G.R. Ferris et al. / Human Resource Management Review 18 (2008) 146–163 147contextual backdrop in order to frame and better comprehend the theoretical process dynamics and outcomes of performanceevaluation. This paper is structured as follows: First, we brieﬂy review the context of performance evaluation. Next, we introduceaccountability theory as a complement to the existing performance evaluation literature. We then consider contextual factors inthe performance evaluation process. Our emphasis here is on context that inﬂuences both the antecedents and process ofperformance evaluation. Finally, we incorporate Affective Events Theory (Weiss & Cropanzano, 1996) and Emotion Cycle Theory(Hareli & Rafaeli, in press) to describe outcomes of the performance evaluation process.1.2. Importance of context in performance evaluation Context is very important in the organizational sciences because it helps frame phenomena in ways that inﬂuence ourperceptions and interpretations of them, which in turn, affect decisions and actions (e.g., Johns, 2006). Supervisor–subordinatedyadic interactions and evaluations take place within a work relationship, and this relationship reﬂects social, emotional, political,and cognitive processes that help explain the decision outcomes. Furthermore, it is important to recognize that organizational contexts are changing as ﬁrms embrace the importance oforganization redesign and restructuring, with concomitant changes in the very nature of how work is designed and organized (e.g.,Bridges, 1994; Cascio, 1995; McLean Parks, Kidder, & Gallagher, 1998). The changing dynamics of organizational contexts includethe breaking down of traditional and rigid job boundaries, and the ﬁxed, static sets of duties and responsibilities that we havereferred to as “jobs,” and the move toward more ﬂuid, constantly changing sets of work roles (Bridges, 1994; Cascio, 1995). Thisnotion, along with an update to the mechanistic model that acknowledges the importance of social inputs, interactions, andcompetencies (Stewart & Carson, 1997), suggests that social dimensions of work are becoming increasingly critical to the verydeﬁnition and interpretation of job performance. Additionally, it is important to appropriately characterize performance evaluation for what it is: a formal accountabilitymechanism in organizations that holds employees answerable for their work-related behavior (e.g., Ferris & Treadway, 2008;Ferris, Mitchell, Canavan, Frink, & Hopper, 1995). As such, like other mechanisms of accountability, it is subject to lapses andineffectiveness, particularly when all of the various inﬂuences that operate on such mechanisms are not fully considered andinvestigated. In this paper, we propose that performance evaluation systems, as accountability mechanisms, are embedded withincomplex social, emotional, cognitive, political, and relationship contexts, which need to be understood in order to adequatelyinterpret the results or outcomes derived from such systems.2. Accountability and performance evaluation Accountability has been referred to as “the adhesive that binds social systems together” (Frink & Klimoski, 1998, p. 3), and itprovides a mechanism through which societies and organizations can direct and control the conduct of their members (Beu &Buckley, 2001). The expectation of not only being held answerable for ones actions, but also receiving consequences as a result, hasobvious implications regarding behavioral motivation, and in a larger sense, social order (Tetlock, 1985). Frink and Ferris (1998)suggested that evaluation is a major component of accountability. As such, the performance evaluation process provides a uniquecontext for investigating accountability-based phenomena in organizations. Indeed, performance evaluation systems are formalmechanisms of accountability in organizations, and are construed as such for purposes of this paper (see also, Ferris & Treadway,2008; Ferris et al., 1995).2.1. Accountability theory and application The relationships present in an accountability context require fundamental assumptions about motives for the parties involved.Early theories of accountability were focused at the ﬁrm-level and based heavily on the positivist branch of agency theory (Jensen &Meckling, 1976). This theory argued that the goals and interests of owners are at odds with the interests of employees, who will actin their own self-interest if not monitored or closely managed (Eisenhardt, 1989). As a result, some form of corporate governancemechanism is needed in order to protect owner interests. Although useful for explaining the development of formal control mechanisms, Tetlock (1985) argued that agency theory andother foundational models of accountability (e.g., the pyramid model and the triangle model of responsibility; Schlenker, Britt,Pennington, Murphy, & Doherty, 1994) ignored the social context in which individuals make decisions. Tetlocks (1985) socialcontingency model presented the “person-as-intuitive-politician” metaphor, which suggested that individuals in organizationsseek to protect their social image and are inherently motivated to seek approval from key constituencies who will be evaluatingthem. This metaphor also supports the assertions of other researchers that organizations are “political arenas” (Mintzberg, 1983;Ferris & Judge, 1991), and helps frame our understanding of performance evaluation as an accountability context. One of the most universal mechanisms by which an organization holds individuals accountable for their behavior isperformance evaluation. Evaluation has been demonstrated to increase performance and effectiveness (Ferris et al., 1995), andrepresents a critical point at which objective job factors interact with the subjective and social elements of the organization. In thetraditional sense, performance evaluation provides supervisor/raters with an opportunity to evaluate subordinate/ratees bycomparing their behaviors to established goals or standards. One interesting, but complicating, feature of this phenomenon is thenotion that, just as the performance evaluation system is a formal mechanism by which supervisors/raters can hold subordinates/
148 G.R. Ferris et al. / Human Resource Management Review 18 (2008) 146–163ratees accountable, so too can supervisors/raters be held accountable by the organization for the quality of their ratings. Forexample, Klimoski and Inks (1990) and Mero, Guidice, and Brownlee (2007) demonstrated that differing levels of accountabilitycan inﬂuence supervisors/raters performance ratings of subordinates, and the accuracy of those ratings. Frink and Klimoski (2004) described the elements of the accountability process that may offer useful insight into theperformance evaluation process: (1) employees being held accountable; (2) evaluative others; (3) standards or expectationsagainst which employees behavior is compared; (4) the belief by employees that their performance will be evaluated; and (5) theexpectation by employees of some outcome based on their performance relative to the standard.2.2. Employees held accountable One fundamental consideration in the accountability literature is that individual differences may explain differentinterpretations of, and reactions to, the same accountability system. To paraphrase Lewin (1936), individuals respond to theirperceptions of accountability, not the formal system of accountability per se. The ﬁlter through which the participant experiencesthe system obviously inﬂuences the effectiveness of that system. As mentioned earlier, typically two levels of accountability exist inthe performance evaluation context: supervisors/raters doing the rating of performance, and subordinates/ratees being rated. Also,subordinates/ratees bring unique personality characteristics and motives, as well as skills, knowledge, and abilities into the ratingrelationship. Each of these factors is likely to inﬂuence the subordinates performance, as well as the formal and informalaccountability system they perceive. However, the accountability of supervisors/raters should not be forgotten in this context. Decisions and ratings by supervisors/raters represent actions that may warrant justiﬁcation. For example, Klimoski and Inks (1990) found that accountability inﬂuencedperformance appraisal rating accuracy. As such, the biases, skills, abilities, and personality traits of supervisors/raters mayinﬂuence the degree to which they are perceived as successful relative to some rating standards.2.3. Evaluative others Frink et al. (2008) stated, “In many cases, accountability episodes include multiple audiences, and these often with conﬂictinginterests” (p. 28). This represents a “web of accountabilities” (Frink & Klimoski, 1998) into which employees often are placed. In theperformance evaluation context, supervisors typically represent the most salient audience. However, other rating sources often areconsidered in more contemporary appraisal systems. Self-appraisal may be included as a formal source for input, although veryseldom is that the only source. Schlenker and Weigold (1992) identiﬁed the “self as audience” concept as well, albeit in a moreinformal capacity. In either situation, the self-evaluation serves as a referent point against which the other formal sources ofevaluation are compared. The notion of multiple, simultaneous sources of accountability highlights the potential for role ambiguity or role conﬂict (Katz &Kahn, 1978), and unfavorable outcomes such as increased stress and decision-evasion tactics (Green, Visser, & Tetlock, 2000).Whether the evaluative audience is a single source or involves multiple sources, the evaluated individuals must address thepotential conﬂicting expectations for performance. In support of Tetlocks (1985) aforementioned “person-as-intuitive politician”metaphor, the ratee will often attempt to protect their social image by acting in ways that garner the approval of key constituencies.Speciﬁcally, when forced to prioritize the audience to which they will attend, employee/ratees often will conform to the audiencewith whom they have the strongest relationship (Frink & Klimoski, 1998), or represents the most salient outcome source. In a traditional performance evaluation scenario, the target audience for subordinates/ratees is the immediate supervisors/raters, assuming the rater is perceived as legitimate and salient with regard to desired outcomes. Depending on the accountabilitystructure in place, supervisors/raters, on the other hand, may be accountable for their ratings decisions to their supervisors, thesubordinates/ratees, or both. Mero, Guidice, and Anna (2006) reported that supervisors/raters adjust their ratings depending onthe characteristics of the audience to whom they are accountable (i.e., high, mixed, or low status), and the form of accountabilityused (i.e., face-to-face justiﬁcation of ratings or written justiﬁcation). They demonstrated that, in conditions where justiﬁcation forratings would be given to high or mixed status audiences, supervisor/rater accuracy was higher than if justiﬁcation was given tosubordinates/ratees (i.e., in which case, ratings tended to be inﬂated). Similarly, if the justiﬁcation was to be given face-to-face tohigh status or mixed audiences, ratings were more accurate than if justiﬁcation was written.2.4. Standards or expectations against which employees are compared Understanding the performance expectations is a fundamental element of any employment relationship. Lerner and Tetlock(1999) identiﬁed several methods by which evaluated individuals respond to evaluative audience expectations. Although framedaround decision justiﬁcation, they argued that individuals conform their attitudes, decisions, or behavior to align with targetevaluative audiences if their opinions are known in advance, or can be guessed. The reduced cognitive complexity associated withthis process reﬂects an effort to please a valued audience. This mindless conformity to audience preference, as well as decreasedaccuracy of objective performance evaluations by supervising raters, represent two of what Frink and Klimoski (1998) describe as“dark side” outcomes of increased accountability (Frink & Klimoski, 1998). One key criterion for a performance evaluation system is the speciﬁcity of the process or outcome for which employees areresponsible (Ferris & Treadway, 2008). Over the years, organizational scholars have offered a number of dimensions by which jobperformance may be deﬁned in an attempt to more clearly specify the performance expectations. For example, Borman and
G.R. Ferris et al. / Human Resource Management Review 18 (2008) 146–163 149Motowidlo (1993) proposed the task and contextual performance as two distinct dimensions of overall job performance (wediscuss task and contextual performance in greater detail in the next section). From the subordinate/ratee perspective, a clearunderstanding of the nature of the evaluated job is critical. Obviously, the less objective the measure of that performance, the morelikely supervisors/raters and subordinates/ratees will differ regarding standards and actual performance.2.5. Employees beliefs their performance will be evaluated A critical motivating element of any accountability system is the awareness that individuals will be evaluated based on theirperformance. Most research on the effects of accountability has used notiﬁcation of an imminent requirement to justify behaviorsor decisions as the manipulation resulting in outcomes of interest. Tetlock (1985) emphasized the importance of the timing ofaccountability notiﬁcation, stating, “Pre-exposure accountability can affect the actual processing of the evidence; post-evidenceaccountability can affect only the recall and analysis of already processed evidence” (p. 230). Advanced knowledge of anaccountability requirement has been related to improvements in a number of judgment and decision-making outcomes (Frink &Klimoski, 1998). For example, Mero and Motowidlo (1995) found that participants who knew their evaluations of videotapedcandidates would need to be justiﬁed were more engaged in tasks, were more attentive, and took more notes than those who didnot believe they were accountable.2.6. Expectation of an outcome The consequences associated with performance evaluation are absolutely critical elements because of their effects on behavior,and must be emphasized in any effective system. Beets and Killough (1990) went so far as to state, “standards without enforcementare window dressing” (p. 116). The awareness and salience of an expected outcome can inﬂuence behavior of supervisors/ratersand subordinates/ratees alike, with the outcome serving as a reinforcing or punishing force. If an evaluation process is known tohave little impact on pay, promotion, or any other salient outcome, for example, it may do little to inﬂuence behavior ofsupervisors/raters or the subordinates/ratees. One of the contextual elements through which the accountability system can be described relates to what Hall, Bowen, Ferris,Royle, and Fitzgibbons (2007) describe as “accountability focus.” This dimension of the organizational environment reﬂects thedegree to which individuals will be accountable for the outcome of their behaviors and decisions, or the process by which theyachieved those results. The literature has demonstrated that outcome-accountability focus has been associated with lower decisionquality (Adelberg & Batson, 1978), unfavorable escalation of commitment in poor decisions (Simonson & Staw, 1992), and lesstruthful behavior (Adelberg & Batson, 1978). Longenecker, Sims, and Gioia (1987) offered a particularly salient example of how outcome-accountability focus inorganizations can drive rater behaviors in unintended directions. In this qualitative study, the authors illustrated how the goalsof the evaluation process (e.g., accuracy) often are not aligned with the goals of the users of the system, who is more focused onoutcomes for which the manager/rater will be held accountable. For example, many managers shared how willing they were tosacriﬁce the accuracy of performance ratings for the promise of a more valuable outcome (i.e., low conﬂict in the work group,strategically positioning someone for promotion or removal, “sending a message” with a low/high rating, giving the impressionthat all employees in the workgroup are outstanding performers, etc.). In such cases, supervisors/raters may be providingunexpected outcomes to subordinates/ratees in an effort to accomplish outcomes for which they ultimately will be heldresponsible by their own supervisors or peers. The issues highlighted above reinforce the notion that performance evaluation is, in fact, an accountability mechanism,operating within a complex context. In the remaining sections of this paper, we sort out and review in detail the social, emotional,cognitive, political, and relationship components of this complex context, and discuss their implications for developing a moreinformed understanding of the performance evaluation process as a formal mechanism of accountability in organizations.3. Characterization of job performance construct Although the very nature of job performance has drawn the attention of organizational scientists for quite some time, soundtheoretical work differentiating the performance construct was lacking until Campbell (1990). He proposed a conceptualization ofjob performance that included dimensions focusing on the execution of substantive tasks, as well as elements focusing onmotivational and interpersonal features. This delineation separated formally prescribed aspects of job performance from thoseelements that, albeit valued by the organization, are neither explicitly designated nor required. Subsequently, Murphy and Cleveland (1995) suggested that two dimensions or categories of job performance can bedistinguished across most all jobs: (1) the explicitly prescribed tasks and duties typically provided in job descriptions (i.e., taskperformance); and (2) those aspects of performance dictated by the social context of the job and work environment. Therefore,they argued that one prominent feature of social contextual performance is social effectiveness (i.e., the effectiveness with whichindividuals develop and maintain good interpersonal relations with others in their work environment). This job performance differentiation between formally prescribed duties of a job, and the non-required aspects suggested bythe social context received more extensive theoretical grounding with the research by Borman & Motowidlo (1993, 1997a,b). Theymore formally and substantially established the two categories as ‘task performance’ (i.e., the set of core duties and tasks that arecentral to a particular job), and ‘contextual performance’ (i.e., behaviors not formally prescribed by any particular job but instead
150 G.R. Ferris et al. / Human Resource Management Review 18 (2008) 146–163informal aspects of all jobs). Borman and Motowidlo (1993) originally conceived of contextual performance to include suchbehaviors as cooperating, helping, volunteering, following rules, and so forth, which appear quite similar to what to organizationalcitizenship behaviors.4. Performance evaluation theory and research A major point of departure occurred in the early 1980s (e.g., Landy & Farr, 1980; Mitchell, 1983, Wexley & Klimoski, 1984),suggesting that performance evaluation is not simply an instrumentation issue, driven by an historically psychometric orientation,but also a “complex process involving social, situational, affective, and cognitive elements” (Ferris, Judge, Rowland, & Fitzgibbons,1994, p. 101). This highlighted the importance of the broad and multifaceted context as background for better understandingperformance evaluation. The ‘context’ is a rather broad term, which encompasses the research examining the cognitive perspective on supervisor ratingprocesses, the social and relationship context, social inﬂuence and politics, the role of affect and emotion in rating processes, andother features of the dyadic context, such as ﬁt, perceived similarity, and distance. In essence, the ‘performance evaluation context’reﬂects the multifaceted background against which formal and informal appraisals of job performance take place. This is not meantto add confusion regarding the aspects of job performance referred to as ‘contextual’ by Borman and his colleagues (e.g., Borman &Motowidlo, 1993), which denote non-required aspects of job performance behavior, as discussed above.4.1. Cognitive, social, emotional, political, and relationship context of performance evaluation In the following sections, we separately present research on the cognitive, social and relationship, affective and emotional, andpolitical and relationship context features, and discuss how each informs the nature of performance evaluation processes andoutcomes. However, the categorization by component is purely for ease of presentation, and not indicative of how these features ofthe context operate. Instead, we need to note here that cognitive, social and relationship, affective and emotional, and political andrelationship-based components of the work context reﬂect an intricately intertwined and integrative contextual backgroundagainst which performance evaluation needs to be perceived and interpreted.4.2. Cognitive context in rating Much research has been conducted in efforts to better understand rater cognition in the performance evaluation process (e.g.,DeNisi & Williams, 1988; Landy & Farr, 1980; Motowidlo, 1986). As noted by Wexley and Klimoski (1984), the general cognitivesequence begins with the presentation of social information (i.e., behavior and performance), which is attended to, encoded,stored, retrieved, and integrated, which then results in the action of rating. Because raters are ﬁnite information processors, problems can emerge concerning the observation, integration, or evaluation ofbehavior as a function of categorization, behavior sampling, encoding of behavior, and precisely how the integration processtranspires. Cognitive models have investigated a number of issues and processes, including automatic and controlled modes ofinformation categorization, initial expectations, schemas, and implicit personality theories, the latter of which includes theconﬁguration of individual characteristics and behavioral tendencies (e.g., DeNisi & Williams, 1988). In a somewhat different approach to rater information processing, Motowidlo (1986) argued that there is a true score domain ofbehaviors, which incorporates a universe of all possible pieces of information about a particular stimulus object (i.e., the ratee). Bitsof information sampled from actual observation or experience with the ratee are stored in the raters memory during theevaluation process, and a sample of such stored impressions is retrieved from memory to render an evaluative judgment. Interestingly, Motowidlo (1986) suggested that the information bits stored in memory do not exist in non-evaluative,descriptive form, but rather tend to be translated into positive or negative evaluative impressions. Of course, this highlights theimportance of affect in cognitive information processing models. Affective and evaluative impressions can be formed almostinstantaneously after initial exposure to the stimulus (Zajonc, 1980), and cognitive categories reﬂecting stronger affect tend to beassociated with greater memory accessibility (Feldman & Lynch, 1988). Although the cognitive information processing perspectivein performance evaluation has been enlightening in itself, it is incomplete when viewed alone and in isolation of the larger andmore multifaceted work context. Thus, we move next to the investigation of the social, emotional, political, and relationshipfeatures of the performance evaluation context.4.3. Social and relationship context Over the past couple of decades, a number of studies have investigated the social context of performance evaluation inwhich some aspect(s) of supervisor–subordinate dyadic work relationships were considered (e.g., Mitchell, 1983; Mitchell &Liden, 1982). Cleveland and Murphy (1992, p. 121) characterized performance evaluation as “a social and communicationprocess” where each member of the dyad is pursuing goals that are inﬂuenced by the social context within which therelationship is embedded. They argued that a key proximal variable affecting the performance evaluation process was thesupervisor–subordinate work relationship, consistent with Ferris and Judge (1991), and they reviewed research that found thesocial context to be of considerable importance (e.g., Freeberg, 1969; Kallejian, Brown, & Weschler, 1953; Mitchell & Liden,1982).
G.R. Ferris et al. / Human Resource Management Review 18 (2008) 146–163 151 Nathan, Mohrman, and Millman (1991) conducted a longitudinal investigation of the effects of the interpersonal relationsbetween supervisors and subordinates on the content features of performance evaluation process and outcomes one to twomonths later. The results demonstrated that supervisor–subordinate interpersonal relations related signiﬁcantly to all threecontent areas (i.e., evaluation criteria, career discussion, and participation in the review), which, in turn, affected performance andsatisfaction. Interpersonal relations between supervisor and subordinate were measured with one item that asked: “At the presenttime, my relationship with my supervisor is …,” followed by seven bipolar adjective pairs, measured on a 7-point scale. Theadjectives were: tense-relaxed, cautious-free, distrusting-trusting, bad-good, productive-destructive, friendly-hostile, andpleasant-unpleasant. Although the relationship variable was merely a one-item measure, which did not explicitly mentionunderlying dimensions, the bipolar adjectives used in response to the item seem to reﬂect some potentially important dimensionsof relationships. Duarte, Goodson, and Klich (1994) studied the potential interactive effects of supervisor–subordinate relationship quality,objective performance, and duration of the work relationship on performance ratings provided by supervisors. The resultssupported the three-way interaction, essentially demonstrating that objective performance evidence dictates subjectiveperformance ratings for subordinates in short-term, low-quality relationships. However, subordinates in longer-term dyadicwork relationships received higher performance ratings regardless of their objective performance, thus suggesting that time is apotentially very important dimension of work relationships. Judge and Ferris (1993) tested a model of the social context of performance evaluation that examined the effects of thesupervisor–subordinate work relationship. They measured the work relationship using ﬁve items developed by Graen and hiscolleagues (e.g., Dansereau, Graen, & Haga, 1975) to assess LMX relationship quality. Although no speciﬁc dimensions wereexplicitly discussed, the ﬁve items seemed to tap dimensions of distance/closeness, dependability, and support. Judge and Ferrisfound that the supervisor–subordinate work relationship demonstrated a signiﬁcant path to supervisors affect toward thesubordinate. As noted by Borman and Motowidlo (1993), contextual performance behaviors support the social fabric of the organization, andas such, should play important roles in job performance ratings. Furthermore, Guion (1983) issued appeals to scholars that causalmodels of job performance ratings should include rater–ratee interpersonal relationship factors. Borman and colleaguesinvestigated these notions, and, not surprisingly, their research demonstrated that these types of behaviors (i.e., contextualperformance behaviors) inﬂuenced supervisor evaluations of employee performance (Borman, White, & Dorsey, 1995; Borman,White, Pulakos, & Oppler, 1991). In an initial study, Borman et al. (1991) tested a model that positioned cognitive ability, job knowledge, and task proﬁciency aspredictors of supervisor ratings of subordinate performance, in addition to achievement orientation, dependability, awardsreceived by ratee, and disciplinary actions taken against the ratee. The social and motivational constructs of achievementorientation and dependability demonstrated interesting indirect effects on performance ratings through awards and disciplinaryaction. Continuing with this program of research, Borman et al. (1995) investigated the effects of a broader array of rater–rateerelationship factors and also ratee personal characteristics on peer and supervisor job performance ratings. For both peer andsupervisor rating models, dependability emerged as the strongest ‘relationship variable’ predictor of performance, and signiﬁcantin both models. Ratee likability and friendliness exhibited signiﬁcant paths in the peer rating model, but neither variabledemonstrated signiﬁcant paths in the supervisor rating model.4.4. Similarity and distance as other social context features It has been over two decades since Schneider (1987) developed the Attraction–Selection–Attrition (ASA) framework, and in thattime, ﬁt has become an increasingly investigated topic in the organizational sciences. The model suggests that organizations attractcertain individuals who possess qualities and characteristics reﬂected, or valued, by such ﬁrms, select those individuals who best ﬁtthose valued attributes, with the acknowledgement that those who are homogenous with the rest of the organization will tend tostay, and those dissimilar will tend to leave. ‘Fit’ can be viewed in a comparable manner to research on similarity (i.e., or perceived similarity), and we can deﬁne it as thecompatibility between subordinates and their supervisors that occurs when at least one of the individuals provides what the otherneeds, they share similar basic characteristics, or both. Much of this work has focused on demographic similarity (e.g., Tsui &OReilly, 1989). These studies examined the supplementary ﬁt between a supervisor and subordinate on such characteristics as age,race, gender, and organizational tenure. Tsui and OReilly (1989) found that increased demographic differences betweensupervisors and subordinates were associated with lower subordinate effectiveness as perceived by supervisors, and decreasedpersonal attraction as rated by supervisors. Consistent with previous research (e.g., Turban & Jones, 1988), perceived similarity is construed as supervisors’ view of howmuch alike their subordinates are to themselves, which raises theoretical arguments of cognitive limitation and informationprocessing. Unlike the emotional considerations of supervisor affect or liking toward a subordinate, similarity perceptions tend torely heavily on cognitive evaluations of a subordinates value to the organization and to the supervisor. Other research has focusedon overall personal perceived similarity as well as perceived attitude similarity (e.g., Pulakos & Wexley, 1983; Wexley, Alexander,Greenawalt, & Couch, 1980). Distance between supervisors and subordinates is a potentially important aspect of the context of performance evaluation, anddistance can be psychological or physical in nature. Nearly half a century ago, Rothaus, Morton, and Hanson (1965) discussed
152 G.R. Ferris et al. / Human Resource Management Review 18 (2008) 146–163psychological distance as an important factor between supervisors and subordinates. Psychological distance refers to a supervisorsperception of similarity with the subordinate on a number of traits. Wexley and Klimoski (1984) mentioned the length of timesupervisors and subordinates have known each other could be construed as a measure of closeness (i.e., psychological distance).More current conceptualizations (e.g., Williams & Bargh, 2008) deﬁne psychological distance as a function of temporal, spatial andsocial components. Physical distance also is important in supervisor–subordinate interactions and evaluations because it can affect the opportunityto observe behavior and performance (Mitchell, 1983; Wexley & Klimoski, 1984). Judge and Ferris (1993) discussed spatial distance(i.e., between supervisors and subordinates) as a potentially important factor in the performance evaluation process. Additionally,Napier and Ferris (1993) proposed a conceptualization of distance in organizations, which included psychological, physical, andfunctional components, the latter reﬂective of an optimal degree achieved in high-quality work relationships. Considerations of physical and psychological distance are complemented by the inﬂuence of time in the performanceevaluation context. Speciﬁcally, researchers have been primarily concerned with variance in evaluations due to temporal effects (cf.Mitchell & James, 2001). For example, Murphy (1989) developed a framework to distinguish between phases of work. He suggestedthat employees work in either transition or maintenance stages of work. Transition stages describe phases of work whereemployees are learning the job or adapting to changes within their work environment. Conversely, the maintenance stage of workdescribes relative stability once an employee has learned the task and contextual requirements of the job. This distinction isimportant because it likely inﬂuences rater/supervisor expectations concerning performance. Maintenance stages are likely tomanifest relative stability in performance evaluations. However, transition stages may exhibit a great deal of variance inperformance evaluations as employees learn, with varying degrees of success, new requirements for the position.4.5. Social inﬂuence and politics Traditional perspectives generally have assumed that performance evaluation systems and processes operate in rational andsystematic fashion. However, other perspectives have argued that performance ratings are susceptible to inﬂuence by suchnonperformance factors as politics and active manipulation. Indeed, several studies have found work contexts containingperformance evaluation systems to be quite political in nature (e.g., Ferris, Fedor, Chachere, & Pondy, 1989; Longenecker et al.,1987). In their model of political inﬂuence in human resources systems, Ferris and Judge (1991) argued that the performanceevaluation process is susceptible to subjective factors and “deliberate manipulations by both evaluators and evaluatees” (p. 461).Furthermore, they suggested that tactics and strategies of inﬂuence demonstrated by subordinates should affect supervisorperceptions of liking of subordinates and perceived similarity, which in turn, affect human resources decisions and actions (e.g.,performance evaluations). Several empirical investigations of portions of this model have provided consistent validation (e.g.,Kolodinsky, Treadway, & Ferris, 2007; Wayne, Liden, Graf, & Ferris, 1997). Wayne et al. (1997) tested portions of the Ferris and Judge (1991) framework. Essentially, they tested a model that examinedsubordinate inﬂuence tactics leading to performance evaluations, promotability ratings, and salary through mediating conditionsof supervisor perceptions of similarity, affect/liking, and subordinate interpersonal skills. The results demonstrated that bothsupervisor perceptions of similarity and subordinate interpersonal skills mediated the inﬂuence tactics—outcomes linkages, butaffect/liking toward subordinates did not. Ferris et al. (1994) formulated and tested a model of the social context of performance evaluation that included inﬂuence tacticsand distance as predictors of supervisor affect toward subordinate, which in turn affected performance ratings provided bysupervisors and provision of resources to subordinate. Therefore, this study examined also issues raised in the Ferris and Judge(1991) framework. Results demonstrated that inﬂuence tactics, but not distance, had signiﬁcant paths to supervisor affect towardsubordinate, which then signiﬁcantly inﬂuenced both performance ratings and provision of resources to subordinate. Dulebohn and Ferris (1999) argued that inﬂuence tactics, exhibited by subordinates, in supervisor–subordinate dyadicperformance evaluation contexts, can assume the role of informal voice, which can lead subordinates to view greater involvementin and control over the evaluation process. Informal voice refers to the provision of input into organizational decision-makingprocesses. Their results demonstrated that inﬂuence tactics as voice demonstrated signiﬁcant relationship with subordinateperceptions of performance evaluation fairness. Other work also has examined social inﬂuence and political processes inperformance evaluation (e.g., Frink, Treadway, & Ferris, 2005).4.6. Affect and emotion The role of affect in performance evaluation has stimulated considerable research, and undergone signiﬁcant evolution (Levy &Williams, 2004). Initially, the role of affect in performance evaluation was seen through an information processing lens, and it wassuggested that our affect towards a stimulus (i.e., either positive or negative) preceded and inﬂuenced the cognitive processing ofinformation concerning that stimulus. The way we felt toward people inﬂuenced how we evaluated their performance. Indeed,there are numerous studies that have suggested a signiﬁcant relationship between positive/negative affect and performanceratings (Cardy & Dobbins, 1986; Kingstrom & Mainstone, 1985; Tsui & Barry, 1986; Wayne & Ferris, 1990). However, this relationshipis now perceived as more than simply a cognitive information processing issue. An alternative suggestion is that we pay signiﬁcantly more attention to those for whom we have positive affect, and that thisresults in more observational opportunities (Antonioni & Park, 2001). Furthermore, as Levy and Williams (2004) have noted, we
G.R. Ferris et al. / Human Resource Management Review 18 (2008) 146–163 153may have misperceived the causal arrow in this relationship. Perhaps we like those who perform better because they are higherperformers. Our positive affect may well be a result of higher levels of performance, not an antecedent factor in the performanceevaluation process. There has been increasing interest in the role of emotions in the workplace, and some research has indicated that thedemonstration of positive emotions results not only in higher employee achievement, but also increased allocation of rewards byothers. Staw, Sutton, and Pelled (1994) found that positive affect signiﬁcantly predicted employees’ pay, and Staw and Barsade(1993) reported that those exhibiting positive emotions at work tended to receive higher performance evaluations from theirsupervisors.4.7. Theoretical foundations Recently, Levy and Williams (2004) helped facilitate the gradual move of the performance evaluation literature away fromthe cognitive processing/rater training/psychometric issues/scale construction emphasis. Indeed, the move was only graduallyaccomplished, with many suggesting the need for such a move along the way (e.g., Ilgen, Barnes-Farrell, & McKellin, 1993).Utilizing this research as their justiﬁcation, Levy and Williams (2004, p. 883) concluded that the most important area toinvestigate is the “performance appraisal context” and the “social milieu or rating environment.” Their work has beeninstrumental in better characterizing the context in which performance evaluation takes place in organizations. In this, theyhave been quite successful, and provided an exhaustive organizing framework for subsequent research in this area. However, we see a need to take this issue a step further, and identify the intricate and multifaceted nature of the variablesinvolved in, and the theoretical approaches underlying, the social context of performance evaluation. Levy and Williams (2004)identiﬁed many of the important variables (i.e., proximal, distal, and structural), but they did not provide a clear idea of how andwhat fashion these various factors operate/interact. Therefore, we believe that even though Levy and Williams gave the ﬁeld agood start in characterizing the context of performance evaluation, more deﬁnition and speciﬁcity is needed. What is necessaryin order to continue the forward momentum that they started is an encompassing theoretical approach concerned with theinteraction of these previously identiﬁed variables. We suggest that a signiﬁcant next step for performance evaluation researchis to evaluate how the outcomes of performance evaluation effect change in the subsequent performance of the ratee and therelational context of the organization.5. Theory and research directions The foregoing literature review categorizes the various research streams that make up the multifaceted context of performanceevaluation. There have been some recent efforts to provide a conceptual framework through which to better understand themechanisms by which performance evaluation processes are framed, inﬂuenced, and produce reactions (e.g., Arvey & Murphy,1998; Levy & Williams, 2004). Unfortunately, although useful in some respects, both efforts provided little discussion of thetheoretical foundations driving such process dynamics. If the multifaceted context serves as the backdrop for, or frames, performance evaluation processes and outcomes, then it isincumbent upon scholars to identify the key components of that context, as well as the factors that predict the formation ofsuch contexts. Additionally, there needs to be sufﬁcient theoretical grounding to the conceptualization to explain theinteractive process dynamics and outcomes. In the following sections, we propose a framework for this area of scientiﬁcinquiry, grounded in Affective Events Theory (Weiss & Cropanzano, 1996) and Emotion Cycle Theory (Hareli & Rafaeli, in press),and we discuss the implications of this conceptualization for future theory and research regarding the context of performanceevaluation.5.1. Model of the performance evaluation context and processes Performance evaluation is arguably an emotional experience. During appraisal, employees face a direct evaluation of their workby one or more raters to whom they are held accountable. These evaluations have potentially signiﬁcant ramiﬁcations for employeepsychological well-being, social status, and the continuation of employment within the organization (Dorfman, Stephan, &Loveland, 1986). Accordingly, performance evaluations represent a “high-stakes” encounter for employees. Characteristics of this encounter also may inﬂuence the potential for emotional reactions by employees within an organization.For example, in anticipation of the event, employees may exhibit apprehension and fear toward the performance evaluation, andappraisal outcomes may elicit an array of potential emotions (e.g., anger, joy, sadness), with ramiﬁcations for subsequent attitudesand behaviors. We propose that these emotional reactions have signiﬁcant implications for organizations as they directly inﬂuencethe relational and social context of work. Our proposed model is presented in Fig. 1.5.2. Ratee affective responses Individual affective responses can be readily described by Affective Events Theory (AET) (Weiss & Cropanzano, 1996).Speciﬁcally, AET predicts that individuals react emotionally, or affectively, to work events that inﬂuence an individuals well-beingand goal attainment. AET was intended as a complement, rather than a substitute, to cognitive theories of judgment inorganizations, and speciﬁcally proposes that organizational events, rather than system features, are proximal causes of affective
154 G.R. Ferris et al. / Human Resource Management Review 18 (2008) 146–163 Fig. 1. Performance evaluation context model.reactions in individuals. Basic assumptions of the model have been empirically supported, and suggest that the frameworkprovides a useful tool in evaluating individual work behavior (e.g., Wegge, van Dick, Fisher, West, & Dawson, 2006). AET makes a distinction between individual emotion and mood reactions. Speciﬁcally, emotions have a target (e.g., a verbalexchange with a co-worker), are relatively intense, and occur over a short duration. Meanwhile, moods lack a speciﬁc target, areless intense than emotions, and may be maintained for a longer duration (Weiss & Cropanzano, 1996). Emotions and moodsinﬂuence one another reciprocally. For example, an emotional response to an event (e.g., joy) may inﬂuence subsequent moods(e.g., positive mood). Relatedly, the disruption of a positive mood by a negative event may result in a highly emotional response(Weiss & Cropanzano, 1996). In the context of performance evaluation, a distinction can be made between affective responses to appraisal (as an event), andaffective responses to a system of performance evaluation. The act of performance evaluation implies an evaluative and discretejudgment of an individuals work. Accordingly, we would anticipate an emotional reaction since the appraisal event is discrete andhas signiﬁcant ramiﬁcations for an individual. Conversely, AET suggests that individuals will exhibit a pattern of moods in responseto organizational features to the extent that these features are relatively stable. This implies that an accountability system will not evoke an emotional response unless individuals are faced with potentialchange in their well-being and employment. Thus, we anticipate that the event of a performance evaluation will drive emotionalreactions by individuals rather than stable reactions to the system, per se. This is not to suggest that individuals fail to reactemotionally to performance evaluation, or accountability, systems. Rather, we suggest that individuals react emotionally tochanges in performance evaluation systems that bear on their psychological well-being and goal attainment. This importantboundary condition distinguishes emotion and mood reactions for individuals in the context of performance evaluation.5.3. Ratee affective outcomes Weiss and Cropanzano (1996) have argued that, once activated, emotional reactions play a functional and dominant role inindividual performance, because emotional states tend to preoccupy individuals from task performance. In AET, individualconsequences of emotional reactions include changes in work attitudes and behaviors through a process of primary and secondaryappraisal (cf. Lazarus, 1991). Speciﬁcally, after initial emotional reaction to an event stimulus (i.e., primary appraisal), Weiss andCropanzano (1996) argued that emotional reactions elicit a cognitive search process-focused on the causes of the event (i.e.,secondary appraisal). Although mentioned brieﬂy in the emotion literature as sense-making (e.g., Hareli & Rafaeli, in press), emotion theoriesgenerally do not consider the attribution process by which individuals consider causes of positive and negative emotion-generating events. In particular, Weiner (1985) argued that unexpected or novel events result in a primary emotional response ofhappiness or sadness. A secondary emotional response then derives from an appraisal concerning attributions made for theunexpected outcome. This attribution component has been previously neglected and may exert an inﬂuence on both the type ofaffective reaction and the severity of an affective reaction to performance evaluation. Weiner (1985) argued that individualsexperience behavioral and attitudinal outcomes based on attributions made for outcomes. His locus of causality (internal orexternal) and stability (stable or unstable over time) dimensions may be especially useful in explicating outcomes of emotionduring a secondary appraisal (cf. Martinko & Thomson, 1998). Further research is needed to understand how attributions inﬂuenceemotional reactions in performance evaluation. Nonetheless, the current discussion is concerned with emotional reactions to performance evaluation. Assuming no change inan emotion-evoking event, theory suggests that individuals ruminate, or dwell, on the event in a series of emotional reactions with
G.R. Ferris et al. / Human Resource Management Review 18 (2008) 146–163 155implications for performance and relationships (Beal, Weiss, Barros, & MacDermid, 2005). Speciﬁcally, Beal and his associates haveargued that emotional reactions consume cognitive resources contributing to decreased task performance. Furthermore, we suggest that the consumption of cognitive resources also may negatively inﬂuence an individuals ability toeffectively contribute to work relationships, especially in settings where others rely on the individual for their own performance.However, these mediated relational effects from task performance must be balanced against the direct effects of emotion onrelationships. For example, research has suggested that emotional reactions inﬂuence relationship satisfaction, quality in groups(Barsade, 2002), and service encounters (Pugh, 2001). Emotional reactions may stimulate individual withdrawal and a decrease inrelationship satisfaction due to their disruptive nature within a social system. Furthermore, the consumption of cognitive resourcesand changes in relationship satisfaction should decrease the quality of relationships within a social system, ceteris paribus. Theoretical support for this assertion derives from Sluss and Ashforth (2007), who developed a model of relational identity,deﬁned as “the nature of ones role relationship within an organization” (p. 11). They argued that individuals’ relational identity isbased on the function, hierarchy, status, and nature of complementarity of their role compared to a social context. Thus,performance evaluations may enact change in the relational identity beliefs of individuals in a social system. For example, high-performing salespersons may feel less valuable to a group or organization as leaders if their performance isappraised poorly or below that of salient peers. Individuals’ role and relational identity as high-performing leaders could bejeopardized by the outcomes of performance evaluation. Consequently, the performance evaluation has relational ramiﬁcationsthat can be evaluated cognitively. However, it is the emotional reaction that signals, or communicates, the social status of theindividual (Hareli & Rafaeli, in press), and enacts change in relationship satisfaction and subsequent quality for individuals. To summarize, certain organizational events, such as performance evaluation, result in emotional reactions by individuals.These emotional reactions inﬂuence the relational identity of ratees and detract cognitive resources from task performance andrelationships, modifying individuals’ subsequent relationship satisfaction and quality.5.4. Social context impacts Whereas AET readily describes the intra-individual process of emotion formation and outcomes, recent theory has highlightedthe inter-individual process of emotion transmission in social systems. An understanding of these processes is important becauseemotion transmissions have the potential to inﬂuence relationships embedded within the social systems of an organization. Inparticular, Hareli and Rafaeli (in press) argued that individual affective reactions by actors (i.e., ratees/subordinates) lead tosubsequent emotion effects in observers, which serve as feedback in subsequent periods to actors. Their theory conceptualizesemotional reactions as an interpersonal communication salient to direct observation by the target of the emotion, witnesses to thereaction, and third-parties. As noted earlier, Hareli and Rafaeli argued that emotions act as a signal regarding an individuals socialstanding. In the social context of performance evaluation, an emotional reaction by a ratee would likely be directed at the rater(s), andsalient to other employees, supervisors, and, potentially, clients. There are at least two considerations guiding the display andreception of emotion. First, theory suggests that emotions can be used as an impression management tactic (e.g., Grandey, 2000).Second, recent theory suggests that emotions are highly salient to observers who react to the emotion episode (e.g., Hareli &Rafaeli, in press). Taken together, we argue that emotional reactions are likely to be salient to an audience, and will enact a responseby that audience with behavioral and social context ramiﬁcations. The use of emotions as impression management was alluded to in work by Grandey (2000), who argued that individuals controltheir display of emotions (i.e., emotional labor) with the intention of inﬂuencing others’ reactions to those emotions. This so-calledsurface acting (Hochschild, 1983) may be viewed as an emotion regulation attempt. Consequently, emotions may be displayed withthe intention of guiding behavioral and/or attitudinal responses of others. Furthermore, the outcomes of performance evaluation are important to employees and supervisors because they provideadditional information on expectancies for their own appraisals, and also may modify individual social status and contributionswithin a group. Thus, we would anticipate a generally high degree of visibility for emotional reactions due to performanceevaluations. These visible emotional reactions elicit responses in others through several potential processes. Speciﬁcally, emotioncycle theory (Hareli & Rafaeli, in press) argues that individual emotion reactions elicit comparable responses in others throughthree processes: mimicking, emotion interpretation, and drawing inferences. Hatﬁeld, Cacioppo, and Rapson (1994) generally are credited with conceptualizing emotional contagion, which describes aconscious or unconscious mimicked response where the emotions of two or more parties converge. Recent research empiricallyhas highlighted the role of emotional contagion within social systems (e.g., Barger & Grandey, 2006; Barsade, 2002; Pugh, 2001),demonstrating that emotions are transferred from person to person. For example, Barger and Grandey (2006) found that emotions were transferred between strangers in a service encounter. Next,Barsade (2002) found evidence for contagion effects between strangers in an experimental setting. In a performance evaluationcontext, this evidence suggests that ratee emotional reactions inﬂuence the emotional or affective state of the rater(s), and likelyothers within a social system. However, it is critical to note that emotional contagion assumes the convergence of similar emotions. Furthermore, susceptibility to emotional contagion varies on the basis of individual differences. For example, Verbeke (1997)found evidence that contagion varies on the basis of empathy and individual charisma. Ilies, Wagner, and Morgeson (2007) foundthat individuals differed in their susceptibility to contagion. Congruent with AET (Weiss & Cropanzano, 1996), we suggest thatindividual dispositions, such as agreeableness and neuroticism, might be especially important in facilitating or inhibiting emotioncontagion. The frequency of exposure to emotional events also may inﬂuence susceptibility to emotion contagion. For example,
156 G.R. Ferris et al. / Human Resource Management Review 18 (2008) 146–163raters in a performance evaluation process may be more or less susceptible to emotion contagion based on the novelty andfrequency of the emotion stimulating event (i.e., the performance evaluation). Emotion interpretation is the second process through which others respond to actor (i.e., ratee) emotional reactions. Hareli andRafaeli (in press) argued that emotion interpretation involves a cognitive evaluation of the actors emotions, which may result inconvergent or divergent responses. They suggested that emotional reactions elicit relational scripts (Lazarus, 1991) that provideinterpretation and response. We would add that emotional reactions may upset the social balance of an organization by changingthe negotiated structure of individual social exchanges (cf., Blau, 1964). That is, emotional reactions deemed as inappropriate mayresult in punitive actions by others in a social system. For example, a co-worker viewing a visibly angry ratee may be incited toanger not at the raters, but at the ratee for disrupting the social balance and mood of the group. This effect is likely to be enhancedthrough the third process of interpersonal emotion response, drawing inferences. Hareli and Rafaeli (in press) suggested that individuals draw inferences regarding the causes and intended effects of emotionalreactions by actors (i.e., the ratee). Although the interpretation of emotions is grounded in heuristics or scripts, emotion inferencesare grounded in cognitive processes that, we would suggest, involve an attributional component (e.g., Weiner, 1985). Speciﬁcally,individuals have a vested interest in ascertaining the causes of behavior in order to guide appropriate future expectancies. Thus, aco-worker might attribute a visibly happy reaction to a positive performance evaluation. The positive appraisal would potentiallyhave signiﬁcant ramiﬁcations for status, power, and credibility (Hareli & Rafaeli, in press). Positive emotions also might signalhigher status and greater power in the organization. However, the highly (perhaps overly) salient display of emotions also mightdecrease the credibility of the actor (i.e., ratee). Thus, cognitive processes also play an important role in the interpretation andresponse to emotional displays. In summary, emotional reactions generally are salient within a social system. Individuals respond to actor (i.e., ratee) emotionalreactions through mimicking, interpretation, and inferences. These reactions have direct and mediated relational effects. As notedearlier, mimicking is likely to elicit a similar emotional response in observers through contagion processes. Thus, we would expectobservers to experience similar emotions that can reinforce actor emotions. Such a response would be characteristic of empathy(Hareli & Rafaeli, in press), and may positively inﬂuence relationship satisfaction for the actor. Speciﬁcally, we would anticipate contagion reactions, even negative emotional contagion, to result in enhanced relationshipsatisfaction for the actor and observer based on a system of reciprocity found in social exchange theory (e.g., Blau, 1964). That is,relationship satisfaction is likely to be optimized when each party can go to the other and receive similar social resources, orsupport. Thus, we suggest that emotion contagion and a visible response of empathy is likely to result in enhanced relationshipsatisfaction. Indirect effects of affective responses are likely to be manifested through the broad categories of avoidance or approachbehaviors (Hareli & Rafaeli, in press). Approach refers to a category of behaviors that engage two people toward afﬁliation, whereasavoidance refers to a category of behaviors that uncouple two people from afﬁliation (cf. Russell & Mehrabian, 1978). Approachbehaviors should result in a higher degree of relationship satisfaction, whereas avoidance behaviors generally reduce relationshipsatisfaction (assuming the individuals like each other to begin). Work by Russell and Mehrabian (1978) found that approach andavoidance afﬁliation behaviors were based on the positive arousal potential from others. In short, individuals are motivated to formrelationships, or afﬁliate, with those who make them feel better. Thus, positive emotional reactions should result in enhancedrelationship satisfaction through approach behaviors, whereas negative emotional reactions may result in decreased relationshipsatisfaction through avoidance behaviors. Emotion cycle theory (Hareli & Rafaeli, in press) suggests that moderators inﬂuence relational outcomes. For example, positiveemotional reactions by one individual may signal a status change due to a positive performance evaluation (e.g., Sluss and Ashforth,2007). To the extent that this status change results in imbalance with others in the organization, we might expect them to exhibitavoidance behaviors. Furthermore, the frequency of emotional display might, by itself, result in avoidance if constant, evenpleasant, emotional reactions detract from an individuals task performance. To summarize, emotional reactions by an actor inﬂuence their relationship satisfaction directly and through approach andavoidance behaviors. These emotional reactions are salient also to observers, and inﬂuence relationship satisfaction directly by thetype of observer emotional reaction or through avoidance or approach behaviors that modify relationship satisfaction. As notedearlier, relationship satisfaction drives the type and quality of individual contributions toward relationship quality. Theoretically,individuals gauge the cost of a relationship relative to the beneﬁts they will receive from that relationship (Blau, 1964). Thus,decreases in relationship satisfaction should lead to decreases in relationship quality as individuals modify their social inputs.Conversely, increases in relationship satisfaction should lead to increases in relationship quality as individuals contribute more tothe dyadic social exchange.5.5. Moderators of the performance evaluation–affective reaction relationship With the primary model of performance evaluation outcomes described, we now turn to a discussion of individual, relational,and accountability features that interact with performance evaluation outcomes to inﬂuence emotional reactions. First, AETproposes that individual dispositions inﬂuence affective reactions to events interactively (Weiss & Cropanzano, 1996). For example,dispositional affect, should inﬂuence negative and positive reactions to events, because it acts as a perceptual ﬁlter for individuals.Likewise, dispositional optimism inﬂuences individual expectancies of good versus bad outcomes. Thus, low levels of optimismmay inﬂuence pessimistic evaluations of future expectancies from performance evaluation, and elicit strong negative emotions.Conversely, high levels of optimism may inﬂuence more optimistic evaluations of future expectancies that elicit positive emotions.
G.R. Ferris et al. / Human Resource Management Review 18 (2008) 146–163 157 Emotional intelligence (e.g., Mayer, Caruso, & Salovey, 1999) should also theoretically and logically inﬂuence outcomes of aperformance evaluation. Emotional intelligence refers to an individual ability that enables individuals to reﬂectively regulate onesown emotions, understand the emotions of oneself and others, assimilate emotions in thought, and perceive and express emotion(Mayer et al., 1999). In a performance evaluation setting, emotional intelligence may play an important role as a moderator ofemotional reactions. In particular, we would expect emotional intelligence of both the rater/supervisor and ratee/subordinate to beinversely related to the type and severity of an affective reaction from performance evaluations. That is, as emotional intelligenceincreases in the rater/supervisor, the outcomes of a performance evaluation will appear less severe to the ratee/subordinate,leading to a tempered emotional response. In addition, we anticipate a tempered affective reaction to performance evaluations asratee/subordinate emotional intelligence increases. Individual differences are not the only moderators of interest. As noted earlier, accountability exerts an important inﬂuence inguiding expectations for behavior and framing the context of performance evaluation (e.g., Frink & Klimoski, 1998). As an inﬂuencefor outcomes, accountability may act as an important moderator between a performance evaluation and the affective reaction ofthe actor. In particular, accountability theory informs qualities of the actual evaluation, the relationship between a rater/supervisorand ratee/subordinate, and expectancies of the performance evaluation by both parties. Ratees relationships with raters also are likely to shape their emotional reactions to performance evaluation. Frink andKlimoski (1998) described this relational context between employees being held accountable and evaluative others, or thoseholding employees accountable during a performance evaluation. Speciﬁcally, we anticipate a non-linear interaction between thequality of the relationship and the emotional reaction to a performance evaluation. Assuming a poor level of relationship qualitywith a rater, a ratee may feel the freedom to emotionally respond in a highly salient and negative manner for a poor performanceevaluation. Likewise, with a high level of relationship quality with a rater, a ratee may feel betrayed and respond emotionally to apoor performance appraisal. At intermediate levels of relationship quality, we anticipate a less salient emotional response toperformance evaluation. Frink and Klimoski (1998) discussed at length the expectations held by employees being evaluated (i.e., ratee/subordinateexpectancies) and those holding employees accountable (i.e., rater/supervisor expectancies). In particular, we expect rateeexpectancies to moderate the emotional reaction and anticipate the emotional reaction to differ in type and degree based on thelevel of discrepancy between rater and ratee expectancies. Thus, ratees who are surprised at the performance evaluation outcomeslikely will exhibit different emotions than ratees who expect the appraisal outcome. Finally, the qualities of the performance evaluation should inﬂuence the type and degree of emotional reaction by the ratee. Inaccountability theory (Frink & Klimoski, 1998), these qualities include the likelihood of evaluation and the standards with whichemployees are held accountable. The structure of a performance evaluation may act as a proxy for these concepts. For example,highly structured performance evaluations may elicit less of an emotional response by the ratee since the rater may have lessperceived control. Likewise, frequent performance evaluations may elicit less of an emotional response as they become routinized.5.6. Accountability system features The fact that the performance evaluation system exists in a political, emotional and relationship-dependent environment doesnothing to simplify its execution. Ferris et al. (1995) highlighted several areas of “tension” that arise in any accountability system,offering some insight into the “poison” that often must be picked by the organization. Earlier in this paper, we presented the notionthat performance evaluation is an accountability mechanism, and as such, these tension points need to be considered in ourdiscussion. There is no doubt that a performance evaluation system, clearly deﬁned and linked to valued outcomes, offers signiﬁcantbehavioral control over the evaluated employees. However, the higher this control, the larger the cost. Reactance represents theemployees attempt to restore the freedom of choice that was lost through the imposition of monitoring and control system. Theresponse may include anything from illegal activity, to simply performing poorly, or tolerating the inappropriate behavior ofothers. Although the vast majority of the employees will not resort to illegal or dysfunctional behavior, there is the potential forresistance and resentment associated with the organization restricting their freedom and self-control (Ferris et al., 1995). Even with appropriate behavior, another tradeoff for the use of a monitoring and evaluation system is that of compliance versusinternalization. Simply said, this is where employees perform because they have to, as opposed to performing because they wantto. The compliance versus internalization balance is one that may be overcome if the evaluation system helps instill, in theemployees, a sense of appreciation and buy-in for the evaluated behaviors. If this occurs, the organization may respond with areduced accountability requirement (i.e., which agency theorists would argue simply opens the door again for self-servingbehaviors from the agents, which is the reason why the accountability system was imposed in the ﬁrst place!). Other dilemmas include the loss of creativity and pro-social behaviors in exchange for conformity. Although Trevino andYoungblood (1990) demonstrated that accountability may increase ethical behavior through conformity, the same conformity hasbeen shown by others to reduce risk taking (Weigold & Schlenker, 1991), “going the extra mile” (Schnake & Dumler, 1993), andcreativity (Staw & Boettger, 1990). In fact, people do respond to the pressures of the context and the system in place, but these forces often may put them at oddswith themselves and the intended behaviors of the organization. Performance evaluation systems, like other accountabilitymechanisms, can pit individual values against organizational values. Similarly, they can intentionally or unintentionally emphasizeoutcome-focused (i.e., “do whatever you have to do to get the job done”) or process-focused (i.e., “how you get the job donematters”) accountability. Finally, in todays work environment, employees often ﬁnd themselves accountable to multiple audiences
158 G.R. Ferris et al. / Human Resource Management Review 18 (2008) 146–163(e.g., supervisors, peers, others), with potentially different expectations for performance. As much as expectations for performancerepresent a basis for performance evaluations, employees often are in a dilemma of making signiﬁcant tradeoffs. All this is not to say that such problems are directly associated with increased levels of accountability. However, there is soundrationale to suggest that, “more is not necessarily better.” The consequences, both positive and negative, of increased controlresulting from systems like performance evaluation must be intentionally considered. For example, it may be argued that there is anon-linear effect of the level of accountability on outcomes like performance, citizenship, well-being, and unethical behavior; thatis, there is an optimal level of accountability, which will produce the correct work behavior. Researchers and practitioners will bewell served to continue to explore the proper balance and degree of accountability, as they design and implement systems ofperformance evaluation.5.7. Work relationship features As we have characterized in the present paper, the supervisor–subordinate relationship represents a critically importantcomponent of the context of performance evaluation that we need to better understand. Interestingly, as important as workrelationships are to the organizational sciences, until quite recently, there was very little research on the topic in this ﬁeld beyondwhat we could garner from the theory and research in LMX. Perhaps that is why the little research that has employed workrelationship quality variables in performance evaluation research contained weak theoretical underpinnings about howrelationships drive rating processes, and used the current version of the LMX measure to assess ‘work relationship quality’ (e.g.,Judge & Ferris, 1993). Fortunately, the ﬁeld is witnessing a ‘relationship revolution’ of sorts, with several pockets of very interesting and innovativework being proposed and conducted on work relationships just within the past few years. Some of this interest has been generatedby the very sound scholarship that has been taking place for years in social psychology on “relationship science” (e.g., Berscheid,1999). Although this work has been focused at interpersonal relationships in everyday life, and without speciﬁc implications forwork organization relationships, the relevance of that research has spilled over and generated interest in the organizationalsciences. Perhaps that work and the beginnings of the new ‘positive psychology’ movement have stimulated work in ‘positiveorganizational scholarship’ and “high-quality connections” initially (Dutton & Heaphy, 2003), followed up by an emphasis on‘positive work relationships’ more recently (Dutton & Ragins, 2007). This work collectively appears to provide much useful thoughtfor building a more informed understanding of work relationships, and their process dynamics and outcomes. Indeed, very recentconceptual work has discussed the favorable effects of high-quality or positive work relationships for health and well-being, whichinclude both psychological and physiological effects (e.g., Heaphy & Dutton, 2008). At present, more excitement and anticipationthan conclusive evidence characterizes this new research on positive and high-quality work relationships, but the zeitgeist mostdeﬁnitely favors continued interest in work relationships today. Furthermore, even though some theory and research has considered what might be the underlying dimensions of workrelationships, much more is needed. For example, Liden, Sparrowe, and Wayne (1997) reported four dimensions that comprisedthe LMX relationship to be affect, loyalty, contribution, and professional respect. Also, Settoon and Mossholder (2002) investigatedrelationship content and context effects on work outcomes, and identiﬁed the three dimensions of co-worker support, trust, andperspective taking (i.e., which appears to be synonymous with empathy), which they found predictive of interpersonal citizenshipbehavior. Certainly, this research provides a good beginning, but in order to identify a more extensive set of critical relationshipdimensions, such as trust, support, empathy, distance, and so forth, much theory and research is needed to more preciselydelineate the speciﬁc dimensions and their interconnections, as well as their speciﬁc and collective implications for performanceevaluations.6. Supervisor and subordinate characteristics and behaviors The characteristics and qualities possessed by both supervisors and subordinates are important to consider in the emergenceand materialization of any work context, because such factors help to form or shape the nature of such contexts. Indeed, a numberof individual difference characteristics of supervisors and subordinates have been suggested in previous models of dyadicinteractions in human resources decisions (e.g., Ferris & Judge, 1991). However, we provide suggestions for some new areas thatresearch should consider in the future, and we highlight qualities and characteristics of dyadic relationship members that cancontribute to more effective social and relationship contexts. Here, we particularly focus on political skill (i.e., as an example of thecategory of social effectiveness constructs), reputation (i.e., also as a framing or context variable), and affect and emotiondemonstration and regulation in dyadic interactions.6.1. Political skill Political skill has been deﬁned as “the ability to effectively understand others at work, and to use such knowledge to inﬂuenceothers to act in ways that enhance ones personal and/or organizational objectives” (Ferris et al., 2005, p. 127). As such, political skillreﬂects an interpersonal style that combines social astuteness with the capacity to adjust, adapt, and calibrate behavior to differentcontextual demands, and to do so employing a manner that is genuine and sincere, inspires conﬁdence and trust, and results ineffective inﬂuence over others at work. Individuals high in political skill tend to reﬂect a sense of calm self-conﬁdence, control, and
G.R. Ferris et al. / Human Resource Management Review 18 (2008) 146–163 159personal security that attracts others, instills in them feelings of comfort, and also contribute to positive affective reactions andtrust (Ferris, Treadway, Perrewé, Brouer, Douglas, & Lux, 2007). Theory and research have argued that politically skilled individuals strategically select the most appropriate methods ofinﬂuence for particular contexts, and then effectively execute such inﬂuence tactics using a style that ensures interpersonal goalaccomplishment. As such, it has been argued by Ferris and colleagues that political skill should demonstrate effects on such socialcontext outcomes as supervisor ratings of subordinate performance, and some research to date has provided support for thesearguments (Ferris et al., 2007). Concerning political skill main effects, Ferris et al. (2005) reported, in two studies, that political skill signiﬁcantly predictedsupervisor ratings of subordinate performance and effectiveness. Additionally, Semadar, Robins, and Ferris (2006) found thatpolitical skill emerged as the strongest predictor of managerial performance ratings than other social competence constructs,including self-monitoring, leadership self-efﬁcacy, and emotional intelligence. Also, in a two-study constructive replication,Jawahar, Meurs, Ferris, and Hochwarter (2008) found that political skill was a stronger predictor of contextual performance thanwas self-efﬁcacy. Other evidence of political skill main effects can be found in Liu, Ferris, Zinko, Perrewé, Weitz, and Xu (2007), who reported, in afour-study investigation, that signiﬁcant relationships between employee political skill and supervisor-rated employee jobperformance were mediated by employee reputation. Also demonstrating mediated effects, Kolodinsky et al. (2007) demonstratedthat political skill inﬂuenced job performance ratings through speciﬁc intermediate linkages (i.e., perceived similarity and affect)suggested by the Ferris and Judge (1991) framework. We suggest here that political skill might be a useful characteristic, which has the potential of contributing to a more favorablesocial, emotional, political, and relationship context within which performance evaluation occurs. The foregoing early evidence,investigating subordinate political skill, seems to be promising. However, the political skill of supervisors in the performanceevaluation context has yet to be examined, and could prove enlightening.6.2. Reputation Characterized as a “complex combination of salient personal characteristics and accomplishments, demonstrated behavior, andintended images presented over some period of time as observed directly and/or as reported from secondary sources” (Ferris, Blass,Douglas, Kolodinsky, & Treadway, 2003, p. 213), the role of personal reputation in organizations recently has been receivingincreased research attention. Individuals who have developed, and enjoy, favorable reputations tend to be perceived by others asreﬂecting a persona characterized by greater legitimacy, competency, trustworthiness, and status. Furthermore, such collectiveperceptions allow for the accumulation of power, inﬂuence, autonomy, and latitude (Ferris et al., 2003), which all are factors thatpromote inﬂuence effectiveness. Although by no means the only outcome of a favorable reputation, certainly a prominent consequence of reputation favorabilityis increased job performance, which typically is assessed as the subjective ratings provided by immediate supervisors. As we haveseen in recent years, supervisory ratings of employee performance have been predicted by a number of job-relevant variables, butalso can be subject to inﬂuence, bias, and distortion. This raises the issue of precisely what role reputation might play as a potentialfactor that operates within the complex performance evaluation context. Because it has been characterized as a multifacetedcollection of different pieces of information aggregated together in some sort of collective whole, it might represent a ‘signal’(Spence, 1974) sent to raters regarding attributes, experience, expected behavior, and so forth that affects ratings in complex ways. Some initial research has hypothesized and found evidence for both main and interaction effects of reputation within theperformance evaluation context. Liu et al. (2007) reported on a four-study research plan to examine the hypothesized chainreﬂecting that personality is antecedent to political skill, which then affects job performance ratings by supervisors through themediating factor of reputation. Collectively, these studies demonstrated support for recent theoretical developments in bothpolitical skill and reputation, suggesting that personality serves as an antecedent of political skill, and that reputation mediates therelationship between political skill and job performance ratings. Subordinate reputation also can serve as a moderator in the process of trying to better understand the performance evaluationcontext, and Hochwarter, Ferris, Zinko, Arnell, and James (2007) recently reported some interesting results in this regard. Theyfound support for reputation as a moderator of the relationships between political behavior and job performance ratings (inaddition to work outcomes of uncertainty and emotional exhaustion). For subordinates with favorable reputations, exhibitingpolitical behavior was associated with increased job performance ratings (and also decreased uncertainty and emotionalexhaustion). Alternatively, for subordinates possessing unfavorable reputations, the demonstration of political behavior wasassociated with decreased job performance ratings (and with increased uncertainty and emotional exhaustion). Future research should investigate the potentially important role of subordinate reputation in the performance evaluationprocess, both in terms of direct effects on supervisor ratings, and also as a moderator of the relationship between other behaviorsand performance ratings. The role of supervisor reputation, in conjunction with subordinate reputation, in future research alsomight be worthy of investigation in order to shed greater light on the context, processes, and outcomes of performance evaluation.6.3. Affect and emotion demonstration and regulation Future research on affect and emotion seems warranted to help better understand the role it plays in the performanceevaluation context. Although considerable research has been conducted on the effects of liking in the performance evaluation
160 G.R. Ferris et al. / Human Resource Management Review 18 (2008) 146–163process, a much deeper understanding is needed concerning why a supervisor develops like or dislike for a particular subordinate,beyond the work that has used the similarity-attraction theory view, whereby similarity is argued to breed attraction and affect(Byrne, 1971). However, the results of the Kolodinsky et al. (2007) study might help shed some initial light on this. Their results indicated thatsupervisors may develop positive affect or liking for subordinates who demonstrate situationally appropriate behavior aroundothers in a socially astute manner. One such behavior, for example, is the use of the inﬂuence tactic of rationality, whereby thesubordinate attempts to inﬂuence through the use of logical persuasion, factual evidence, and thorough explanations. That is, thepolitical skill x rationality tactic interaction suggested that the skillful presentation of the rationality inﬂuence tactic contributed togreater supervisor liking of subordinates. Thus, Kolodinsky et al. noted that if subordinates demonstrate behavior that is consistentwith the supervisory expectations, subordinate politically skilled behavior most likely will be instrumental regarding relationalperceptions and, ultimately, in helping supervisors meet their own goals. Additional ideas for future research on the role affect and emotion plays in performance evaluation can be gleaned from theexcellent review and conceptualization presented by Arvey et al. (1998). As Arvey et al. suggested, the demonstration of emotion bysubordinates tends to trigger emotions in supervisors rating their performance, which in turn may inﬂuence the cognitiveprocesses supervisors use to formulate ratings. Furthermore, Arvey et al. suggested that besides affecting job performance ratings,subordinate emotional display also may inﬂuence the supervisors’ assessments of subordinate ‘ﬁt’ with the job and organization,drawing upon the ASA framework (e.g., Schneider, 1987). Such ‘ﬁt’ assessments based on emotional display might be referenced byany formal or informal expectations or norms regarding emotional behavior in the job and/or organization context. Arvey et al. (1998) suggested future research should investigate the potential effects of different types of emotions on differentjob performance criteria, with the argument that positive emotional demonstration would most likely be more strongly andpositively related to contextual performance dimensions. Also, we recommend research that investigates the distinction betweenspontaneous emotional displays versus emotion demonstration that is calculated, and strategically presented, as a form ofinterpersonal inﬂuence (e.g., Grandey, 2000). We also recommend that research examine the range of emotion presented, and its potential non-linear effects on outcomes. Inthis case, although prior research reviewed earlier in the paper regarding affect and performance ratings demonstrated positivelinear effects of the ratings provided by supervisors, it might be the case that a non-linear relationship exists between subordinatedemonstration of emotion and actual performance and effectiveness (i.e., not just supervisory subjective ratings), suggesting anoptimal level of affect/emotion in such contexts.7. Theory development and integration As noted by Levy and Williams (2004), contextual factors represent an oft-neglected area of performance evaluation. Priorreviews of performance evaluation (e.g., Arvey & Murphy, 1998; Levy & Williams, 2004) have expounded many of the distaland proximal features inﬂuencing processes and outcomes of interest. However, the utility of these valuable contributionslargely is constrained by a lack of unifying theoretical frameworks on which to base speciﬁc assumptions and boundaryconditions. Consequently, this paper has focused on two contributions that serve to address the current status of performance evaluationtheory and research. First, we argue that accountability theory (Frink & Klimoski, 1998; Frink et al., 2008) represents a viable andextant tool for evaluating performance evaluation processes from a cognitive theoretical perspective. Second, we propose that thesocial and relational outcomes of performance evaluation are effectively described by theories of intrapersonal (Weiss &Cropanzano, 1996) and interpersonal (Hareli & Rafaeli, in press) affect. In this section, we brieﬂy review the four categories of socialcontext proposed by Levy and Williams (2004), and integrate them within accountability and affect theoretical frameworks. Thesecategories include distal variables, process proximal variables, structural proximal variables, and rater and ratee behavior.7.1. Distal variables Levy and Williams (2004) referred to distal variables as macro-level factors that inﬂuence the performance evaluation system.These are classiﬁed as environmental factors in accountability theory (Frink et al., 2008), and include social norms, regulatorymandates, culture and climate effects, strategy, and network relationships. Environmental factors generally shape performanceevaluations as antecedents to formal appraisal systems and informal appraisal norms. Theoretically, environmental factors shouldinﬂuence affective reactions only when they modify performance evaluation systems or inﬂuence perceptions of change.Congruent with prior research (Frink et al., 2008; Levy & Williams, 2004), we anticipate these effects on individuals to be largelymediated through structural and process proximal variables.7.2. Process and structural proximal variables Structural proximal variables are performance evaluation features that inﬂuence the nature and content of the performanceevaluation (Levy & Williams, 2004). Examples include performance standards, the frequency of evaluation, legitimacy of theevaluation, and evaluation system features. Relatedly, process proximal variables inﬂuence the conduct of a performanceevaluation (Levy & Williams, 2004). Examples of process proximal variables include the rater–ratee relationship, performanceexpectancies, task characteristics, and rater and ratee affect.