1207_ftp Textos Da Wiley B-On

Perceived Purposes ofPerformance Appraisal:Correlates of Individual- andPosition-Focused Purposes onAttitudinal Outcomes

Satoris S. Youngcourt, Pedro I. Leiva, Robert G. Jones

Performance appraisals have traditionally been directed at individuals,serving either an administrative or developmental purpose. They may servea role definition purpose as well. This study sought to identify andmore broadly define the purposes of performance appraisals to include thisrole definition purpose. Furthermore, this study examined purposes ofperformance appraisals as perceived by the role incumbent, as opposedto the stated organizational purposes. The relationships between theseperceived purposes with several attitudinal outcomes, including satisfactionwith the performance appraisal, job satisfaction, affective commitment, androle ambiguity, are reported. Data from 599 retail service employees wereused to test the hypothesized relationships. Results suggested support for amodel consisting of three performance appraisal purposes having differentialrelationships with the outcomes examined, suggesting the purpose of theperformance appraisal may influence ratees’ perceptions of and attitudestoward their jobs.

Over two decades ago, Bernardin and Beatty (1984) identified many interde-pendent purposes of performance appraisal, including to improve the use ofresources and serve as a basis for personnel actions. Similarly, Cleveland and

HUMAN RESOURCE DEVELOPMENT QUARTERLY, vol. 18, no. 3, Fall 2007 © Wiley Periodicals, Inc.Published online in Wiley InterScience (www.interscience.wiley.com) • DOI: 10.1002/hrdq.1207 315

A previous version of this article was presented at the annual meeting of the Academy ofManagement, New Orleans, Louisiana, in August 2004.

hrdq18303.qxd 8/4/07 2:43 PM Page 315

316 Youngcourt, Leiva, Jones

colleagues (Cleveland, Mohammed, Skattebo, & Sin, 2003; Cleveland, Murphy, &Williams, 1989) have shown that in practice, performance appraisals appear tobe directed to four purposes: to make distinctions among people, distinguish aperson’s strengths from his or her weaknesses, implement and evaluate humanresource systems in organizations, and document personnel decisions. Indeed,yet another goal of performance appraisals may ultimately be to increaseperformance at the individual and, subsequently, the organizational level(see DeNisi & Gonzalez, 2000; Meyer, Kay, & French, 1965).

These distinctions among purposes of the performance appraisal areimportant for human resource development (HRD) research and practice formany reasons. Once performance ratings are made, the adequacy of manage-ment decisions about how to develop, reward, train, and otherwise attempt toaffect the behavior of job incumbents is at stake. Having a clear idea about thelinks between initial rating decisions and later administrative and develop-mental decisions has been a major focus of considerable research literature(DeNisi & Gonzalez, 2000; Murphy & Cleveland, 1995). Because the purposeof performance appraisal is one of the factors that affect the characteristics andquality of the ratings, the practical implications of this research for profes-sionals have included numerous attempts to improve performance ratings andmeasurement.

For example, as numerous researchers have demonstrated, the purpose ofthe appraisal affects rating processes and outcomes (Bernardin & Beatty, 1984;DeNisi, Cafferty, & Meglino, 1984; Murphy, Balzer, Kellam, & Armstrong,1984; Sharon & Bartlett, 1969; Williams, DeNisi, Blencoe, & Cafferty, 1985;Zedeck & Cascio, 1982). Appraisals conducted for developmental purposes,for example, are less prone to rating biases (say, elevation or leniency) than areappraisals conducted for administrative decision-making purposes (Meyer et al.,1965; Zedeck & Cascio, 1982).

The most common purposes of performance appraisal tend to be aimedat the measurement of individuals, with the focus being on distinguishingeither among individuals (the administrative purpose) or within individuals(the developmental purpose; Cleveland et al., 1989, 2003). This emphasis onindividual-focused purposes, however, may be insufficient. The uncertain busi-ness environment that contemporary companies face has forced researchersand practitioners to conceptualize organizations as open systems with objec-tives and goals that continually need to adapt to the changing environment(Katz & Kahn, 1979; Schneider & Klein, 1994). Furthermore, from a moremacro strategic perspective, the resource-based view of the firm (Barney, 1991;Barney, Wright, & Ketchen, 2001) has challenged traditional views of compa-nies. This view proposes that the competitive advantage of a company dependson how valuable, rare, difficult to imitate, and difficult to replace are itsresources. From this perspective, certain human resources are a cornerstonefor the development of the company’s core competencies, on which thecompany invests the most. This has forced researchers to identify employees

HUMAN RESOURCE DEVELOPMENT QUARTERLY • DOI: 10.1002/hrdq

hrdq18303.qxd 8/4/07 2:43 PM Page 316

as a source of competitive advantage in unstable business environments(Barney, 1991; Barney et al., 2001), especially employees who are highlyvaluable and unique for the company’s core competency (Lepak & Snell, 1999).

In this context, employees may play a role in the definition of their ownjobs. Popular human resource practices such as quality circles (Schneider,Brief, & Guzzo, 1996) have demonstrated that incumbents are aware of theirenvironment and are able to capture customer needs and propose changes inthe work design and their own positions, thus improving organizationalperformance. From this perspective, performance appraisal theory can beexpanded to suggest that performance appraisals have a third purpose that ismore focused on the position than on the individual within the position. Thatis, change or clarification of the structure or content of the work associatedwith a position is an important and relatively overlooked purpose of perfor-mance appraisal and likely has a different impact on incumbents than do thetraditional administrative and developmental purposes.

Research on performance appraisal has tended to focus on how various pur-poses of performance appraisals influence the appraiser and the rating outcome,overlooking the impact of the perceived intent of the appraisal on the employeebeing evaluated. Whereas several researchers have examined employee reactionsto the performance appraisal system (Dipboye & De Pontriand, 1981; Mount,1983), employee attitudes related to the purpose of the appraisal remainrelatively unexplored (see Boswell & Boudreau, 2000, for an exception). Thus,another gap that needs to be addressed in performance appraisal theory is theimpact that the purpose of the appraisal may have on employee attitudes towardthe appraisal and the organization.

Study Purpose

The overarching goal of this article is to expand the traditional way of lookingat the purposes of performance appraisal in order to increase managerialunderstanding of its potential effects and subsequently enhance its use for HRD.Beyond this objective, we have two specific purposes. The first is to expandperformance appraisal theory by identifying and more broadly defining thepurposes of performance appraisals to include a position-focused purpose inaddition to the traditional incumbent-focused purposes. Although a purposeaimed at role definition has important implications for organizations, it hasremained relatively unexplored.

The second purpose is to expand performance appraisal theory by exam-ining the differential relationships that employees’ perceptions of the variouspurposes have with several criteria of interest (satisfaction, commitment, androle ambiguity). Several researchers have demonstrated that employees’ per-ceptions of the workplace are more important than the workplace itself inaffecting attitudinal and behavioral organizational responses (Allen, 2001;James & McIntyre, 1996). From this standpoint, two individuals working for

Perceived Purposes of Performance Appraisal 317


hrdq18303.qxd 8/4/07 2:43 PM Page 317


the same organization could perceive the purpose of the performance appraisalas being completely different. Regardless of the intended purpose of theappraisal, which may in fact serve both purposes, it is the perceived purposeof the appraisal that will likely influence the employee’s attitudes towardthe performance appraisal process and, ultimately, the organization.

Whereby performance appraisals have traditionally held either adminis-trative and developmental purposes, they speak directly to organizationaldevelopment and individual career development, respectively. Expanding per-formance appraisal theory to include a third purpose that goes beyond thesetraditional approaches adds another dimension in which individuals mayadvance their own careers (by learning what is expected of them, they knowwhere to focus their efforts) and identify areas in which supervisors maychoose to offer additional training and development for their employees (forexample, when the role definition purpose reveals additional requirementsfor a job that the incumbent does not already possess). In addition, a purposeaimed at role definition allows information for the organization as a whole todevelop because the information gained from this purpose may show how dif-ferent positions are increasing or decreasing in role breadth, thus indicatingwhere more or fewer resources should be allocated. This additional purpose,aimed at the position rather than at the individual within the position, is clearlyan overlooked area of research within the HRD literature. Therefore, we aimto address this gap and provide an understanding of how perceptions of thesedifferent purposes relate to important outcomes for organizations.

Literature Review

A review of the literature presents individual- versus position-focusedpurposes of performance appraisal, theoretical rationale for role defini-tion purpose, and differential relationships of perceived purposes of theperformance appraisal.

Individual- versus Position-Focused Purposes of Performance Appraisal.From the traditional perspective, performance appraisals have been concep-tualized to be intended to change the behavior of the incumbent in a position.Following Campbell’s model of job performance (Campbell, 1990; Campbell,McCloy, Oppler, & Sager 1993), performance appraisals are aimed at alteringthe motivation or the knowledge, skills, and abilities (KSAs) of the personin the position (or both). For example, supervisors may give rewards-basedfeedback based on performance ratings, which is aimed at enhancing motiva-tion of a job incumbent. Or a supervisor may decide to provide training anddevelopment to a subordinate based on skills deficits identified throughperformance appraisal. These performance appraisal outcomes have beenvirtually the entire focus of performance appraisal research, in terms of eitheraccuracy (Landy, Barnes, & Murphy, 1978) or utility (Greller, 1978; Nathan,Morhman, & Milliman, 1991), or more direct measures of behavior intention


hrdq18303.qxd 8/4/07 2:43 PM Page 318

(such as fairness perceptions; Burke, Weitzel, & Weir, 1978; Taylor, Tracy,Renard, Harrison, & Carroll, 1995).

But in a dynamic job context, another plausible purpose of performanceappraisals may be focused on the position itself rather than on the individualoccupying the position. There are several reasons a position-focused purposemay be necessary. For example, the position requirements, for which anincumbent is evaluated in incumbent-focused appraisals, may change.As advancements are made in technology, the requirements for a job are likelyto change. And just because the position requirements are changing does notmean that the employee within the position is changing too. Many employeesremain in the same positions long after initial requirements have changed andnew demands are placed on them.

Although some changes are readily apparent, such as the increased use oftechnology or increasing skills required to perform a task or job, some changesin the job are not as apparent. It is in these latter cases that a position-focusedappraisal would be most useful. For example, employees may become accus-tomed to assisting their fellow employees in ways that slowly transfer theresponsibilities of one employee’s job to that of another. Employees from twoseparate positions within an organization may realize that although one ofthem is supposed to be responsible for a particular task, the other actually hasmore resources (such as time, opportunity, or KSAs) in order to complete thetask more efficiently or with higher quality. These changes may result withoutupper management’s knowledge or approval and therefore may not be reflectedin the appraisal system for the two positions (the one with added responsibil-ity, as well as the one with less responsibility). This may be even more the casewhen employees perform such activities as contextual or citizenship behaviors(Bateman & Organ, 1983; Borman & Motowidlo, 1993; Organ, 1988; Smith,Organ, & Near, 1983). According to Hough and Oswald (2000), organizationsclearly require both task and citizenship performance. Because employees mayassist each other and complete each other’s tasks, organizational decision mak-ers would be well served to allow for and accept that jobs will slowly changeand task requirements may switch hands. This sort of employee developmentby position redefinition and citizenship may have important effects on bothindividual attitudes and organizational effectiveness.

In the event that employees do become responsible for additional tasks orchanges in technology or protocol occur, the performance appraisal should notonly reflect these changes; they may be a meaningful means for incorporatingthese changes through the performance review process. At a minimum, anemployee should not be punished for not completing aspects of a job that areno longer necessary to fulfillment of the purpose of the position. Similarly,an employee should be assessed on aspects of the job that are essential foroptimal performance, regardless of whether these tasks were part of an originaljob description. As long as the manager is aware of changes in position require-ments and the employee is accurately assessed on his or her performance in



hrdq18303.qxd 8/4/07 2:43 PM Page 319


the job in its current form, such problems are less likely to arise. The interactionthat occurs during feedback following the performance appraisal may be aprime time for discussions of these changes in job and position requirements.In the best instance, the supervisor may choose to alter the position based ona particular strength of a particular incumbent. All of these examples illustrateways that the performance appraisal itself can be considered an occasion forposition, or role, definition.

From another point of view, in tight labor markets, only some incumbentsmay be able to fulfill all functions of a position as it has been conceptualizedprior to their being employed. In such cases and in lieu of firing, managersmay be forced to reduce position expectations in terms of level or number.When done systematically and with full understanding of the specific limita-tions of incumbents, alterations of the position and its design may serve toaccomplish the organizational functions without making unreasonabledemands on the incumbents. Similarly, in an uncertain business environment,altering a position to take full advantage of particular incumbents’ talents mayhelp the organization gain a competitive advantage, if only by retaining avalued incumbent.

In summary, in addition to individual-focused purposes of performanceappraisals, another possible purpose could be change in the structure ofthe work associated with the position. From this perspective, performanceappraisal outcomes may ultimately lead to training an employee to improve hisor her performance (a developmental decision), removing an employee froma position (administrative decisions such as firing, promotion, and layoff ) oraltering position requirements (role definition decisions such as reengineeringor reorganization). The former cases are akin to the individual-focused devel-opmental and administrative purposes of the performance appraisal; the lattercase is more directed at the position, whereby the position itself may be alteredin the interest of improving organizational performance or retaining a valuableincumbent. Thus, performance appraisal decisions are oriented toward theeffectiveness of the position as a whole, in addition to its subset of individualincumbent effectiveness (DeNisi, 2000; Ilgen, 1993; Murphy & Cleveland,1992, 1995).

Theoretical Rationale for Role Definition Purpose. Some models of jobperformance (Campbell et al., 1993; Day, Sin, & Chen, 2004) suggest that thedeterminants of performance include not only abilities and motivations, butpeople’s declarative knowledge of what is expected of them. In their compre-hensive theory of organizational behavior, Katz and Kahn (1979) describedthis in terms of social information and refer to it as role definition: the extentto which relevant job behaviors are clearly defined within a social environment.To the extent that role expectations are well understood, incumbents maydemonstrate both their abilities and the level of effort they exert to meet theseexpectations. If people do not know what is expected of them, there is a goodchance that their behavior will not conform to important others’ expectations,


hrdq18303.qxd 8/4/07 2:43 PM Page 320

and effort and abilities will not be expressed appropriately in work behavior.Put simply, it is important to be willing and able to do a job, but it is essentialto know what is expected in the job. It would be hard to demonstrate one’sability and motivation without a clear idea of how and when to apply them.

But where and when do people receive role information in organizations?This question is far from new, as discussed over a quarter of a century ago byKatz and Kahn (1979), who noted, “Research on role-sending in organizationshas thus far concentrated more on side effects than main effects (that is, roleenactment per se) and more on particular inadequacies in the expectationalpattern and sending process (conflict, ambiguity, overload) than on construc-tive elements” (p. 203). These authors cited Likert’s suggestion (1967) thatsupportive supervision may contribute to effective role enactment. This notionhas received attention in the literature on participation in performanceappraisal (see Cawley, Keeping, & Levy, 1998). However, the basic construc-tion of roles has received less attention.

One possible exception to this is Management by Objectives (MBO;McConkie, 1979; Murphy & Cleveland, 1995), which explicitly requiressupervisors and subordinates to negotiate mutually agreeable goals in order toenhance incumbent effectiveness. Although this comes close to describing therole definition purpose, there are notable differences in terms of formality,prescription, and criteria of importance. Using Katz and Kahn’s terms (1979),MBO is a formal, prescriptive system aimed at individual effectiveness. Therole definition purpose, however, may be a more nonnormative, informal partof discussions about performance that follow from the appraisal and are aimednot only at individual effectiveness but at the effectiveness of the position morebroadly. That is, the purpose is to alter the position and clarify actual tasks andrequirements of the position, not necessarily the agreed-on goals of themanager and employee. In a very rough sense, this role definition purpose actsas an ongoing job or position analysis.

In this manner, performance appraisals can serve to identify aspects of thejob that are no longer required or areas where job requirements haveexpanded. We propose that the performance appraisal is an important toolthrough which employees may receive information regarding their roles andsuperiors may adjust role expectations. This evaluation, along with discussionswith the incumbent about organizational expectations regarding his or her role,is an important opportunity for role definition and adjustment.

Differential Relationships of Perceived Purposes of the PerformanceAppraisal. The second purpose of this article is to examine the differentialrelationships of the three performance appraisal purposes (administrative,developmental, and role definition purposes), with several attitudinal outcomesof interest (employees’ commitment, satisfaction, and role clarity or ambigu-ity). The following sections describe the relationships that would be expectedalong with the theoretical and empirical justification for these expectations.Although testing the relationships of the different attitudinal outcomes



hrdq18303.qxd 8/4/07 2:43 PM Page 321


considered in this study was not an aim, because previous research hasdemonstrated that some of the attitudinal outcome variables share a significantamount of variance, the hypotheses suggested in this study were tested aftercontrolling for some of these relationships. Figure 1 shows the hypothesizedmodel. We next describe the relationships among dependent variables thatneed to be controlled to test the hypotheses. Then we provide the rationale toexpect the three perceived purposes of the performance appraisal system to becorrelated. Finally, we present the empirical and theoretical support to suggestthe differential relationship between each perceived performance appraisalpurpose and each of the dependent variables.

Relationships Among Dependent Variables. Numerous studies have demon-strated that job satisfaction correlates positively with commitment (Green,Anderson, & Shivers, 1996; McFarlin & Sweeney, 1992; Mowday, Steers, &Porter, 1979). Nevertheless, because of the difficulty of specifying the causalprecedence of different affective responses (Mathieu & Zajac, 1990; Meyer,Stanley, Herscovitch, & Topolnytsky, 2002) and the fact that job satisfactionand affective commitment contribute uniquely to withdrawal behaviors suchas turnover intentions (Tett & Meyer, 1993), the hypotheses suggested in thisstudy are tested after controlling for the bidirectional relationship betweenaffective commitment and job satisfaction.

Role ambiguity has been found to correlate negatively with job satisfaction(Abramis, 1994; Jackson & Schuler, 1985). In this case, because role ambigu-ity is a variable that captures the perceptions individuals have regarding their


Figure 1. Hypothesized Model for the Study

AdministrativePurpose

DevelopmentalPurpose

Role DefinitionPurpose

JobSatisfaction

PerformanceAppraisal

Satisfaction

AffectiveOrganizationalCommitment

RoleAmbiguity

hrdq18303.qxd 8/4/07 2:43 PM Page 322

job and job satisfaction is a variable that captures an affective response towardthe job, the effect of role ambiguity on job satisfaction will be controlled whentesting the hypotheses suggested in this study.

Finally, individuals’ job satisfaction has been researched based on thereports of workers regarding their job in general and with different facets oftheir job, such as satisfaction with their pay, their supervisor, and the organi-zational climate. Because satisfaction with the job is broader than satisfactionwith individual facets of the job, it is logical to assume that job satisfaction isa more distal variable than is the satisfaction with specific job facets (see Chen,Gully, Whiteman, & Kilcullen, 2000, for similar thinking with general andspecific self-efficacy). Thus, job satisfaction can be considered an antecedentof job satisfaction. Because satisfaction with the performance appraisal systemcan be considered a facet of the job, we posit that this variable has a directeffect on job satisfaction.

Relationships Among the Three Perceived Purposes. Researchers (for example,Cleveland et al., 1989, 2003) have found that performance appraisals oftenserve multiple purposes rather than one sole purpose. Therefore, individualsmay perceive, perhaps quite accurately, that the same appraisal is used formultiple purposes. For example, an employee may perceive a performanceappraisal as serving a developmental purpose in that it identifies his or herstrengths and weaknesses, and simultaneously perceive the ratings as beingused to determine a raise. Similarly, the employee may perceive theperformance appraisal as indicating problems in the position (for example,because of low ratings in an area where the employee lacks controlor opportunities to excel), thereby leading to changes in the job. Conse-quently, the employee may also view the same appraisal as serving a roledefinition purpose. In this manner, we expect individuals’ perceptions of anyone purpose of the performance appraisal to be related to both of the others.Nevertheless, we maintain that the purposes of performance appraisal aredistinct, separate purposes, as conceptualized and demonstrated by previousresearchers (Boswell & Boudreau, 2000; Cleveland et al., 1989, 2003;Ostroff, 1993).

Effects of Administrative Purpose on Dependent Variables. Administrativedecisions can lead to such things as pay increases, bonuses, and promotions,which are often valued outcomes to employees. Thus, it is logical to assumethat perceptions of such purposes would be related to job satisfaction.However, we do not expect a direct relationship between the administrativepurpose and job satisfaction. Instead, we hypothesize that the relationshipbetween administrative purpose and job satisfaction is mediated bysatisfaction with the performance appraisal. As several researchers havenoted, a performance appraisal system will be less successful if it is notaccepted (Carroll & Schneier, 1982; Murphy & Cleveland, 1995). To theextent employees accept or are satisfied with the appraisal, they are expectedto be more satisfied with their job in general. That is, a lack of satisfaction



hrdq18303.qxd 8/4/07 2:43 PM Page 323


with the appraisal suggests that employees do not accept the rating or feelthe appraisal is not just or accurate, and it may subsequently lead to feelingsof overall job dissatisfaction because the valued outcomes are not as likely.

Effects of Developmental Purpose on Dependent Variables. We hypothesizea relationship between perceptions of a developmental purpose and affectivecommitment. Affective commitment has been defined as the emotional bondindividuals experience with an organization concerning their involvementin, identification with, and attachment to their organization (Meyer & Allen,1991). Under the tenets of social exchange theory (Blau, 1964), it isreasonable to expect that if an individual feels the organization is providingsupport and opportunities for improvement, evidenced by the appraisalbeing used for developmental purposes, he or she may respond attitudinally.That is, because the organization is viewed as caring for the employee(through the developmental appraisal), the employee will reciprocate withcommitment to the organization (Cialdini, 2001).

We also posit that perceptions of the performance appraisal as serving adevelopmental purpose are related to job satisfaction through satisfaction withthe performance appraisal. Considering that developmental performanceappraisals are aimed at identifying strengths and weaknesses for the purposesof improvement and enhancement, individuals who perceive the appraisal asbeing used for this purpose may believe the organization is interested in theirimprovement and well-being. We believe this may be reflected in part throughtheir satisfaction with the appraisal. This satisfaction with the appraisal is thenexpected to lead to job satisfaction.

Effects of Role Definition Purpose on Dependent Variables. Because perfor-mance appraisals aimed at role definition purposes are defining andclarifying roles, we posit that individuals who perceive the performanceappraisal as being for this purpose will have lower levels of role ambiguity.Thus, by definition, if an individual views the appraisal as being at least inpart intended for role clarification, then it is likely he or she will experienceless ambiguity regarding the expectations of the job.

We also posit that individuals who view the purpose of the appraisal asbeing for the clarification of roles will be more satisfied with the performanceappraisal and consequently more satisfied with their job. This may result fromseveral mechanisms. First, having a chance to communicate their opinionsregarding the expectations and requirements of the job has been shown tohave a positive relationship with satisfaction. For example, in a sample ofmanagement-level employees, Korsgaard and Roberson (1995) found that bothinstrumental voice and noninstrumental voice were important in predictingsatisfaction with the evaluation. Second, there is some reason to believethat having an opportunity to help craft the job through role definition inperformance appraisal may have an effect on satisfaction as well. Using meta-analytic procedures, Spector (1986) found that high levels of perceived controlare positively related to job satisfaction. The feedback discussion following the


hrdq18303.qxd 8/4/07 2:43 PM Page 324

performance appraisal may be a good time for employees to be activelyinvolved in the definition of their jobs and may give employees a sense of con-trol over both the performance appraisal process and their job. Although voiceand actual control over the process may not truly exist, the perceptions thatthe appraisal is helping to clarify roles is the key to these affective reactions.Thus, perceptions that the performance appraisal serves a role-definitionpurpose are expected to relate positively to performance appraisal satisfaction.Satisfaction with the appraisal is believed to then lead to job satisfaction, as wehave already noted.

We further propose that when individuals perceive the performanceappraisal as serving role definition purposes, they will have greater levels ofaffective commitment. When employees view the performance appraisal beingused to make changes in their job, they may feel a sense of ownership and con-trol over the appraisal and the job. Spector (1986) showed that high levels ofperceived control are related to greater commitment toward the organization.Consequently we posit that employees will feel a greater sense of commitmenttoward the organization to the extent that the performance appraisal clarifiesexpectations and make necessary changes in the job.

Summary of the Problem and Hypotheses

As part of our overarching goal to expand the traditional way of looking atperformance appraisal purposes, we aim to increase managerial understandingof the potential effects of there being a role definition purpose in addition to thetraditional purposes. As such, this study posed two primary research questions:(1) Do employees perceive performance appraisals to be used for these three pur-poses? and (2) How do these different purposes relate to employee perceptionsof job satisfaction, satisfaction with the performance appraisal, commitment toone’s job, and understanding of the expectations of their role?

Boswell and Boudreau (2000) found that the perception of the develop-mental use of the performance appraisal influenced employee satisfaction withthe performance appraisal and satisfaction with the appraiser but found nosuch relationship between the perception of evaluative use of performanceappraisals and these two attitudinal outcomes. We expand on their researchby examining the relationships between perceptions of the two traditionalpurposes of the performance appraisal (administrative and developmental), aswell as the third purpose of role definition that we have proposed. Further-more, we expand the attitudinal outcomes to include overall job satisfaction,affective commitment, and role ambiguity.

In the light of our research questions, and based on the rationale we haveoutlined, we proposed and tested the following hypotheses:

HYPOTHESIS 1. The three performance appraisal purposes are separate but relatedconstructs.



hrdq18303.qxd 8/4/07 2:43 PM Page 325


HYPOTHESIS 2. The relationship between perceptions of an administrative purposeof the performance appraisal and job satisfaction is mediated by satisfaction withthe performance appraisal.

HYPOTHESIS 3. There is a positive relationship between perceptions of a develop-mental purpose of the performance appraisal and affective commitment.

HYPOTHESIS 4. The relationship between perceptions of a developmental purpose ofthe performance appraisal and job satisfaction is mediated by satisfaction with theperformance appraisal.

HYPOTHESIS 5. There is a negative relationship between perceptions of a role defin-ition purpose of the performance appraisal and role ambiguity.

HYPOTHESIS 6. The relationship between perceptions of a role definition purpose ofthe performance appraisal and job satisfaction is mediated by satisfaction with theperformance appraisal.

HYPOTHESIS 7. There is a positive relationship between perceptions of a role defini-tion purpose of the performance appraisal and affective commitment.

Method

Data for this study were collected as part of a larger examination of the effec-tiveness of a company’s performance appraisal system. Questionnaires weresent through an interstore mail system to all employees and managers at 298randomly selected automotive parts stores in a single organization. Originally300 stores were selected, but 2 stores could not be used because they were notyet open. A total of 3,556 questionnaires were distributed to stores in Iowa,Kansas, Missouri, Nebraska, and Texas. Participants were told that this wasuniversity-based research designed to evaluate the company’s performanceappraisal system, that completion of questionnaires was voluntary, and that allindividual responses would be kept confidential. Participants were given twoweeks to return the questionnaire, with an e-mail reminder from humanresources sent two days before the deadline. The questionnaires were returnedto the human resource department in envelopes marked as confidential andwere seen only by the researchers responsible for data entry.

Participants. A total of 1,047 individuals returned questionnaires, for aresponse rate of 29 percent. Only individuals classified as employees wereincluded in this study, however, because we were interested in employees’perceived purposes of the appraisal, as opposed to the intended purposes ofthe appraisal. Consequently, managers and respondents who did not reporttheir position were eliminated. From the sample of 643 employees whoreturned the questionnaires, 559 employees reported complete information forthe variables of interest. Most of these respondents were white (79 percent)and male (70 percent), representative of the organization as a whole.


hrdq18303.qxd 8/4/07 2:43 PM Page 326

Measures. We next present each of the measures used in this study.We have organized them by the independent variables (which include the threeperceived purposes of the performance appraisal) and the dependent variables(which include the attitudinal outcomes of role ambiguity, job satisfaction,satisfaction with the performance appraisal, and affective commitment).

Independent Variables. Perceptions regarding the three purposes of thecompany’s performance appraisal system were measured with nine items: six(three administrative and three developmental) were taken from existingresearch (Cleveland et al., 1989; Longenecker, Liverpool, & Wilson, 1988),and three role definition purpose items were developed for this study.Response options ranged from 1 (strongly disagree) to 5 (strongly agree).The items taken from previous research were chosen based on howrepresentative they were of the constructs of interests. The items used toassess perceptions of an administrative purpose were (1) “Performanceappraisals help determine whether to promote, retain or terminate anemployee,” (2) “Performance appraisals determine what raise someoneshould receive,” and (3) “The performance appraisal process documents andrecognizes employee performance.” The items assessing perceptions of adevelopmental purpose were (4) “Performance ratings let employees knowwhere they stand,” (5) “Performance ratings are used to provide feedbackabout employee performance,” and (6) “Performance appraisals identifyindividual strengths and weaknesses.” Finally, the three items selected tomeasure perceptions of a role definition purpose were (7) “Performanceappraisals provide information about what employees are responsible foraccomplishing,” (8) “Performance appraisals provide information that helpsmake positive changes in the job itself,” and (9) “Performance appraisalsprovide information about what employees actually do in their jobs.”Reliability estimates for administrative, developmental, and role definitionpurposes were .70, .75, and .73, respectively.

Dependent Variables. Three items measuring perception of role ambiguitywere taken from Rizzo, House, and Lirtzman’s measure (1970) of roledefinition problems. The three role ambiguity items were (1) “I know exactlywhat is expected of me,” (2) “I feel certain about how much authority I haveon the job,” and (3) “I know what my responsibilities are.” Only three itemswere used due to space considerations in the final questionnaireadministered. These items were chosen because of their relevance to theconstruct of interest and because they were among the items with the highestfactor loadings in Rizzo et al.’s original article (.61, .51, and .61, respectively).Response choices ranged from 1 (strongly disagree) to 5 (strongly agree).All items were reverse-coded to reflect ambiguity rather than clarity.The reliability estimate for this scale for the current study was .75.

Job satisfaction was measured with three items previously used by Trembleand Alderks (1992). These items were (1) “I usually feel that my job isworthwhile,” (2) “I am interested in my present job,” and (3) “Overall, I am



hrdq18303.qxd 8/4/07 2:43 PM Page 327


satisfied with my current job,” Response choices ranged from 1 (stronglydisagree) to 5 (strongly agree). The reliability estimate for this scale was .81.

Satisfaction with the performance appraisal was measured with thefollowing two items: (1) “I am pleased with how performance appraisals arehandled,” and (2) “I am generally satisfied with the performance appraisalprocess.” A response set from 1 (strongly disagree) to 5 (strongly agree) wasused. The reliability estimate for this scale was .86.

Affective commitment to the organization was assessed using three itemsfrom Allen and Meyer’s affective commitment scale (1990): (1) “I really feel asif this company’s problems are my own,” (2) “This company has a great dealof personal meaning for me,” and (3) “I do not feel ‘emotionally attached’ tothis company,” (reverse-coded) with a scale ranging from 1 (strongly disagree)to 5 (strongly agree). Allen and Meyer’s full eight-item scale was not includeddue to space limitations in the questionnaire. However, the three items wereshown in their original study as being related to a single factor, with loadingsof .52, .79, and .81, respectively. The reliability estimate for this scale for thecurrent study was .69.

Analyses and Results

To test the hypothesized model (see Figure 1), analyses of covariance structureswere performed in LISREL 8.3 (Jöreskog & Sörbom, 1993). To control forcapitalizing on chance when sequentially introducing modifications tothe model, we followed MacCallum’s suggestion (1992), performing theanalyses in three sequential steps. In the first step, tau-equivalent measurementmodels were estimated for the independent variables and the dependentvariables separately using a first subsample (n � 155). In the second step,the hypothesized model was tested with a second subsample (n � 197). And in the third step, the results were cross-validated with a third subsample(n � 207). These three subsamples were obtained randomly from the originalsample. To control for the threat of common method bias due to the cross-sectional nature of the data, as recommended by Podsakoff, MacKenzie, Lee,and Podsakoff (2003), all analyses were performed controlling for the effects ofa single unmeasured latent method factor. In addition, to control for item carry-over effects, we assessed the correlations between the errors of sequential itemsbut retained only the significant paths for the analyses.

To test the appropriateness of the models estimated through the analyses,we provide several indicators of the goodness of fit. First, we provide the goodness-of-fit index (GFI), which deals with error in reproducing thevariance-covariance matrix and should be equal to or greater than .90 to acceptthe model. However, because GFI tends to be sensitive to the sample sizeand the distribution of the data, we also provide the comparative fit index (CFI),which has been shown to be among the measures least affected by sample size(Fan, Thompson, & Wang, 1999). By convention, the CFI should be .90 or


hrdq18303.qxd 8/4/07 2:43 PM Page 328

greater to accept the model. Another index that we included, which is lessaffected by sample size and deviations from the normal distribution, is thenonnormed fit index (NNFI), in which values below .90 indicate a need torespecify the model, and values .95 and above indicate a good model (Hu &Bentler, 1999). Finally, we use the root mean square error of approximation(RMSEA) as an indicator of goodness of fit. Researchers (Hu & Bentler, 1999)have suggested that an RMSEA value of .06 or less indicates good model fit,although there is a tendency for it to overestimate goodness of fit for smallsample sizes (Fan et al., 1999). To test the improvement of the goodness offit of a model when removing (or adding) parameters, we examine the increaseof the chi square of the nested model at a level of significance of .05. There-fore, if a path is removed from a model and the increase on the chi square isnot significant, the more parsimonious model is considered a better solution.

Table 1 shows scale means, standard deviations, reliability coefficients, andintercorrelations among the variables of interest. As shown, the correlationsamong variables were all significant at a level of .01 even after controlling forcommon method variance, which tends to contribute to inflated correlations.

Independent Variables Measurement Model. We first estimated the inde-pendent variables measurement model assuming tau equivalence amongthe items for each scale. To do this, a three-factor model underlying the nineperformance appraisal purpose items was estimated with a four-factor confir-matory factor analysis (CFA), which controlled for a single unmeasured latentfactor. In this analysis, the paths between the items and their hypothesized factor,as well as the paths between each item and the unmeasured latent factor, wereconstrained to be equal, the correlations among the three hypothesized factorswere left free for estimation, and the correlations of the three factors with thefourth single unmeasured latent method factor were constrained to zero. Also,the correlations between the errors of two pairs of sequential items were left free



Table 1. Descriptive Statistics, Reliabilities, and Correlations for the Variables Examined

M SD 1 2 3 4 5 6 7

1. Administrative purpose 3.65 0.72 (.70)

2. Development purpose 3.84 0.62 .73 (.75)

3. Role definition purpose 3.60 0.70 .60 .67 (.73)

4. Affective commitment 3.32 0.86 .36 .37 .40 (.69)

5. Job satisfaction 3.95 0.71 .39 .46 .45 .48 (.81)

6. Role ambiguity 4.00 0.67 .31 .42 .36 .29 .48 (.75)

7. Performance appraisal 3.39 0.98 .43 .43 .49 .39 .45 .30 (.86)satisfaction

Note. N � 599. All correlations are significant at a level of .01. Reliability coefficients are along thediagonal.

hrdq18303.qxd 8/4/07 2:43 PM Page 329


for estimation to test the presence of item carry-over effects. To test hypothesis 1,the goodness of fit for this model was compared with (1) one two-factor modelwith one common factor (representing administrative, developmental, and roledefinition purposes combined into a single purpose) and a second single unmea-sured latent method factor, and (2) three three-factor models obtained with allpossible combinations of two of the three factors (administrative and develop-mental combined versus role definition separate, developmental and role defin-ition combined versus administrative separate, and administrative androle definition purposes combined versus developmental separate) and a thirdsingle unmeasured latent factor. After testing for the appropriateness of thehypothesized three-factor structure, the loadings obtained for the items in eachscale and the correlations among the three factors were retained to constrainthese paths when testing the hypotheses with the second subsample.

The four-factor CFA performed on the nine independent variable itemsyielded an appropriate fit. However, because the correlations between theerrors of two pairs of consequential items were not statistically significant,the model was estimated again without leaving free for estimation thoseparameters. As shown in Table 2, when compared with the two-factor solution,in which all the items were associated with one common factor and a singleunmeasured latent factor, the four-factor solution resulted in a statisticallysignificant chi-square increase, suggesting that the hypothesized solution hasa better fit. Furthermore, change in chi square was statistically significant whencompared with any of the possible combinations of the three-factor model.

The four-factor solution showed that each of the three perceived adminis-trative purpose items loaded 0.59 on their latent factor, the three perceiveddevelopmental purpose items each loaded 0.55 on their factor, and theperceived role definition purpose items all loaded 0.66 on their factor. Inaddition, as expected, the three hypothesized factors were highly correlated.Specifically, the correlation between perceived administrative purpose andperceived developmental purpose factors was .98, the correlation betweenthe perceived developmental purpose and the perceived role definitionpurpose factors was .92, and the correlation between the perceived develop-mental purpose and the perceived role definition purpose factors was .76.


Table 2. Independent Variables Measurement Models’ Goodness-of-Fit Indexes

Models �2 df p GFI CFI NNFI RMSEA �, �2* df p

Four factors 30.95 30 .42 .96 1.00 1.00 .014 — — —

Two factors 57.51 33 .00 .92 .98 .96 .007 26.56 3 .00

Note: GFI � goodness-of-fit index; CFI � comparative fit index; NNFI � nonnormed fit index;RMSEA � root mean square error of approximation.

hrdq18303.qxd 8/4/07 2:43 PM Page 330

The strong correlations among the three purposes can be explained tosome extent by the statistical technique used to perform the analysis. Becausecovariance structural analysis is a statistical technique in which the error ofmeasurement is considered and partialed out from the item variances whenanalyses are performed, the correlations between latent factors tend to be muchstronger than the zero-order correlation obtained from the scale scores. In thiscase, the correlation between administrative purpose and developmentalpurpose, initially estimated as being .73 when error of measurement is notremoved, is .97 after correcting for the item error of measurement. In addition,the correlation between developmental purpose and role definition purpose,initially estimated as being .67, is .92 after controlling for the item error ofmeasurement. Finally, the correlation between administrative purpose and roledefinition purpose, initially estimated as being .60, is .76 after controllingfor the error of measurement. Despite the strong correlations among theindependent variables, the fact that the four-factor measurement model (threehypothesized factors and a single unmeasured latent factor) demonstratedstatistically better fit than any other possible factor structure gives strong supportfor our first hypothesis. Thus, it appears that the three perceived performanceappraisal purposes, although highly correlated, are empirically distinct.

Dependent Variables Measurement Model. In the second set of analyses,similar to the procedure used to estimate the independent variables measure-ment model, a series of CFAs was performed to test the appropriateness of thehypothesized four-factor solution for the dependent variable items (that the fourdependent variables—job satisfaction, satisfaction with the performanceappraisal, affective commitment, and role ambiguity—are all separateconstructs). In this case, a four-factor model underlying the eleven dependentvariable items was estimated performing a five-factor CFA, which controlled fora single unmeasured latent factor (to control for the threat of common methodbias, as described earlier). As with previous analyses, the paths between theitems and its hypothesized factor and those between each item and the unmea-sured latent factor were constrained to be equal, the correlations among the fourhypothesized factors were left free for estimation, and the correlations of thosefactors with the fifth single unmeasured latent method factor was constrainedto zero. Also, the correlations between the errors of two pairs of sequential itemswere left free for estimation. The goodness of fit for this model was comparedwith (1) a two-factor model with one common factor (all dependent variableitems loading on the same factor) and a second single unmeasured latentmethod factor, (2) six possible four-factor models obtained with all combina-tions of three out of four factors and a fourth single unmeasured latent factor,and (3) seven possible three-factor models obtained with all combinations oftwo out of the four factors and a third single unmeasured latent factor. Aftertesting for the appropriateness of the hypothesized three-factor structure, theloadings obtained for the items in each scale were retained to constrain thesepaths when testing the hypotheses with the second subsample.



hrdq18303.qxd 8/4/07 2:43 PM Page 331


The five-factor CFA performed with the eleven dependent variable itemsobtained an appropriate goodness of fit. This analysis also showed thatthe error of the two pairs of sequential items was not statistically significant,so the analysis was performed constraining to zero the relationship betweentheir errors of measurement. Table 3 shows that when a two-factor solution inwhich the eleven items load on a common factor and a single unmeasuredlatent factor was compared to the hypothesized factor structure, the increasein chi square was statistically significant, suggesting that the hypothesized fac-tor structure had a better fit than the alternative factor structure. Furthermore,the comparison of the hypothesized factor structure with all other possiblealternative factor structures demonstrated that the hypothesized model yieldedbetter fit than any other possible factor structure. In terms of specific factorloadings, the hypothesized five-factor solution showed that the three commit-ment items each loaded 0.72 on their latent factor, the three satisfaction itemsloaded 0.62 on their factor, the role ambiguity items all loaded 0.65 on theirfactor, and the satisfaction with performance appraisal items loaded .90 ontheir latent factor.

Structural Equation Model: Testing the Hypotheses. After estimating theindependent variable and dependent variable measurement models separately,using the second subsample, we assessed the hypothesized relationshipsbetween the independent and dependent variables estimating a null model, inwhich the paths between all the latent exogenous factors and the latent endoge-nous factors were constrained to zero, while the hypothesized relationshipsbetween dependent variables were left free for estimation. In order to estimatethe suggested bidirectional relationship between organizational commitmentand job satisfaction, the beta paths between these two variables were con-strained to be equal. The results obtained for each of these paths were retainedto test the hypothesized model. Then we assessed the goodness of fit of thehypothesized model structure with all the proposed relationships. Parameterestimates were reviewed and modification indexes explored to test each of theremaining six hypotheses. Once the final model was identified, the solution wascompared with a competing model, in which all the paths between the latentexogenous factors and the latent endogenous factors were left free for estima-tion. The final solution was cross-validated using the third subsample.


Table 3. Dependent Variables Measurement Models’ Goodness-of-Fit Indexes

Models �2 df p GFI CFI NNFI RMSEA �, �2* df p

Five factors 94.66 45 .00 .90 .96 .95 .08 — — —

Two factors 298.46 51 .00 .74 .77 .65 .17 203.81 6 .00

Note: GFI � goodness-of-fit index; CFI � comparative fit index; NNFI � nonnormed fit index;RMSEA � root mean square error of approximation.

hrdq18303.qxd 8/4/07 2:43 PM Page 332

The estimation of the null model showed that the suggested relationshipswere statistically significant. The bidirectional relationship between organiza-tional commitment and job satisfaction was .22, the effect of satisfaction withperformance appraisal on job satisfaction was .22, and the effect of role ambi-guity on job satisfaction was .42. Therefore, the hypothesized model shown inFigure 1 was estimated with structural equation modeling techniques, leavingfree for estimation the hypothesized relationships between exogenous andendogenous latent variables. In this way, the parameter estimates suggestedthat the solution was not appropriate. That is, several estimated paths were notsignificant, and the estimation of simultaneous relationships forced the modelto overestimate some of the parameters. One of the explanations for the over-estimation of the parameters is the high correlations among the independentvariables, which leave little variance to estimate the relationships with thedependent variables. Therefore, we had to adopt a different strategy to testthe hypotheses. First, we estimated an alternative model in which theadministrative purpose (satisfaction with performance appraisal), develop-mental purpose (affective commitment), and role definition purpose(role ambiguity) paths were included. Second, we estimated the increase in thechi square when each of the other hypothesized relationships was successivelyincluded in the model.

The suggested alternative model obtained acceptable goodness of fit (x2(186) �

288.67, p � .01, GFI � .86, CFI � .96, NNFI � .96, RMSEA � .05).This provided support for hypothesis 3, which suggested that perceiveddevelopmental purpose was significantly related to affective commitment, andsupport for hypothesis 5, which suggested that perceived role definitionpurpose was significantly related to role ambiguity.

To test hypotheses 2, 4, 6, and 7, we compared this alternative model suc-cessively with other models to test each of our hypotheses. When the directpath between administrative purpose and job satisfaction was included in themodel, the increase in chi square was not statistically significant, suggestingthat the relation between perceived administrative purpose and job satisfac-tion was fully mediated by satisfaction with the performance appraisal system(x2

(185) � 285.15, p � .01, GFI � .86, CFI � .96, NNFI � .96, RMSEA � .05,dx2

(1) � 3.42, p � .06). This provides full support to hypothesis 2, whichsuggested that the perceived administrative purpose of performance appraisalis related to job satisfaction and mediated by satisfaction with the performanceappraisal.

When the relationship between perceived developmental purpose and jobsatisfaction was included in the alternative model, the increase in chi square wasnot statistically significant (x2

(185) � 285.60, p � .01, GFI � .86, CFI � .96,NNFI � .96, RMSEA � .05, dx2

(1) � 3.07, p � .08). Furthermore, when the pathbetween perceived developmental purpose and satisfaction with theperformance appraisal was included in the alternative model, the increase in chisquare was not statistically significant (x2

(185) � 285.16, p � .01, GFI � .86,



hrdq18303.qxd 8/4/07 2:43 PM Page 333


CFI � .96, NNFI � .96, RMSEA � .05, dx2(1) � 3.51, p � .06). Thus, hypoth-

esis 4, which suggested that the relationship between perceived developmen-tal purpose and job satisfaction was mediated by satisfaction with performanceappraisal, was not supported.

When the relationship between the perceived role definition purpose andjob satisfaction was included in the alternative model, the increase in chisquare was not statistically significant (x2

(185) � 286.80, p � .01, GFI � .86,CFI � .96, NNFI � .96, RMSEA � .05, dx2

(1) � 1.07, p � .17). In addition,the inclusion of a path between perceived role definition purpose andsatisfaction with the performance appraisal into the alternative model did notyield a statistically significant increase in chi square (x2

(185) � 287.06, p � .01,GFI � .86, CFI � .96, NNFI � .96, RMSEA � .05, dx2

(1) � 1.61, p � .21).Thus, we found no support for hypothesis 6, which suggested that the rela-tionship between perceived role definition purpose and job satisfaction wasmediated by satisfaction with the performance appraisal.

Finally, when the relationship between the perceived role definition purposeand affective commitment was included in the alternative model, the chi-squareincrease was not statistically significant (x2

(185) � 288.65, p � .01, GFI � .86,CFI � .96, NNFI � .96, RMSEA � .05, dx2

(1) � 0.02, p � .89). Therefore,hypothesis 7, which suggested a significant relationship between perceived roledefinition purpose and affective commitment, was not supported.

The total unstandardized effects of the exogenous variables on the endoge-nous variables showed that perceived administrative purpose had a significantand strong effect on satisfaction with performance appraisal (.53), a signifi-cant but small effect on job satisfaction (.14), and a significant but very smalleffect on affective commitment (.03). Perceived developmental purpose had asignificant and strong effect on affective commitment (.49) and a significantbut small effect on job satisfaction (.11). Finally, perceived role definitionpurpose had a significant and moderate effect on role ambiguity (�.33),significant but small effect on job satisfaction (.15), and significant butvery small effect on affective commitment (.03). Figure 2 shows the completelystandardized solution for the final model.

To test the stability of the final model, a cross-validation was performedwith the third subsample. The model fit was marginally acceptable (x2

(186) �415., p � .01, GFI � .81, CFI � .93, NNFI � .93, RMSEA � .07), and allpaths between the exogenous variables and endogenous variables were signif-icant at a level of .05. This result offers stronger support for the alreadysupported hypotheses: 1, 2, 3, and 5.

Discussion

Our first objective was to identify and more broadly define the purposes of per-formance appraisals. In addition to the traditional administrative anddevelopmental purposes aimed at incumbent performance management, we


hrdq18303.qxd 8/4/07 2:43 PM Page 334

proposed that another purpose of the appraisal may be directed toward definingthe position itself: a role definition purpose. We assessed employee perceptionsof these three performance appraisal purposes and found that performanceappraisal systems were not only perceived to provide information regardingthe employees’ administrative future and their own development, but also weredirected at the position as a whole within its organizational context.

Although we expected the three purposes to be correlated with oneanother, they were correlated to a greater extent than is ideal. Specifically, thetwo individual-focused purposes were the most correlated (.98), followed bythe relationship between developmental purpose and role definition purpose(.92), and then the relationship between administrative purpose and role def-inition purpose (.76). We assert, however, that they are still separate constructsfor several reasons. First, although they are the highest correlations within thisarticle, previous researchers have demonstrated that the two individual-focusedpurposes (administrative and developmental) are empirically distinct. Giventhat these two purposes are both individual focused, it is not altogether sur-prising that they are highly correlated. The distinction between these twopurposes, however, is not the focus of this study. The high correlations betweenthe role definition purpose and the other two purposes are the primaryconcern here. Nevertheless, they are clearly separate from a theoretical stance.Furthermore, we refer to previous studies (Bosson & Swann, 1999; Chen,Gully, & Eden, 2004) that also had high correlations between their constructs



Figure 2. Completely Standardized Solution for the Final Model

AdministrativePurpose

DevelopmentalPurpose

Role DefinitionPurpose

JobSatisfaction

PerformanceAppraisal

Satisfaction

AffectiveOrganizationalCommitment

RoleAmbiguity

.58

.98

.92

.76

.47

.22

–.42

–.33

.22

hrdq18303.qxd 8/4/07 2:43 PM Page 335


of interest. Specifically, Bosson and Swann (1999) found that people’s feelings ofself-liking and self-competence were highly correlated with one another(.76) yet independently related to self-verification strategies. Similarly, Chenet al. (2004) found that general self-efficacy and self-esteem were highlycorrelated (.80) but were considered to be empirically and conceptuallydistinguishable because they related to task performance in predictably distinctprocesses. Thus, in both examples, although highly intercorrelated, they wereconsidered to be distinct constructs because they were differentially related tooutcomes of interest or related in manners that were predictably different. Also,the error of measurement is considered and partialed out from the itemvariances when analyses are performed using covariance structural analysis;thus, the correlations between latent factors are inflated compared to what onewould find with zero-order correlations. Indeed, our analyses comparingmodels indicate that the three purposes are best thought of as being empiri-cally distinct. Therefore, although they are highly intercorrelated, they can beconsidered separate constructs.

The second objective of this study was to examine the differentialrelationships these various purposes have with several criteria of interest.We hypothesized numerous relationships and tested these using structuralequation modeling techniques. Although the first model did not support all ofour specific predictions, the model supported our overall propositions:(1) employees distinguish among the three proposed performance appraisalpurposes, and (2) the perceptions of different purposes for the performanceappraisal system are related differentially to the different attitudinal variables.Our results suggest that the perception of each of the three purposes ofperformance appraisal systems relates differently to attitudinal outcomes.Specifically, perceived administrative purpose was related to job satisfactionthrough satisfaction with the performance appraisal; perceived developmentalpurpose yielded direct relationships with job satisfaction and affectivecommitment; and perceived role definition purpose was directly related to roleambiguity, which related to job satisfaction.

Implications for HRD Research and Practice. This study contributes toHRD research and practice in several ways. In terms of implications for HRDresearch, our second objective at least begins to answer Cleveland et al.’s longoverdue call (1989) for more research on the correlates and consequences ofusing performance appraisals for potentially conflicting uses. Future researchis needed to explore the consequences of using performance appraisalsfor multiple purposes. An extension of these effects to nonattitudinalconsequences might be particularly useful. Along these lines and in line withimplications for HRD practice, to the extent that there are any differences inthe perceived purpose of the performance appraisal, employers may want tobe clearer in communicating the intended purpose of the system dependingon the outcome they hope to achieve. For example, in times of change,managers may find that employees are confused about what their actual job


hrdq18303.qxd 8/4/07 2:43 PM Page 336

responsibilities entail. Under these conditions, managers may want not onlyto build a role definition purpose into the performance review system but tocommunicate this intent to employees in order to increase employee percep-tions of role clarity.

Another implication for our study with HRD research and practice involvesthe criteria that we chose to use. The criteria of interest in this study(satisfaction, commitment, and role ambiguity) are primarily attitudinalvariables, which have broad implications for their responses to individual devel-opment systems, organizational development, and ultimately organizationaleffectiveness. Employees’ attitudes, especially to the organization, are increas-ingly relevant for companies in the contemporary unstable and competitivebusiness environment. It has been demonstrated that organizational attitudespredict turnover (Griffeth, Hom, & Gaertner, 2000), which may hurt companyperformance from at least two standpoints. In the short term, turnover has afinancial impact on an organization because of the costs associated with recruit-ing, selecting, hiring, and training a new employee (Abbasi & Hollman, 2000).Moreover, these costs are usually higher for valuable and unique humanresources, which demand more investment (Lepak & Snell, 1999, 2002).Turnover also has a financial impact on the organization in the long term, as itmay represent a loss of a company’s competitive advantage when valuable, rare,hard-to-replace, and inimitable employees leave the company (Barney, 1991;Barney et al., 2001).

Our research also provides recommendations regarding perceptions ofappraisals that individuals may have. This research focused on employeeperceptions of performance appraisal purposes rather than on actual intendedpurposes of the appraisals, as designated by the organization. Therefore, ourresults support the importance of employee perceptions, regardless of actualuse, a finding that is similar to that of Boswell and Boudreau (2000). Depend-ing on the outcomes they are interested in (for example, role clarity versusaffective commitment), organizations could make certain aspects of theperformance appraisal process clear so that employees understand and believethat the appraisal is truly for that purpose, such as personal and career devel-opment. However, we echo Boswell and Boudreau’s recommendation (2000)that organizations should be mindful in matching employee perceptions towhat the organization is really doing in order to avoid possible feelings ofmanipulation or injustice.

Finally, HRD practitioners may want to take note of the versatility that athird purpose offers: it allows many different decisions to be made from thesame evaluation (such as whom to promote, areas for employee development,or what parts of the job to restructure). For example, in tight labor markets,only some incumbents may be able to fulfill all functions of a position as it hasbeen conceptualized prior to their being employed. In such cases, and especiallywhere training is not feasible because costs are too high or no trainers areavailable, for example, managers may be forced to reduce position expectations



hrdq18303.qxd 8/4/07 2:43 PM Page 337


in terms of level or number. When done systematically and with full under-standing of the specific limitations of incumbents, alterations of the positionand its design may serve to accomplish the organizational functions withoutmaking unreasonable demands on the incumbents. Similarly, in an uncertainbusiness environment, altering a position to take full advantage of particularincumbents’ talents may help the organization gain a competitive advantage, ifonly by retaining a valued incumbent. Thus, this broader conceptualization ofthe purposes of the performance appraisal may be advantageous to employeeswithin the positions as well as the organization as a whole.

Strengths, Limitations, and Future Research. This study has severalstrengths. First, data were obtained from employees within a single organiza-tion across the Midwest. By limiting the sample to a single organization, we canbe assured that relationships between the perceived purposes of the appraisaland the attitudinal outcomes are truly representing the perceptions and are notconfounded by the actual purpose of the appraisal. That is, the actual purposeof the performance appraisal did not vary by store in this organization,although clearly the perceived purposes varied in their relationships withoutcomes.

Another strength is that the relatively large sample that was available allowedus to perform cross-validations, in order to account for several potentialproblems, including common-method bias and item carry-over effects. An evenlarger sample size would have been ideal because larger subsamples may haveprevented instability in the expected relationships among variables. Still, therepresentativeness of a sample to the overall organization was certainly a strength.Yet another strength is the use of covariance structure analysis to test ourseven hypotheses. This statistical analysis allows us to test multiple relationshiphypotheses at the same time while controlling for type 1 error, which is inflatedwhen multiple relationships are tested sequentially using different statistics (as intypical regression techniques used to test mediation, for example).

This study is not without limitations. First, although we maintain that thethree purposes are distinct, as demonstrated by the criterion-related validityand results of the structural equation models provided, they were still corre-lated to a greater extent than is ideal. Nevertheless, future researchers may wantto consider using different items that are even more distinct. In addition,researchers may want to focus on whether there are differences between actualand perceived purposes and whether there is less of a distinction whenconsidering employee perceptions. Unfortunately, we did not collect data onthe actual uses of the performance appraisal and are therefore unable toexamine this within the current study context.

Another limitation with our study was that all data were cross-sectional,and therefore we are unable to infer causality. For example, satisfied and com-mitted employees may perceive performance appraisals purposes differently,rather than the perception of purpose driving satisfaction and commitment.Clearly a longitudinal design would need to be employed in the future versus


hrdq18303.qxd 8/4/07 2:43 PM Page 338

the cross-sectional one used in the current study. In addition, although ourfinal sample size was more than sufficient, our response rate (29 percent) wassomewhat lower than we would have liked. That is, although some researchers(among them, Steeh, 1981) have noted that for most mailed surveys, aresponse rate of 10 percent is typical and 20 percent is the best that can usuallybe expected, others (including Baruch, 1999; Kerlinger, 1986) have set thetypical response rate at closer to 40 or 50 percent. Nonetheless, despite effortsto increase the return rate (reminder e-mails and organizational support), ourresponse rates remained somewhat low. Future researchers may want toconsider offering incentives or conducting data collection on-site in orderto achieve higher response rates.

Two other limitations in this study concern our measures. First, our mea-sure of role definition does not clarify for whom role definition is intended.For example, the items for this purpose all say, “Performance appraisals provideinformation . . . ,” but they do not state the target of this provided information.It could be that information provided primarily to the organization wouldserve as documentation function, whereas information provided to themanager-employee dyad actually leads to a renegotiation of the role itself.Thus, although the performance appraisal purpose questions were constructedin the same fashion, the role definition purpose scales could have included thetarget of the provided information.

A second limitation concerns our measure of satisfaction with the perfor-mance appraisal not clearly differentiating between satisfaction with the processand satisfaction with the outcome. These two components are embeddedwithin the scale, with satisfaction with the process implied in the first item andsatisfaction with the process and outcome in the second item. Thus, it is notpossible to separate out these different elements of the performance appraisal.This was not a key component of our study, however, and thus was notdeemed to be a substantial limitation. Nevertheless, future researchers shouldmeasure process and outcome satisfaction separately and build the twovariables into the theoretical model.

Finally, we limited the outcome variables used in this study to attitudinalvariables. Although the focus of the research is on the purposes of the perfor-mance appraisal, not on the outcomes themselves, future research in this areamay benefit from moving beyond attitudinal outcome variables. That is,although we addressed outcomes that are quite relevant for contemporarycompanies and have been linked with numerous important outcomes, it wouldbe worthwhile to examine such aspects as individual and firm performance.In addition, performance ratings would be interesting to examine within theframework used here. For example, one could argue that the relationshipbetween the perceptions people have regarding the performance appraisalsystem and attitudinal variables may be moderated by previously obtained per-formance appraisals ratings (Boswell & Boudreau, 2000, 2002). Unfortunately,this information was not available for the study examined here.



hrdq18303.qxd 8/4/07 2:43 PM Page 339


References

Abbasi, S. M., & Hollman, K. W. (2000). Turnover: The real bottom line. Public PersonnelManagement, 29, 333–343.

Abramis, D. J. (1994). Work role ambiguity, job satisfaction, and job performance: Meta-analysesand review. Psychological Reports, 75, 1411–1434.

Allen, N. J., & Meyer, J. P. (1990). The measurement and antecedents of affective, continuance,and normative commitment to the organization. Journal of Occupational Psychology, 63, 1–18.

Allen, T. D. (2001). Family-supportive work environments: The role of organizationalperceptions. Journal of Vocational Behavior, 58, 414–435.

Barney, J. B. (1991). Firm resources and sustained competitive advantage. Journal of Management,17, 99–111.

Barney, J., Wright, M., & Ketchen, D. J. (2001). The resource-based view of the firm: Ten yearsafter 1991. Journal of Management, 27, 625–641.

Baruch, Y. (1999). Response rates in academic studies—a comparative analysis. Human Relations,52, 421–434.

Bateman, T. S., & Organ, D. W. (1983). Job satisfaction and the good soldier: The relationshipbetween affect and employee “citizenship.” Academy of Management Journal, 26, 587–595.

Bernardin, H. J., & Beatty, R. W. (1984). Performance appraisal: Assessing human behavior at work.Boston: Kent.

Blau, P. M. (1964). Exchange and power in social life. Hoboken NJ: Wiley.Borman W. C., & Motowidlo S. J. (1993). Expanding the criterion domain to include elements

of contextual performance. In N. Schmitt & W. Borman (Eds.), Personnel selection in organiza-tions (pp. 71–98). San Francisco: Jossey-Bass.

Bosson, J. K., & Swann, W. B. (1999). Self-liking, self-competence, and the quest for self-verification. Personality and Social Psychology Bulletin, 25, 1230–1241.

Boswell, W. R., & Boudreau, J. W. (2000). Employee satisfaction with performance appraisals andappraisers: The role of perceived appraisal use. Human Resource Development Quarterly, 11,283–299.

Boswell, W. R., & Boudreau, J. W. (2002). Separating the developmental and evaluative perfor-mance appraisal uses. Journal of Business and Psychology, 16, 391–412.

Burke, R. J., Weitzel, W., & Weir, T. (1978). Characteristics of effective employee performancereview and development interviews: Replication and extension. Personnel Psychology, 31,903–919.

Campbell, J. P. (1990). Modeling the performance prediction problem in I/O psychology. In M. D. Dunnette & L. Hough (Eds.), The handbook of industrial and organizational psychology(Vol. 1, pp. 704–715). San Francisco: CPI.

Campbell, J. P., McCloy, R. A., Oppler, S. H., & Sager, C. E. (1993). A theory of performance. InN. Schmitt, W. C. Borman, & Associates (Eds.), Personnel selection in organizations (pp. 35–70).San Francisco: Jossey-Bass.

Carroll, S. J., & Schneier, C. E. (1982). Performance appraisal and review systems: The identification,measurement and development of performance in organizations. Glenview, IL: Scott, Foresman.

Cawley, B. D., Keeping, L. M., & Levy, P. E. (1998). Participation in the performance appraisalprocess and employee reactions: A meta-analytic review of the field investigations. Journal ofApplied Psychology, 83, 615–633.

Chen, G., Gully, S. M., & Eden, D. (2004). General self-efficacy and self-esteem: Toward theo-retical and empirical distinction between correlated self-evaluations. Journal of OrganizationalBehavior, 25, 375–395.

Chen, G., Gully, S. M., Whiteman, J. A., & Kilcullen, R. N. (2000). Examination of the relation-ships among trait-like individual differences, state-like individual differences, and learningperformance. Journal of Applied Psychology, 85, 835–847.

Cialdini, R. B. (2001). Influence: Science and practice (4th ed.). Needham Heights, MA: Allyn & Bacon.


hrdq18303.qxd 8/4/07 2:43 PM Page 340

Cleveland, J. N., Mohammed, S., Skattebo, A. L., & Sin, H. P. (2003, April). Multiple purposes ofperformance appraisal: A replication and extension. Poster presented at the annual conference forthe Society for Industrial and Organizational Psychology, Orlando, FL.

Cleveland, J. N., Murphy, K. R., & Williams, R. E. (1989). Multiple uses of performance appraisal:Prevalence and correlates. Journal of Applied Psychology, 74, 130–135.

Day, D. V., Sin, H. P., & Chen, T. T. (2004). Assessing the burdens of leadership: Effects of formalleadership roles on individual performance over time. Personnel Psychology, 57, 573–605.

DeNisi, A. S. (2000). Performance appraisal and performance management. In K. J. Klein & S.W.J. Kozlowski (Eds.), Multilevel theory, research, and methods in organizations (pp. 121–156).San Francisco: Jossey-Bass.

DeNisi, A. S., Cafferty, T. P., & Meglino, B. M. (1984). A cognitive view of the performanceappraisal process: A model and research propositions. Organizational Behavior and HumanPerformance, 33, 360–396.

DeNisi, A. S., & Gonzalez, J. A. (2000). Design performance appraisal systems to improveperformance. In E. A. Locke (Ed.), Handbook of principles of organizational behavior (pp. 60–72).Oxford: Blackwell.

Dipboye, R. L., & De Pontriand, R. (1981). Correlates of employee reactions to performanceappraisals and appraisal systems. Journal of Applied Psychology, 66, 248–251.

Fan, X., Thompson, B., & Wang, L. (1999). Effects of sample size, estimation method, and modelspecification on structural equation modeling fit indexes. Structural Equation Modeling, 6,56–83.

Green, S. G., Anderson, S. E., & Shivers, S. L. (1996). Demographics and organizationalinfluences on leader-member exchange and related work attitudes. Organizational Behavior andHuman Decision Processes, 66, 203–214.

Greller, M. M. (1978). The nature of subordinate participation in the appraisal interview. Academyof Management Journal, 21, 646–658.

Griffeth, R. W., Hom, P. W., & Gaertner, S. A. (2000). A meta-analysis of antecedents andcorrelates of employee turnover: Update, moderator tests, and research implications for thenext millennium. Journal of Management, 26, 463–488.

Hough, L. M., & Oswald, F. L. (2000). Personnel selection: Looking toward the future—remembering the past. Annual Review of Psychology, 51, 631–664.

Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis:Conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1–55.

Ilgen, D. R. (1993). Performance appraisal accuracy: An illusive or sometimes misguided goal?In H. Schuler, J. L. Farr, & M. Smith (Eds.), Personnel selection and assessment: Individual andorganizational perspectives (pp. 233–252). Mahwah, NJ: Erlbaum.

Jackson, S. E., & Schuler, R. S. (1985). A meta-analysis and conceptual critique of research onrole ambiguity and role conflict in work settings. Organizational Behavior and Human DecisionProcesses, 36, 16–78.

James, L. R., & McInytre, M. D. (1996). Perceptions of organizational climate. In K. R. Murphy(Ed.), Individual differences and behavior in organizations (pp. 416–450). San Francisco: Jossey-Bass.

Jöreskog, K. G., & Sörbom, D. (1993). Lisrel-8 [computer program]. Chicago: Scientific Software.Katz, D., & Kahn, R. (1979). The social psychology of organizations (2nd ed.). Hoboken, NJ: Wiley.Kerlinger, F. N. (1986). Foundations of behavioral research (3rd ed.). New York: Holt.Korsgaard, M. A., & Roberson, L. (1995). Procedural justice in performance evaluation: The role

of instrumental and noninstrumental voice in performance appraisal discussions. Journal ofManagement, 21, 657–669.

Landy, F. J., Barnes, J., & Murphy, K. (1978). Correlates of perceived fairness and accuracy ofperformance appraisals. Journal of Applied Psychology, 63, 751–754.

Lepak, D. P., & Snell, S. A. (1999). The human resource architecture: Toward a theory of humancapital allocation and development. Academy of Management Review, 24, 31–48.



hrdq18303.qxd 8/4/07 2:43 PM Page 341


Lepak, D. P., & Snell, S. A. (2002). Examining the human resource architecture: The relation-ships among human capital, employment, and human resource configurations. Journal ofManagement, 28, 517–543.

Likert, R. (1967). The human organization. New York: McGraw-Hill.Longenecker, C. O., Liverpool, P. R., & Wilson, K. Y. (1988). An assessment of manager/

subordinate perceptions of performance appraisal effectiveness. Journal of Business andPsychology, 2, 311–320.

MacCallum, R. C. (1992). Model modifications in covariance structure analysis: The problem ofcapitalization on chance. Psychological Bulletin, 111, 490–504.

Mathieu, J. E., & Zajac, D. M. (1990). A review and meta-analysis of the antecedents, correlates,and consequences of organizational commitment. Psychological Bulletin, 108, 171–194.

McConkie, M. L. (1979). A clarification of the goal-setting and appraisal process in MBO.Academy of Management Review, 4, 29–40

McFarlin, D. B., & Sweeney, P. D. (1992). Distributive and procedural justice as predictors ofsatisfaction with personal and organizational outcomes. Academy of Management Journal, 35,626–637.

Meyer, H. H., Kay, E., & French, J.R.P. (1965). Split roles in performance appraisal. HarvardBusiness Review, 43, 123–129.

Meyer, J. P., & Allen, N. J. (1991). A three-component conceptualization of organizationalcommitment. Human Resource Management Review, 1, 61–89.

Meyer, J. P., Stanley, D. J., Herscovitch, L., & Topolnytsky, L. (2002). Affective, continuance, andnormative commitment to the organization: A meta-analysis of antecedents, correlates,and consequences. Journal of Vocational Behavior, 61, 20–52.

Mount, M. K. (1983). Comparison of managerial and employee satisfaction with a performanceappraisal system. Personnel Psychology, 37, 687–702.

Mowday, R., Steers, R., & Porter, L. (1979). The measurement of organizational commitment.Journal of Vocational Behavior, 14, 224–247.

Murphy, K. R., Balzer, W. K., Kellam, K., & Armstrong, J. (1984). Effect and purpose of ratingon accuracy in observing teacher behavior and evaluating teaching performance. Journal ofEducational Psychology, 76, 45–54.

Murphy, K. R., & Cleveland, J. N. (1992). Performance appraisal: An organizational perspective.Needham Heights, MA: Allyn & Bacon.

Murphy, K. R., & Cleveland, J. N. (1995). Understanding performance appraisal: Social, organiza-tional, and goal-based perspectives. Thousand Oaks, CA: Sage.

Nathan, B. R., Mohrman, A. M., & Milliman, J. (1991). Interpersonal relations as a context forthe effects of appraisal interviews on performance and satisfaction: A longitudinal study.Academy of Management Journal, 34, 352–369.

Organ, D. W. (1988). Organizational citizenship behavior: The good soldier syndrome. Lanham, MD:Lexington Books.

Ostroff, C. (1993). Rater perceptions, satisfaction and performance ratings. Journal of Occupa-tional and Organizational Psychology, 66, 345–356.

Podsakoff, P. M., MacKenzie, S. B., Lee, J., & Podsakoff, N. P. (2003). Common method biasesin behavioral research: A critical review of the literature and recommended remedies. Journalof Applied Psychology, 88, 879–903.

Rizzo, J. R., House, R. J., & Lirtzman, S. I. (1970). Role conflict and ambiguity in complexorganizations. Administrative Science Quarterly, 15, 150–153.

Schneider, B., Brief, A. P., & Guzzo, R. A. (1996). Creating a climate and culture for sustainableorganizational change. Organizational Dynamics, 24(4), 7–19.

Schneider, B., & Klein, K. J. (1994). What is enough? A systems perspective on individual-organizational performance linkages. In D. H. Harris (Ed.), Organizational linkages: Understandingthe productivity paradox (pp. 81–104). Washington, DC: National Academies Press.


hrdq18303.qxd 8/4/07 2:43 PM Page 342

Sharon, A., & Bartlett, C. J. (1969). Effect of instructional conditions in producing leniency intwo types of rating scales. Personnel Psychology, 22, 251–263.

Smith, C. A., Organ, D. W., & Near, J. P. (1983). Organizational citizenship behavior: Its natureand antecedents. Journal of Applied Psychology, 68, 653–663.

Spector, P. E. (1986). Perceived control by employees: A meta-analysis of studies concerningautonomy and participation at work. Human Relations, 39, 1005–1016.

Steeh, C. (1981). Trends in nonresponse rates, 1952–1979. Public Opinion Quarterly, 45, 40–57.Taylor, M. S., Tracy, K. B., Renard, M. K., Harrison, J. K., & Carroll, S. J. (1995). Due process in

performance appraisal: A quasi-experiment in procedural justice. Administrative Science Quar-terly, 40, 495–523.

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

Tremble Jr., T. R. & Alderks, C. E. (1992). Measures for research on small unit preparedness forcombat effectiveness. Alexandria, VA: U.S. Army Research Institute for the Behavioral and SocialSciences.

Williams, K. J., DeNisi, A. S., Blencoe, A. G., & Cafferty, T. P. (1985). The role of appraisal pur-pose: Effects of purpose on information acquisition and utilization. Organizational Behavior andHuman Decision Processes, 35, 314–339.

Zedeck, S., & Cascio, W. F. (1982). Performance appraisal decisions as a function of rater train-ing and purpose of the appraisal. Journal of Applied Psychology, 67, 752–758.

Satoris S. Youngcourt is assistant professor of psychology at Kansas State University.

Pedro I. Leiva is professor of psychology at P. Universidad Catolica de Valparaíso in Chile.

Robert G. Jones is associate professor of psychology at Missouri State University.



hrdq18303.qxd 8/4/07 2:43 PM Page 343

1207_ftp Textos Da Wiley B-On

Documents

tight labor markets

structural equation modeling

response choices ranged

human decision processes

human resource architecture

making unreasonable demands

unmeasured latent factor

uncertain business environment