf. Hom & Griffeth, 1995; Price & Mueller,1986), and other essential bases of workplace fairness, like the interpersonalcontext surrounding the enactment of rules for awarding rewards, are ne-glected (Greenberg, 1990; Tyler & Bies, 1990). Known as interactional justice(Bies & Moag, 1986), this construct represents a person's evaluation of the

Page 3: Integrating Justice Constructs Into The

1210 Academy of Management foumal October

truthfulness, impartiality, and respect an authority has shown to him or her.One advantage of referent cognitions theory over previous formulations isthat it accounts for how different types of justice perceptions might relate toturnover.

Referent Cognitions Theory

Referent cognitions theory introduces the evaluation of procedures intoa relative deprivation framework and more completely describes how injus-tice prompts dissatisfaction. It extends traditional turnover models that im-plicate outcome or procedural unfairness as undermining morale and reten-tion (cf. Hom & Griffeth, 1995; Hulin, Rosnowski, & Hachiya, 1985; Price &Mueller, 1986) hy elaborating the psychological processes that generate dis-satisfaction. Specifically, referent cognitions theory explains how dissatis-faction arises when a person compares existing reality to a more favorablealternative.

According to the theory, people perform three mental simulations in-volving referent cognitions, justifications, and the likelihood of ameliora-tion. Referent cognitions are alternative, imaginable circumstances that dif-fer from a person's actual circumstances. People are most likely to be dis-satisfied when imagined results are more attractive than existing reality.They become aware of alternatives by discovering that others are receivingrewards different from their own. As referent outcomes are compared toexisting outcomes, people think about "what might have been."

Just as referent outcomes are compared to existing outcomes, so too arethe procedures that produced those outcomes. The critical question under-lying such comparisons is the extent to which referent procedures are morejustifiable tban those that produced existing outcomes. If actual proceduresare judged to be morally inferior to referent procedures, then there will below justification for existing outcomes. Conversely, superior existing proce-dures will be associated with high justification. If the rationale for an exist-ing procedure is judged less appropriate or convincing than that for thereferent procedure, dissatisfaction results. When the rationale is consideredappropriate, convincing, and hence justifiable, dissatisfaction with presentoutcomes can diminish (Greenberg, 1987; Folger & Martin, 1986; Foiger,Rosenfield, & Robinson, 1983).

A third tenet of referent cognitions theory is that people may viewexisting outcomes as temporary because satisfaction may be influenced bywhat they expect to receive in tbe future. Mental simulations involvingfuture states are presented in tbe model as beliefs regarding tbe likelihood ofamelioration. Likelihood of amelioration occurs because, when people ex-pect outcomes to improve, they are less dissatisfied than wben they see littlechance for improvement in their circumstances (Folger, Rosenfield,Rheaume, & Martin, 1983). That likelihood of amelioration will be related todissatisfaction parallels Martin's (1981) argument that people's responses tofelt deprivation are influenced by their beliefs about whether their organ-

Page 4: Integrating Justice Constructs Into The

1997 Aquino, Griffeth, Allen, and Ham 1211

izations are amenable to change. If they believe that the organizations canchange, then inferior outcomes may not necessarily produce dissatisfaction.Instead, they may motivate constructive attempts at improvement (Martin,1981). But if employees do not hold this belief, poor outcomes can producenegative responses directed inward, such as stress and depression, or nega-tive responses directed outward—absenteeism, poor performance, and res-ignations.

Thus far, we have used referent cognitions theory to explain how com-parisons between existing and imagined outcomes and procedures, as wellas the likelihood of amelioration, are related to dissatisfaction and turnover.To pursue tbis argument and lay the groundwork for hypothesized mediatedrelationships between referent cognitions constructs and turnover, we nextconsider bow employee dissatisfaction with work outcomes and supervisorsincreases actual resignations.

Outcome and Supervisor Satisfaction as Predictors of Turnover

We theorize tbat perceptions of distributive and procedural justice af-fect employees' satisfaction with both outcomes and supervisors. Althoughseveral studies have examined tbe relationship between justice perceptionsand facets of job satisfaction {e.g., Alexander & Ruderman, 1987; McFarlin &Sweeney, 1992), few have investigated how the processes underlying rela-tive deprivation culminate in turnover (cf. Hom & Griffetb, 1995). This studythus contributes to tbe turnover literature by integrating referent cognitionstheory and withdrawal by empirically verifying those links. Using tbe theoryas a framework, our model, presented in Figure la, shows how outcome andsupervisor satisfaction mediate referent cognitions theory constructs andturnover.

Our model specifies tbat referent cognitions, justification, and the like-libood of amelioration influence outcome and supervisor satisfaction. Thesefacets of job satisfaction have been documented as turnover determinations(Hom & Griffetb, 1995). Indeed, several leading perspectives on withdrawalhave implicated reward satisfaction (e.g., Rusbult & Farrell, 1983) and lead-ership treatment (e.g., Graen & Ginsburgb, 1977; Graen, Liden, & Hoel, 1982)as bases of retention. Our model further interposes withdrawal cognitionsbetween satisfaction facets and turnover. The sequence is consistent withFindings of recent literature reviews (Hom & Griffetb, 1995; Tett & Meyer,1993) showing tbat witbdrawal cognitions, a combination of quitting inten-tions and cognitions, is the most immediate precursor to turnover. Thus, lowjob satisfaction should increase resignations by stimulating withdrawal cog-nitions.

We predict tbat referent outcomes will affect outcome satisfaction butnot supervisor satisfaction. Tbe effect is sucb tbat high referent cognitionswill decrease outcome satisfaction. This prediction is based on the argumentthat referent cognitions underlie judgments of distributive ratber than pro-cedural fairness (Folger & Konovsky, 1989; Konovsky, Folger, & Cropanzano,

Page 5: Integrating Justice Constructs Into The

1212 Academy of Management journal October

FIGURE 1

Hypothesized Referent Cognitions Models ofEmployee Turnover

(la) Theoretical Model

ReferentOutcomes

Likelihood ofAmelioration

ProceduralJustification

InterpersonalJustification

J OutcomeSatisfaction

SupervisorSatisfaction

WithdrawalCognitions

Turnover

(lb) More Constrained Alternative Model

ReferentOutcomes

Likelihood ofAmelioration

ProceduralJustification

InterpersonalJustification

OutcomeSatisfaction

SupervisorSatisfaction

WithdrawalCognitions

Turnover

(lc) Less Constrained Alternative Model

ReferentOutcomes

Likelihood ofAmelioration

ProceduralJustification

InterpersonalJustification

WithdrawalCognitions

Turnover

1987). Consequently, they should more strongly predict satisfaction withtangible outcomes like pay or promotions than interpersonal outcomes likesatisfactory treatment by a supervisor. After all superiors are not entirelyresponsible for their subordinates' pay and promotional opportunities.

Page 6: Integrating Justice Constructs Into The

3997 Aquino. Griffeth, Allen, and Hom 1213

which frequently depend on decisions ahout internal and external job worthand the availability of positions outside the work unit.

With regard to tbe justification construct, our model distinguishes be-tween two components of procedural justice. One component represents thestructural characteristics of formal procedures; the second is based on thetreatment a person receives from an authority, like a supervisor. We hypoth-esize that a favorahle evaluation of formal procedures will promote satisfac-tion with outcomes but not satisfaction with a supervisor. Alternatively,favorable justifications provided by interpersonal treatment will foster su-pervisor satisfaction, but not outcome satisfaction.

Our hypothesis relating procedural justification to outcome satisfac-tion is supported by research showing that fair procedures can minimizethe dissatisfaction resulting from poor outcomes (Folger. Rosenfield, Grove,& Corkran, 1979; Greenberg & Folger, 1983). The hypothesis relating in-terpersonal justification to supervisor satisfaction is consistent with Biesand Shapiro's (1987) argument that judgments of interpersonal fairnessmore strongly affect a person's evaluation of a decision maker or author-ity than the person's evaluation of a particular job outcome. The morefavorable the evaluation, tlie more satisfied the person will be with thatauthority.

Lastly, our model depicts likelihood of amelioration as increasing bothoutcome and supervisor satisfaction. This prediction is based on the work ofseveral writers (e.g., Croshy, 1984; Homans, 1974) who have contended thatfairness perceptions depend on both the evaluation of present outcomes andthe expectation of future outcomes. For example, a study by Bernstein andCrosby (1980) found that feelings of deprivation increased as the perceivedfeasibility of obtaining a desired outcome decreased.

Our proposition, that future outcome expectations promote satisfactionwith current job outcomes and with superiors who may authorize such im-provements, extends prevailing thinking about relative deprivation (cf.Davies, 1969; Gurr, 1970). This reasoning conforms to Mobley, Griffeth,Hand, and Meglino's (1979) view that expected future attainment of superiorwork circumstances can bond an employee to an employer, even thoughpresent conditions are deficient. Hence, when employees expect the qualityof work-related outcomes to improve, they hecome less distraught over theirpresent circumstances. In sum. we posited the following relationshipsamong constructs;

Hypothesis 1. High referent outcomes are negatively asso-ciated with outcome satisfaction but are not related tosupervisor satisfaction.

Hypothesis 2. Procedural justification is positively asso-ciated with outcome satisfaction but is not related to sat-isfaction with a supervisor. Interpersonal justification ispositively associated with satisfaction with a supervisorbut is not related to outcome satisfaction.

Page 7: Integrating Justice Constructs Into The

1214 Academy of Management Journal October

Hypothesis 3. Likelihood of amelioration is positively as-sociated with both outcome and supervisor satisfaction.

Hypothesis 4. Both outcome and supervisor satisfactionare negatively related to withdrawal cognitions.

Hypothesis 5. Withdrawal cognitions is positively relatedto turnover.

We used structural equation modeling to assess how well our concep-tual integration of referent cognitions theory constructs with the turnoverprocess accounted for data. Structural equation modeling techniques allowsimultaneous tests of theoretical propositions about relationships amongmultiple variables in a complex model (Bentler, 1990). In addition, a re-searcher can compare the explanatory power of a model to that of competingmodels using an array of adequacy indices. One limitation is that structuralequation modeling cannot firmly establish that a model supported by data isin fact the "true" model because several models may fit the data equallywell. Hence, the correct model cannot be identified on purely statisticalgrounds (MacCallum, Wegener, Uchino, & Leandre, 1993). Nonetheless, theresearcher can reduce uncertainty about a plausible model by testing a prioricompeting models, comparing their meaningfulness, parsimony, and ahilityto account for the data. Eliminating competing rivals strengthens support fora model. Anderson and Gerhing (1988) outlined a procedure for testing al-ternative models using sequential chi-square difference tests. We usednested model comparisons to compare the relative fit of competing modelspositing different structural networks than the original.

Alternative Models

We identified and tested two rival models. One was a more constrainedversion of the original model. Operationally, this meant one pathway wasdeleted from the original model. The other was a less constrained version,meaning a new pathway was added to the original formulation. In the moreconstrained model, we fixed the pathway from likelihood of amelioration tosupervisory satisfaction at zero (see Figure lb). Tbis specification of norelationship between amelioration and supervisory satisfaction is consistentwith prominent theories of relative deprivation (e.g., Davies, 1969; Gurr,1970) that make no predictions about the effects of future expectations ondissatisfaction with authorities. Instead, the theories emphasize relation-ships hetween likelihood of amelioration and outcomes. By including apathway from likelihood of amelioration to satisfaction with supervisor inour original model, we extended these theories by proposing that futureexpectations affect not only satisfaction with tangible job outcomes, but alsowith the authority who may be instrumental for change.

In the third, less constrained model, we introduced a direct effect oflikelihood of amelioration on withdrawal cognitions (see Figure lc). Writ-ings on relative deprivation and turnover converge to support this predic-

Page 8: Integrating Justice Constructs Into The

1997 Aquino. Griffeth, Allen, and Hom 1215

tion. First, Martin (1981) contended tbat turnover and absenteeism are mostlikely to occur wben employees believe organizations are unreceptive tocbange. Similarly, several relative deprivation studies bave suggested thatpessimism about improvement can trigger discontent and anger (Bernstein &Crosby, 1980; Crosby & Gonzalez-Ital, 1984). Relatedly, some turnover the-orists (e.g., Mobley et al., 1979) bave noted that anticipated attainment of amore valued outcome or role can inhibit exits, despite current job dissatis-faction; others (Hulin, 1991; Steers & Mowday, 1981) bave argued that dis-satisfied employees' optimistic beliefs tbat they can take effective action torectify tbeir situation can deter their withdrawal. Accordingly, we comparedthis model predicting that likelihood of amelioration weakens withdrawal tothe original model.

METHODS

Procedures and Respondent Characteristics

A survey was administered to 192 employees of a large northeasternhospital; employees were in small groups, and the survey was given onhospital time. Respondents provided their names on the questionnaires sotbat we could match survey responses to later organizational exits. We guar-anteed the confidentiality of individual survey responses, emphasizing thatstudy participation was voluntary. Usable data on all study variables wereobtained from 150 employees. Men constituted 21 percent and Caucasians95 percent of the respondents. Tbeir average age was 31 years, and theiraverage organizational tenure was 5.1 years. The average number of depen-dents (counting the respondent) was 1.94, and 52 percent were married.Sixty-six percent bad formal education beyond high school, 90 percentworked full-time, and 35 percent were nurses.

Measurement of Model Constructs

Referent outcomes. Six items similar to those used in past referent cog-nitions theory resetu-ch (Folger, 1977) were used to measure referent out-comes. Respondents compared their pay, education, and experience to thoseof others both internal and external to their work groups and organizationusing a five-point Likert-type scale (1 = strongly agree, 5 = strongly disagree).An exemplary item is "The pay rate for my job is higher than the pay rate forthe same job in otber companies in this area." The Likert scale was anchoredin such a way that high ratings on tbe items indicated tbe referent receivedhigher pay than the respondent.

Procedural justification. Three items assessed perceived proceduraljustification for outcomes. Like other procedural justice measures (e.g., Mc-Farlin & Sweeney, 1992; Nieboff& Moorman, 1993), the items asked respon-dents to report how consistently procedures governing work standards, ef-fort expended, and behavior were applied within the organization. A sampleitem is "Some people in my department get by without meeting work stan-

Page 9: Integrating Justice Constructs Into The

1216 Academy of Management Journal October

dards." Items were rated so that a bigh value indicated high proceduraljustification.

Interpersonal justification. Six items measured justification based ontbe personal treatment employees received from tbeir immediate supervi-sors. Consistent with previous measurements (Bies & Shapiro, 1987; Moor-man, 1991), these items asked respondents to rate their supervisors' behaviorwith regard to fairness, impartiality, and consistency. A sample item is "Mysupervisor sees to it that all of us meet work standards." A high scale ratingindicated high interpersonal justification.

Likelihood of amelioration. Tbis construct was measured by askingrespondents to "list five of tbe most important things tbat you consider whenmaking a cboice between jobs" (e.g., interesting work). Respondents thenestimated their chances of attaining these outcomes from their present jobson a scale ranging from 1 (no chance) to 5 (100% chance). These likelihoodratings were averaged. For some respondents, elicited outcomes might cor-respond imperfectly to measured referent outcomes. Although the measureof amelioration was partially detached from referent outcomes, having re-spondents provide an idiosyncratic set of ameliorating outcomes reducedresponse sets that might inflate relationships between this measure and tbesatisfaction indexes. Moreover, the scale format captured job characteristicstbat more strongly impacted satisfaction and withdrawal cognitions.

Outcome satisfaction. Following Mobley and colleagues' (1979) concep-tion of satisfaction as a purely affective reaction to a job, we used six 7-pointbipolar (e.g., 1 = bad, 7 = good) adjective sets to measure satisfaction withpay and promotions. This index corresponds to Scott's semantic measure ofmorale (Scott, 1967; Scott & Rowland, 1970).

Supervisory satisfaction. Three 7-point bipolar adjective sets, similar tothe above scale, measured this attitude.

Withdrawal cognitions. We used Hom and Griffeth's (1991) six-itemvalidated measure of this construct. Two items measured eacb of tbe fol-lowing components: intention to quit, intention to search, and thoughts ofquitting. All items bad a five-point likelibood scale (1 = no chance, 5 = 100%chance).

Turnover. Since we were only concerned about modeling voluntaryturnover, it was essential to differentiate between voluntary and involuntaryexits. This determination was made from confidential exit interviews tbepersonnel department conducted witb departing employees. The variablevoluntary turnover was coded 1 for quitters and 0 for others. Five percent ofthe sample voluntarily left one year after the survey.

Statistical Analyses

Because the sample size was modest, we evaluated manifest rather thanlatent structural models using LISREL 8 (Joreskog & Sorbom, 1993). To re-duce interpretational confounding, we performed a confirmatory factoranalysis (CFA) to verify our items' factor structure (Anderson & Gerbing,1988). The measurement model assessed whether all items in a given scale

Page 10: Integrating Justice Constructs Into The

1997 Aquino, Griffeth. Allen, and Hom 1217

represented the same latent factor. We then aggregated only those itemsreflecting a common construct to derive unidimensional composite scalesfor the structural model tests (Anderson & Gerbing, 1988).

To evaluate the structural model, we followed Bollen's (1990) recom-mendation to interpret multiple indexes of model fit. We reviewed LISRELfit statistics, such as the chi-square test and root-mean-square residual(RMSR). To supplement these indices, we examined the normed fit index(NFI; Bentler & Bonett, 1980) because, unlike the chi-square test, its calcu-lation does not depend on sample size (Bollen, 1990). We also consideredthe goodness-of-fit index (GFI: Joreskog & Sorbom, 1984) and the compara-tive fit index (CFI; Bentler, 1990), both of which are relatively stable insamples smaller than 250 (Hu & Bentler, 1995).

We evaluated the absolute fit of our theoretical model hy followingAnderson and Gerbing's (1988) recommendations. We assessed a sequenceof nested alternative models via sequential chi-square difference tests. First,our theoretical model was compared to a null model that includes no struc-tural relations among the constructs. The theoretical model was then com-pared to (1) a saturated model specifying all unidirectional relations amongconstructs, (2) a more constrained alternative model in which one path wasremoved from the theoretical model, and (3) a less constrained alternativemodel in which an extra pathway was added to the theoretical model.

RESULTS

According to the confirmatory factor analysis, the initial measurementmodel fit the data poorly (x^ = 1,538.43, df= 539, p < .05; RMSR = .11. NFI= .63, GFI = .72). This poor fit was not surprising since measurement modelrespecification is typically required (Anderson & Gerbing, 1988). Respecifi-cation decisions should follow both content and statistical considerations.Deleting problem indicators is the preferred solution for poor fit {Anderson& Gerbing, 1988). On the basis of item content evaluation and inspection ofmodification indexes, we dropped six indicators from the model. Doing thisresulted in a substantially improved fit for the measurement model (x^ =440.80, df= 356, p < .05; RMSR = .07, NFI = .84, GFI = .96). Although thechi-square statistic is still significant, the measurement model was deemedacceptable, given the other supportive indexes (Anderson & Gerbing, 1988).

Table 1 presents the means, standard deviations, reliabilities, and cor-relations among the model components.

Only withdrawal cognitions significantly predicted turnover (r= .24, p< .05), indicating that it was a more proximal determinant of turnover thanreferent cognitions theory constructs or job satisfaction (Hom & Griffeth,1995; Tett & Meyer, 1993). Also, as expected, the referent cognitions con-structs were all significantly related to both outcome and supervisory satis-faction. Tbe measures of procedural and interpersonal justification weremoderately correlated (r = .58, p < .05). Although this pattern is not sur-prising, given that a well-regarded supervisor is likely to use fair procedures,it may indicate a lack of discriminant validity in the measures.

Page 11: Integrating Justice Constructs Into The

O

1218 \ \ Academy of Management Journal October

TABLE 1Means, Standard Deviations, and Correlations*

Variable Mean s.d. 8

1. Referent outcomes 3.33 0.62 (.70) I '•,2. Likelihood of

amelioration 3.24 0.88 -.29* (.74)3. Procedural

justification 3.53 0.89 -.26* .20* (.78)4. Interpersonal

justification 2.90 0.73 -.14 .23* .58* (.80)5. Outcome

satisfaction 3.81 1.61 -.58* .50* .25* .27* (.90)6. Supervisory

satisfaction 4.87 1.55 -.19* .33* .31* .44* .35* (.96)7. Withdrawal

cognitions 2.55 1.02 .20* -.41* -.19* -.33* -.31 -.36* (.88)

8. Turnover 0.05 0.23 .06 -.03 -.09 -.09 -.10 .01 .24*

" Cronbach alpha reliabilities are shown along the diagonal.* p < .05

To test for discriminant validity, we fixed the correlation parameterbetween the two constructs at 1.0 and used a chi-square difference test tocompare the constrained and unconstrained models (Bagozzi & Phillips,1982). The test showed that discriminant validity was achieved betweenprocedural and interpersonal justification (x^d = 70.50, df= 1. p < .001).

According to Folger, Rosenfield, and Rheaume (1983), likelihood ofamelioration exerts an interactive effect on relative deprivation, with feel-ings of deprivation experienced only when there is a high referent outcomeand low likelihood of amelioration. In our model, however, we treated like-lihood of amelioration as a main effect. To test for a possihle interaction, weperformed moderated regression analyses on both satisfaction variables. Nosignificant interactions were found, supporting the predicted effect of like-lihood of amelioration on satisfaction.

After achieving a satisfactory measurement model and ruling out theinteractive effects of likelihood of amelioration, we evaluated the structuralmodel. Our theoretical model fit the data well (x^ = 29.49, df= 14, p < .05;GFI - .96, RMSR = .08, NFI = .90, and CFI = .94). Equally important, thestandardized parameter estimates shown in Figure 2a indicate that all butone of the hypothesized relationships were significant and in the predicteddirection.

Both outcome and supervisor satisfaction were significantly predictedby amelioration. Referent outcomes also decreased outcome satisfaction.With regard to procedural and interpersonal justification, parameter esti-mates showed that interpersonal justification affected satisfaction with su-pervisor. The proposed effect of procedural justification and outcome satis-faction was refuted, however. Consistent with prior research, both outcomesatisfaction and supervisor satisfaction lowered withdrawal cognitions

Page 12: Integrating Justice Constructs Into The

1997 Aquino, Griffeth, Allen, and Hom 1219

FIGURE 2Estimated Referent Cognitions Models of Employee Turnover

{2a) Original Theoretical Model

ReferentOutcomes

Likelihood ofAmelioration

ProceduralJustification

InterpersonalJustification

- . 2 1

•OutcomeSatisfaction

SupervisorSatisfaction

WithdrawalCognitions

.24*

Turnover

(2b) Final Alternative Model

ReferentOutcomes

Likelibood ofAmelioration

ProceduralJustification

InterpersonalJustification

Satisfaction /-.23/

Supervisor /Satisfaction

WithdrawalCognitions

.24*

Turnover

*p < .05

(Graen & Ginsburgh, 1977; Rusbult & Farrell, 1983). Finally, withdrawalcognitions increased resignations, as predicted hy our model. In sum, themodel satisfactorily fit the data, and its parameter estimates upheld nearlyall pathways.

Table 2 shows the results of the nested model comparisons. In step 1, wecompared our theoretical model to the null model. As expected, the theo-retical model showed significant improvement over the null model (x^d =169.13, df= 8, p < .05). In the next step, we compared our theoretical modelto a saturated model. The saturated model showed significant improvementover the theoretical model (x^a = 29.49, df= 14, p < .05). Following Andersonand Gerbing (1988), we then compared our theoretical model with a moreconstrained alternative. Results indicated tbat the theoretical model repro-duced observed covariances more precisely than the more constrained al-ternative (x^^ = 10.80, df=\,p< .05).

Next, we compared our theoretical model with the less constrained

Page 13: Integrating Justice Constructs Into The

1220 Academy of Management foumal

TABLE 2Nested Model Comparisons^

October

Model

1. Null model2. Referent cognitions model of the

turnover processModel 1 versus model 2

3. Saturated mode!Model 2 ver.sus model 3

4. More constrained alternative modelModel 4 versus model 2

5. Less constrained alternative modelModel 2 versus model 5Model 5 versus model 3

x'198.62*

29.49*

0.00

40.29'

17.78^

RMSR

.30

.08

.14

.08

GFI

.74

.96

.94

.97

NFI

.32

.90

.86

.94

CFI

.33

.94

.90

.98

AX^

169.13*

29.49*

10.80*

11.71*17.78''

"* RMSR = root-mean-square residual; GFI = goodness-of-fit index; NFI = normed fit index;CFI = comparative fit index.

*• Values are not significant.*p < .05 '

alternative. This alternative produced a fit superior to that of the theoreticalmodel (x̂ d = 11.71, df = 1, p < .05). Thus, our second alternative modelappeared to explain the data better than tbe theoretical model. Thus, we nextcompared the less constrained alternative model witb the saturated model.The saturated model did not show significant improvement over the lessconstrained alternative model (x^| = 17.78, df= 13, n.s.). Hence, nestedmodel comparisons indicate that this alternative model explains the data aswell as the best-fitting (saturated) model, but more parsimoniously. Thisconclusion is reinforced by the alternative model's high fit indexes (x^ =17.78, df= 13, n.s.; GFI = .97, RMSR = .08, NFI = .94, CFI = .98).

Figure 2b displays standardized parameter estimates of the less con-strained alternative model. Consistent with theoretical expectations, referentoutcomes and likelihood of amelioration predicted outcome satisfaction.Likelihood of amelioration and interpersonal justification also predicted sat-isfaction with supervisor. Withdrawal cognitions remained a significant pre-dictor of turnover. But once again we found that procedural justification didnot significantly predict outcome satisfaction. Tbe additional path in thealternative model was supported, as withdrawal cognitions were signifi-cantly and negatively related to likelihood of amelioration. Finally, althoughsatisfaction with supervisor remained a significant predictor of withdrawalcognitions, these cognitions were no longer significantly predicted by out-come satisfaction.

DISCUSSION

Tbe present study provides encouraging evidence for our integration ofreferent cognitions theory constructs witb turnover. Tbe fit indexes of ourmodel fell within acceptable ranges, and all the hypothesized relationships

Page 14: Integrating Justice Constructs Into The

1997 Aquino, Griffeth, Allen, and Hom 1221

among the constructs but one were supported. Consistent with referent cog-nitions theory predictions, the relationship we found between referent out-comes and outcome satisfaction was negative, and that between interper-sonal justification and supervisory satisfaction was positive. As predicted,likelihood of amelioration also increased both satisfaction facets. Tbe onlyhypothesized relationship not supported was the path relating proceduraljustification and outcome satisfaction.

Despite the good fit obtained for our theoretical model, a less con-strained alternative model fit the data better. Unlike the original formulation,this model contains a direct negative relationsbip between likelibood ofamelioration and withdrawal cognitions. Except for tbis path and tbe non-significant path from outcome satisfaction to withdrawal cognitions, whichis discussed below, parameter estimates of the rival model did not varymaterially from those of the theoretical model.

In both the alternative and original models, procedural justice did notincrease outcome satisfaction. This result is consistent with previous re-search that has shown tliat perceptions of procedural justice are less pivotaldeterminants of pay satisfaction than distributive justice (Folger &Konovsky, 1989; Konovsky et al., 1987; McFarlin & Sweeney, 1992). Takentogether, the findings suggest that distributive and procedural justice exertdifferent effects depending upon the outcome considered. Similarly, themarked contrast between procedural justification's nonsignificant effect andinterpersonal justification's significant impact on supervisory satisfactionsuggests that favorable interpersonal treatment from authorities overridesconcerns employees may have about the structural characteristics of formalprocedures. Another plausible explanation for the nonsignificant effect ofprocedural justice, however, may be tbat it is a more distal cause of otherexogenous variables. For example, perceptions of high procedural justicemay raise the likelihood of amelioration. Unfortunately, tbis bypotbesiscould not be tested in the present study because both procedural justiceperceptions and likelibood of amelioration were measured at the same time.

In the original and unconstrained alternative models, likelihood of ame-lioration affected both satisfaction facets. Tbese relationships may refiectwbat Foiger, Rosenfield, and Rbeaume (1983) referred to as the "replace-ment effect": when a person believes an inadequate outcome has a goodchance of being ameliorated, a positive affect is induced. Our findings ac-cord witb research showing that affective states determine whether peopleare dissatisfied with their jobs more strongly than objective features of tbejobs themselves do (Staw, Bell, & Clausen, 1986). Besides corroborating priorwork, we also discovered that likelihood of amelioration directly discour-aged withdrawal cognitions. The latter finding upholds a longstanding con-tention by Mobley and colleagues (1979) that employees' expectations aboutthe future attainment of desirable roles or role outcomes can promote theirretention.

In our final models, outcome satisfaction did not decrease withdrawalcognitions. This result appears to show that outcome satisfaction does not

Page 15: Integrating Justice Constructs Into The

1222 Academy of Management Journal October

fully mediate the effects of referent cognitions theory constructs and theturnover process as we predicted. Moreover, if we also take into account thatsatisfaction showed virtually no correlation to actual turnover, our argumentfor a mediated process is further weakened. However, we suggest that it isnot necessarily inaccurate to contend that satisfaction is indirectly related toturnover through withdrawal cognitions.

Mediation is customarily thought of in its strongest version, which re-quires that X be significantly related to Y as a precondition for sbowing thata third variable (M) mediates this link (Baron & Kenny, 1986). However, wesuggest that mediation also exists when X is not directly related to Y but isonly related through M. Such mediation was claimed by Wright, who as-serted that goal setting mediated the effects of incentives on performancedespite a nonsignificant relationship between incentives and performance.Moreover, the idea that the variable withdrawal cognitions mediates theinfluence of satisfaction on turnover is consistent with prior empirical evi-dence (Hom & Griffeth, 1991; Tett & Meyer, 1993) and thought about turn-over (Hom & Griffeth, 1991; Mobley, 1977). Finally, it should be recalled thatthis pathway was significant in our original model, which suggests that itmay be premature to reject the hypothesized effect of mediation via out-come satisfaction outright.

A second possihle interpretation of the nonsignificant relationship be-tween outcome satisfaction and withdrawal cognitions is that interpersonalconcerns (supervisor satisfaction) may exert the dominant effect on thesecognitions. In both the original and unconstrained models, supervisor sat-isfaction influenced withdrawal cognitions more strongly than did outcomesatisfaction. These results paralleled our findings regarding the relative ef-fects of interpersonal and procedural justification on the two facets of sat-isfaction. Taken together, these results suggest that the way people aretreated in organizations exerts a powerful effect on their reactions that isindependent of the objective features of the work environment.

Although the data better supported the less constrained alternativemodel, these results should be regarded cautiously. First, the alternativemodel needs to be replicated in a new sample to determine if its structuralrelationships are robust. Second, we omitted other turnover determinantsfrom the model that might have improved its capacity to predict turnover. Ofcourse, failure to represent every conceivahle determinant is a prohlem ofany model test. Third, two of the proposed relationships in the alternativemodel were repudiated. We offered several plausihle explanations for thenull results, but research with different samples and measurements of con-structs is needed to provide more definitive evidence of the validity of theseexplanations.

Finally, it might he argued that the absence of relationships betweensatisfaction facets and turnover is problematic. However, a recent meta-analysis (Hom & Griffeth. 1995) reported tbat pay and promotion satisfactionwere weakly related to turnover (r's = -.04 and -.14, respectively, aftercorrection for measurement errors), with the 95 percent credibility interval

Page 16: Integrating Justice Constructs Into The

1997 Aquino. Griffeth. Allen, and Hom 1223

for both including zero. The relationship between supervisory satisfactionand turnover was also very low (-.10). Moreover, from one-third to overone-half of the bivariate correlations were nonsignificant. Hom and Criffeth(1995) identified several important factors as having contributed to thesenull findings: a low turnover base rate, its dichotomous measurement, pre-dictor utireliabiiity, and, of course, sample size, all of which can influencethe magnitude of correlations between predictors and turnover. Thus, ourfindings are not entirely inconsistent with previous research.

The present research makes several theoretical and practical contribu-tions to the literature. First, it supports Martin's (1981) argument that rela-tive deprivation can increase turnover. Second, it fully describes and pro-vides evidence for the psychological mechanisms by which felt deprivationemerges and activates quitting. Third, it outlines a model for future researchin which the importance of perceived prospects for the amelioration ofpresent circumstances, an overlooked turnover antecedent introduced hyMobley and colleagues (1979), is recognized. One interesting question posedby the alternative model is. what happens when loyal organization membersexpect their situations to improve and those expectations not met? Davies's(1969) theory of rising expectations suggests that such an event can lead tohigh levels of frustration and acts of collective violence. In an organizationalcontext, there may be an increase in deviant behaviors like theft, industrialsabotage, sick-outs, and strikes (Martin, 1981).

A managerial implication is that the present research affirms the pow-erful effect the interpersonal context can have on employee attitudes andbehaviors. The relative strength of the effects of interpersonal and proce-dural justification suggests that the actions of managers rather than the for-mal structure of procedures tend to elicit the strongest responses irom em-ployees. This interpretation coincides with Sheppard, Lewicki. and Mint-on's (1992) argument that people are more likely to hlame people thansystems for poor outcomes. Indeed, Bies (1987) argued that managers canreduce feelings of discontent resulting from poor outcomes or unfair proce-dures by providing the aggrieved with social accounts that appear legitimateand sincere. Thus, our model suggests that poor management of employeeperceptions of interactional justice at the supervisory level may ultimatelyinspire employees to quit. Consequently, for organizations facing economicconstraints or unresponsive formal structures, training and rewarding man-agers for promoting interactional justice may be a cost-effective way to detervoluntary organizational exit.

REFERENCES

Adams. J, S. 1965. Inequity in social exchange. In L. Berkowitz (Ed.). Advances in experimen-tal psychology, vol. 2: 267-299. New York: Academic Press.

Alexander. S.. & Ruderman. M. 1987. The role of procedural and distributive justice in organ-izational behavior. Social Justice Research, 1: 177-198.

Anderson. J. G.. & Gerbing. D. W. 198fl. Stnictural equation modeling in practice: A review andrecommended two-step approach. Psychological Bulletin, 103: 411-423.

Page 17: Integrating Justice Constructs Into The

1224 Academy of Management Journal October

Bagozzi, R. P., & Pbiltips, L. W. 1982. Representing and testing organizational theories; A ho-listic constmai. Administrative Science Quarterly, 27: 459-489.

Baron. R.M., & Kenny, D.A. 1966. The moderator-mediator variable distinction in social psy-chological research: Conceptual, strategic, and statistical considerations. Jaumal of Per-sonality and Social Psychology, 51: 1173-1182.

Bentler, P. M. 1990. Comparative fit indexes in structural models. Psycbological Bulletin, 107:238-246.

Bentler, P. M.. & Bonett, D. G. 1980. Significance tests and goodness of fit in the analyses ofcovariance structures. Psychological Bulletin, 88: 588-606.

Bernstein, M., & Crosby, F. 1980. An empirical examination of relative deprivation theory.Journal of Experimental Social Psychology, 16: 442-456.

Bies, R. J, 1987. The predicament of injustice: The management of moral outrage. In L, L.Cummings & B. M. Staw (Eds.), Research in organizatianal behavior, vol. 9: 289-319.Greenwich, CT: JAI Press.

Bies, R. J., & Moag. J. S. 1986. Interactional justice: Communication criteria of fairness. In R.Lewicki, B. Sheppard, & M. Bazerman (Eds.), Research on negotiation in organizations:43-53. Greenwich, CT: JAI Press.

Bies, R. J., & Sbapiro, D. L. 1987. Interactional fairness judgements: The influence of causalaccounts. Social Justice Besearch, 1: 199-218.

Bollen, K. A. 1990. Overall fit in covariance structural models: Two types of sample size effects.Psychological Bulletin, 107: 256-259.

Cropanzano, R.. & Folger, R. 1989. Referent cognitions and task decision autonomy: Beyondequity theory. Journal of Applied Psychology, 74: 293-299.

Crosby, F. 1994. Relative deprivation in organizational settings. In L. L. Cummings & B. M. Staw(Eds.), Besearch in orgunizationol behavior, vol. 6: 51-94. Creenwicb, CT: JAI Press.

Crosby, F., & Gonzalez-Ital. A. M. 1984. Relative deprivation and equity theories. In R. Folger(Ed.), The sense of injustice: Social psychological perspectives: 141-166. New York:Plenum.

Davies, |. C. 1969. Toward a theory of revolution. American Sociological Beview, 27; 5-19.

Dittrich, J. E., & Carrell, M. R. 1979. Organizational equity perceptions, employee job satisfac-tion, departmental absence and turnover rates. Organizational Behavior and Human Per-formance, 24:29-40.

Folger, R. 1977. Distributive and procedural justice; Combined impact of "voice" and peeropinions on responses to inequity. Journal of Personality and Social Psychology, 35:108-119.

Folger. R. 1987. Reformulating the preconditions of resentment: A referent cognitions model. In]. C. Masters & W. P. Smitb (Eds.), Social comparison, social justice, and relative depri-vation: Theoretical, empirical, and policy perspectives: 183-215. Hillsdale, NJ: Ertbaum.

Folger, R.. & Konovsky, M. A. 1989. Effects of distributive and procedural justice on reactionsto pay raise decisions. Academy of Management Journal, 32: 115-130.

Folger. R.. & Martin, C. 1986. Relative deprivation and referent cognitions: Distributive and

procedural justice effects. Journal of Experimental Social Psychology, 22: 531-546.

Folger. R.. Rosenfield, D.,Crove, J., & Corkran, L. 1979. Effects of "voice" and peer opinions on

responses to inequity. Journal of Personality and Social Psychology, 37: 2253-2261.

Polger, R., Rosenfield, D., & Rheaume, K. 1983. Role-playing effects of likelihood and referentoutcomes on relative deprivation. Bepresentative Research in Social Psychology, 13:2-10.

Folger, R., Rosenfield, D., Rbeaume, K., & Martin. C. 1983, Relative deprivation and referentcognitions. Journal of Experimental Social Psychology, 19: 172-184.

Page 18: Integrating Justice Constructs Into The

1997 Aquino. Griffeth, Allen, and Hom 1225

Folger, R., Rosenfield, D., & Robinson, T. 1983. Relative deprivation and procedural justifica-tions, foumal of Personality and Social Psychology, 45: 268-273.

Graen, G. B.. & Ginsburgh, S. 1977. Job resignation as a function of role orientation and leaderacceptance: A longitudinal investigation of organizational assimilation. OrganizationalBehavior and Human Performance, 19: 1-17.

Graen, G. B., Liden.R.. &Hoel, W. 1982. Role of leadership in the employee withdrawal process.foumal of Applied Psychology, 67: 868-872.

Greenberg, J. 1987. A taxonomy of organizational justice theories. Academy of ManagementReview, 12: 9-22.

Greenberg, J. 1990. Organizational justice: Yesterday, today, and tomorrow, foumal of Man-agement, 16: 399-432.

Greenberg, J.. & Folger, R. 1983. Procedural justice, participation, and the fair process effBct ingroups and organizations. In P. B. Paulus (Ed.). Basic group processes: 235-256. NewYork: Springer-Verlag.

Gurr. T. R. 1970. Why men rebel. Princeton: Princeton University Press.

Horn, P. W., & Griffeth, R. W. 1995. Employee tumover. Cincinnati: South-Western.

Hom, P. W., & Griffeth, R. W. 1991. Structural equations modeling test of a turnover theory:Gross-sectional and longitudinal analyses. Journal of Applied Psychology, 76: 350-366.

Homans, G.L. 1974. Social behavior: Its elementary forms. New York: Harcourt Brace Jo-van ovich.

Hu, L.. & Bentler. P. M. 1995. Evaluating model fit. In R. H. Hoyle (Ed.), Stmctural equationmodeling: Concepts, issues, and applications. Thousand Oaks, GA: Sage.

Hulin, G. L. 1991. Adaptation, persistence, and commitment in organizations. In M, D. Dunnette& L. M. Hough (Eds.), Handbook of industrial and organizational psychology (2nd ed.).vol. 2: 445-505. Palo Alto, GA: Gonsulting Psychologists Press.

HuUn, G. L., Roznowski, M., & Hachiya, D. 1985. Alternative opportunities and withdrawaldecisions: Empirical and theoretical discrepancies and an integration. Psychological Bui'letin, 97: 233-250.

loreskog, K. G.. & Sorbom. D. 1993. LISREL 8: Structural equation modeling with theSIMPLIS command language. Hiilsdale. NJ: Scientific Software.

Jiireskog, K. G., 8E Sorbom, D. 1984. LISREL VI user's guide. Mooresville, IN: Scientific Soft-ware.

Konovsky, M. A., & Gropanzano. R. 1991. Perceived fairness of employee drug testing as apredictor of employee attitudes and job performance, foumal of Applied Psychology,76: 698-707.

Konovsky, M. A.. Folger, R., & Gropanzano. R. 1987. Relative effects of procedural and distribu-tive justice on employee attitudes. Representative Research in Social Psychology, 17:15-25.

Lee, T. W. 1996. Family characteristics and the prediction of voluntary employee turnover.Paper presented at the annual meeting of the Academy of Management, Gincinnati.

Lee. T. W., & Mitchell, T. R. 1994. An alternative approach: The unfolding model of voluntaryturnover. Academy of Management Review, 19: 51-89.

Lind, E. A., & Tyler, T. R. 1988. The social psychology of procedural justice. New York:Plenum Press.

MacGallum. R. G.. Wegener, D. T., Uchino, B. N., & Leandre, R. F. 1993. The problem of equiva-lent models in applications of covariance structure analysis, fournal of Applied Psychol-ogy, 114: 185-199.

Martin, J. 1981. Relative deprivation: A theory of distributive justice for an era of shrinking

Page 19: Integrating Justice Constructs Into The

•̂ 226 Academy of Management Journal October

resources. In L. L. Cummings & B. M. Staw (Eds.), Research in organizational behavior,vol. 3: 53-108. Greenwich, CT: JAI Press.

McFarlin. D. B., & Sweeney, P. D. 1992. Distributive and procedural Justine as predictors ofsati.sfaction with personal and organizational outcomes. Academy of Management Jour-nal, 35: 626-637.

Mobley, W. 1977. Intermediate linkage in the relationship between job satisfaction and em-ployee turnover. Journal of Applied Psychology, 62: 237-240.

Mobley, W. H., Griffetb. R. W.. Hand, H. H., & Meglino. B. M. 1979. Review and conceptualanalysis of the employee turnover process. Psychological Bulletin, 86: 493-522.

Moorman, R. H. 1991. The relationship between organizational justice and organizational citi-zenship behaviors: Do fairness perceptions influence employee citizenship? Journal ofApplied Psychology, 76: 845-855.

Niehoff, B. P., & Moorman, R. H. 1993. Justice as a mediator of the relationship between meth-ods of monitoring and organizational citizenship behavior. Academy of ManagementJournal, 36: 527-556.

O'Reilly, C. A. 1991. Organizational behavior: Where we've been, where we're going. In M.Rosenzweig & L. Porter (Eds.). Annual review of psychology: 427-458. Palo Alto, CA:Annual Reviews.

Price, J. L., & Mueller, C. W. 1986. Absenteeism and turnover of hospital employees. Green-wich. CT: JAI Press,

Rusbult. C. E., & Farrell, D. 1983. A longitudinal te.st of the investment model; The impact of jobsatisfaction, job commitment, and turnover on variations in rewards, costs, alternatives,and investments. Journal of Applied Psychology, 68: 429^38.

Scott, W. E. 1967. The development of semantic differential scales on measures of "morale."Personnel Psychology, 20: 179-198. I

Scott, W. E.. & Rowland. K. 1970. The generality and significance of semantic differential scalesas measures of morale. Organizational Rehavior and Human Performance, 5: 576-591.

Sheppard, B,, Lewicki, R., & Minton, J, 1992. Organizational justice: The search for fairnessin the workplace. Toronto: Lexington Books.

Staw, B. M., Bell, N. E., & Clausen. J. A. 198B. The dispositional approach to job attitudes: Alifetime longitudinal test. Administrative Science Quarterly, 31: 56-77.

Steers, R. M., & Mowday, R. T. 1981. Employee turnover and postdecision accommodationprocesses. In L. L. Cummings & B. M. Staw (Eds.), Research in organizational behavior,vol. 3: 235-281.

Tett. R. P., & Meyer. J. P. 1993. Job satisfaction, organizationa) commitment, turnover Intention,and turnover: Path analyses based on meta-analytic findings. Personnel Psychology, 46:259-293,

Thibaut, J, W,, & Walker. L, 1975. Procedural Justice: A psychological analysis. Hillsdale, NJ:Erlbaum.

Tyler, T, R,, & Bies, R, J, 1990. Beyond formal procedures; The interpersonal context of proce-dural justice. In J. S. Carroll (Ed.). Applied social psychology in business settings: 77-98.Hillsdale, N): Erlbaum.

Walster, E,, Walster. G. W., & Berscheid. E. 1978. Equity: Theory and resetirch. Boston: Allyn& Bacon.

Wright. P. 1989, Test of the mediating role of goals in the incentive-perfonnance relationship.Journal of Applied Psychology, 74: 699-705,

Page 20: Integrating Justice Constructs Into The

1997 Aquino, Griffeth, Allen, and Hom 1227

Karl Aquino is an assistant professor in the Department of Management at Georgia StateUniversity. He received his Ph.D. in organization behavior from Northwestern Univer-sity. His current research interests include organizational justice, negotiation and con-flict resolution, employee victimization, and revenge and forgiveness in the workplace.

Rodger VV. GrifTeth is a professor of management and senior associate of the W. T.Buebe Institute of Personnel and Employment Relations at Georgia State University. Hereceived his doctorate in social-industrial psychology from the University of SouthGarolina. His major research interests include employee turnover, motivation, and jobredesign.

David G. Allen is a doctoral candidate at the W. T. Beebe Institute of Personnel andEmployment Relations at Georgia State University, His current research interests in-clndn organizational justice, employee turnover, and part-time employment.

Peter W. Hom is a professor of management at Arizona State University. lie received hisdoctorate in industrial-organizational psychology from the University of Illinois. Hisresearch interests inchide employee turnover and performance appraisals.

Page 21: Integrating Justice Constructs Into The