360° Feedback: Accuracy, Reactions, and...

13
Journal of Applied Psychology 2001, Vol. 86, No. 5. 930-942 Copyright 2001 by the American Psychological Association, Inc. 0021-9010/01/S5.00 DOI: 10.1037//0021-9010.86.5.930 360° Feedback: Accuracy, Reactions, and Perceptions of Usefulness Joan F. Brett and Leanne E. Atwater Arizona State University West This study examined how 360° feedback ratings and self-other rating discrepancies related to reactions to feedback, perceptions of feedback accuracy, perceived usefulness of the feedback, and recipients' receptivity to development. The results indicated that less favorable ratings were related to beliefs that feedback was less accurate and to negative reactions. Negative reactions and perceptions that feedback was less accurate were related to beliefs that the feedback was less useful. Those who found feedback less useful were perceived by a facilitator as less development-focused. Goal orientation did not moderate the relationship between ratings and perceptions of accuracy or reactions to feedback. Goal orientation was related to perceptions of usefulness of the process several weeks after receipt of feedback. The results question widely held assumptions about 360° feedback that negative and discrepant feedback motivates positive change. The use of 360° feedback, the process in which subordinates, peers, and bosses provide anonymous feedback to managers, con- tinues to grow in popularity. Antonioni (1996) reported that an estimated 25% of companies use some type of upward or 360° feedback process. One of the reasons for its increased popularity is the belief that developmental feedback can improve performance. Studies have reported improvements in overall performance fol- lowing this type of feedback (e.g., Atwater, Roush, & Fischthal, 1995; Reilly, Smither, & Vasilopoulos, 1996). Positive results have been shown to be most notable for those who initially rated themselves higher than the ratings they received from others (Johnson & Ferstl, 1999). An assumption of the 360° feedback process is that negative or discrepant feedback (ratings from others that are lower than self- ratings) will create awareness and motivate individuals to change behaviors. Although studies such as those presented above have shown that this is sometimes the result, research on performance appraisal feedback suggests that positive results are only one possible outcome. Kluger and DeNisi (1996) concluded that in more than one third of the cases, performance appraisal feedback actually resulted in decreased performance. Rather than being motivated to change, managers who received negative feedback were often discouraged and not motivated to improve. These negative reactions were particularly the case when feedback con- cerned personal characteristics rather than task behaviors (Kluger & DeNisi, 1996). Do these same reactions to negative performance Joan F. Brett and Leanne E. Atwater, School of Management, Arizona State University West. We thank Ellen Jackofsky, Don VandeWalle, David Waldman, and Fran Yammarino for their comments on this research. We also thank Paula Hill of the Southern Methodist University Business Leadership Center and Tom Daniels of Personnel Decision International for their support on this project. Correspondence concerning this article should be addressed to Joan F. Brett, School of Management, Arizona State University West, P.O. Box 37100, Phoenix, Arizona 85069-7100. Electronic mail may be sent to [email protected]. appraisal feedback occur when individuals receive negative 360° feedback that is confidential, developmental, and not used for reward or promotion decisions? Is negative 360° feedback that highlights weaknesses or development needs perceived as con- structive? Or is negative feedback, regardless of its purpose, poorly received? Although there is a growing body of research on 360° feedback, little is known about whether 360° developmental feedback has effects similar to performance appraisal feedback in terms of reactions to feedback. A better understanding of how managers perceive and react to feedback is needed if 360° feed- back is to have the expected value to individuals and organizations. The purpose of this study was to examine the relationships between 360° feedback ratings and self-other rating discrepancies and reactions to feedback, perceptions of feedback accuracy and usefulness, and recipients' receptivity to development following feedback. We also investigated an individual-differences variable, goal orientation, which we believed would moderate the relation- ships between feedback and reactions, accuracy, and usefulness (Button, Mathieu, & Zajac, 1996; Farr, Hoffmann, & Ringenbach, 1993). Overview The relationships among variables examined in this study are presented in Figure 1. This is not a causal model but rather an exploratory model of hypothesized relationships. We proposed that both rating favorability and the extent to which self-ratings and others' ratings diverge will be related to perceptions of feedback accuracy and to positive and negative reactions to feedback. We expected perceptions of feedback accuracy to be positively related to positive reactions (e.g., pleased) and negatively related to neg- ative reactions (e.g., anger). Both reactions and accuracy will, in turn, be related to perceptions of the usefulness of the feedback. In addition, usefulness will be related to the feedback recipient's responsiveness to development. The feedback recipient's goal orientation, a mental framework influencing how an individual interprets and responds to achieve- ment situations, was expected to serve as a moderator. Dweck (1986) and Dweck and Leggett (1988) identified two broad classes 930

Transcript of 360° Feedback: Accuracy, Reactions, and...

Page 1: 360° Feedback: Accuracy, Reactions, and …eportfolio.lib.ksu.edu.tw/~T093000259/repository/fetch...360 Feedback: Accuracy, Reactions, and Perceptions of Usefulness Joan F. Brett

Journal of Applied Psychology2001, Vol. 86, No. 5. 930-942

Copyright 2001 by the American Psychological Association, Inc.0021-9010/01/S5.00 DOI: 10.1037//0021-9010.86.5.930

360° Feedback: Accuracy, Reactions, and Perceptions of Usefulness

Joan F. Brett and Leanne E. AtwaterArizona State University West

This study examined how 360° feedback ratings and self-other rating discrepancies related to reactionsto feedback, perceptions of feedback accuracy, perceived usefulness of the feedback, and recipients'receptivity to development. The results indicated that less favorable ratings were related to beliefs thatfeedback was less accurate and to negative reactions. Negative reactions and perceptions that feedbackwas less accurate were related to beliefs that the feedback was less useful. Those who found feedback lessuseful were perceived by a facilitator as less development-focused. Goal orientation did not moderate therelationship between ratings and perceptions of accuracy or reactions to feedback. Goal orientation wasrelated to perceptions of usefulness of the process several weeks after receipt of feedback. The resultsquestion widely held assumptions about 360° feedback that negative and discrepant feedback motivatespositive change.

The use of 360° feedback, the process in which subordinates,peers, and bosses provide anonymous feedback to managers, con-tinues to grow in popularity. Antonioni (1996) reported that anestimated 25% of companies use some type of upward or 360°feedback process. One of the reasons for its increased popularity isthe belief that developmental feedback can improve performance.Studies have reported improvements in overall performance fol-lowing this type of feedback (e.g., Atwater, Roush, & Fischthal,1995; Reilly, Smither, & Vasilopoulos, 1996). Positive resultshave been shown to be most notable for those who initially ratedthemselves higher than the ratings they received from others(Johnson & Ferstl, 1999).

An assumption of the 360° feedback process is that negative ordiscrepant feedback (ratings from others that are lower than self-ratings) will create awareness and motivate individuals to changebehaviors. Although studies such as those presented above haveshown that this is sometimes the result, research on performanceappraisal feedback suggests that positive results are only onepossible outcome. Kluger and DeNisi (1996) concluded that inmore than one third of the cases, performance appraisal feedbackactually resulted in decreased performance. Rather than beingmotivated to change, managers who received negative feedbackwere often discouraged and not motivated to improve. Thesenegative reactions were particularly the case when feedback con-cerned personal characteristics rather than task behaviors (Kluger& DeNisi, 1996). Do these same reactions to negative performance

Joan F. Brett and Leanne E. Atwater, School of Management, ArizonaState University West.

We thank Ellen Jackofsky, Don VandeWalle, David Waldman, and FranYammarino for their comments on this research. We also thank Paula Hillof the Southern Methodist University Business Leadership Center and TomDaniels of Personnel Decision International for their support on thisproject.

Correspondence concerning this article should be addressed to Joan F.Brett, School of Management, Arizona State University West, P.O. Box37100, Phoenix, Arizona 85069-7100. Electronic mail may be sent [email protected].

appraisal feedback occur when individuals receive negative 360°feedback that is confidential, developmental, and not used forreward or promotion decisions? Is negative 360° feedback thathighlights weaknesses or development needs perceived as con-structive? Or is negative feedback, regardless of its purpose,poorly received? Although there is a growing body of research on360° feedback, little is known about whether 360° developmentalfeedback has effects similar to performance appraisal feedback interms of reactions to feedback. A better understanding of howmanagers perceive and react to feedback is needed if 360° feed-back is to have the expected value to individuals and organizations.

The purpose of this study was to examine the relationshipsbetween 360° feedback ratings and self-other rating discrepanciesand reactions to feedback, perceptions of feedback accuracy andusefulness, and recipients' receptivity to development followingfeedback. We also investigated an individual-differences variable,goal orientation, which we believed would moderate the relation-ships between feedback and reactions, accuracy, and usefulness(Button, Mathieu, & Zajac, 1996; Farr, Hoffmann, & Ringenbach,1993).

Overview

The relationships among variables examined in this study arepresented in Figure 1. This is not a causal model but rather anexploratory model of hypothesized relationships. We proposed thatboth rating favorability and the extent to which self-ratings andothers' ratings diverge will be related to perceptions of feedbackaccuracy and to positive and negative reactions to feedback. Weexpected perceptions of feedback accuracy to be positively relatedto positive reactions (e.g., pleased) and negatively related to neg-ative reactions (e.g., anger). Both reactions and accuracy will, inturn, be related to perceptions of the usefulness of the feedback. Inaddition, usefulness will be related to the feedback recipient'sresponsiveness to development.

The feedback recipient's goal orientation, a mental frameworkinfluencing how an individual interprets and responds to achieve-ment situations, was expected to serve as a moderator. Dweck(1986) and Dweck and Leggett (1988) identified two broad classes

930

Page 2: 360° Feedback: Accuracy, Reactions, and …eportfolio.lib.ksu.edu.tw/~T093000259/repository/fetch...360 Feedback: Accuracy, Reactions, and Perceptions of Usefulness Joan F. Brett

360° FEEDBACK 931

13 "82 "̂E o>o. m-i s P

erce

ived

Use

fuln

ess

ofFe

edba

ckTi

me

2

0) =

fes

oC5

O.Ic

aI

Page 3: 360° Feedback: Accuracy, Reactions, and …eportfolio.lib.ksu.edu.tw/~T093000259/repository/fetch...360 Feedback: Accuracy, Reactions, and Perceptions of Usefulness Joan F. Brett

932 BRETT AND ATWATER

of underlying goal orientations: (a) a learning goal orientation, adisposition to develop competence by acquiring new skills andmastering new situations, and (b) a performance goal orientation,a disposition to demonstrate and validate one's competence byseeking favorable judgments and avoiding negative judgments.Goal orientation was expected to moderate the relationship be-tween both rating favorability (e.g., high or low) and self-otherrating discrepancy and accuracy and negative reactions. The hy-pothesized relationships are described in detail below.

Favorability of Others' Ratings

A number of studies of performance appraisal ratings havesuggested that more positive evaluations are seen as more accurateand are better accepted than negative ratings (Halperin, Snyder,Shenkel, & Houston, 1976; Stone & Stone, 1985). Positive ratingsalso have been related to satisfaction with the appraisal or theappraisal process (Dipboye & de Pontbriand, 1981; Ilgen & Ham-stra, 1972; Morran & Stockton, 1980; Podsakoff & Farh, 1989).Because performance appraisal ratings often provide the basis forevaluative outcomes such as raises or promotions, it is not sur-prising that more positive ratings are better received. Whether thisfinding holds for ratings that have no ties to evaluative outcomesis uncertain. In one study with a small sample of managers,Facteau, Facteau, Schoel, Russell, and Poteet (1998) found thatfeedback from subordinates and peers was related to acceptance;that is, more favorable ratings were seen by managers as moreaccurate.

Self-enhancement theory (Schrauger, 1975) suggests that indi-viduals will react more positively to higher ratings regardless oftheir purpose. We proposed that higher ratings from multisourceanonymous, developmental feedback from others will be per-ceived. We proposed that higher ratings will be perceived as moreaccurate and will be related to more positive reactions than lowerratings. We expected lower ratings to be perceived as negativefeedback and related to more negative reactions. Kernis and John-son (1990) found that failure feedback resulted in more unpleasantemotions such as shame, anger, and overall unpleasantness. Meyer(1980) found that employees who were given below-average meritratings became alienated and demoralized. Taylor, Fisher, andIlgen (1984) warned that negative feedback may be perceived asthreatening and may result in defiant opposition. Research has notexamined whether these findings hold for negative feedback that isdesigned for developmental purposes (e.g., 360° feedback).

The research on reactions to feedback, limited as it is, suggeststhat reactions to negative versus positive feedback may includemore affect intensity and arousal (see Russell, 1980). For example,low ratings may evoke emotional arousal such as anger or defiantopposition. It is less likely that high ratings will evoke emotionalarousal such as joy or glee, but these ratings may result in positivereactions such as pleased or inspired. Thus, we proposed distincthypotheses for low and high rating favorability and reactions tofeedback. We hypothesized the following:

Hypothesis 1: Higher ratings will be perceived as more accurate.

Hypothesis 2: Higher ratings will be related to positive reactions.

Hypothesis 3: Lower ratings will be related to negative reactions.

Self-Other Rating Discrepancy

Rating favorability may have less of an influence on managers'attitudes toward feedback than the discrepancies between theirexpected ratings and the actual ratings received from varioussources. Control theory (Carver & Scheier, 1981) suggests thatself-other discrepancies will motivate the managers to reduce anydiscrepancies by changing their behaviors to bring them in linewith others' expectations. However, behavioral change is only onestrategy for discrepancy reduction. Another approach under con-trol theory is for the managers to seek feedback from the environ-ment that would disconfirm the discrepancy. Waldman, Atwater,Clements, and Atwater (1996) found support for this discrepancyreduction tactic in a study in which those with lower ratings thanexpected were more likely to seek additional feedback. Cognitivedissonance theory (Festinger, 1954) suggests that another way toreduce the discrepancy between self-ratings and others' ratings isto discount the feedback from others as inaccurate. The extent towhich feedback ratings deviate from one's expectations has beenshown to be related to perceived accuracy of the feedback and torecipient reactions. Bernstein and Lecomte (1979) found that feed-back that was more negative than expected was valued less thanfeedback that was similar to or more positive than what wasexpected. Meyer (1980) found that self-other rating discrepancieswere associated with defensiveness on the part of the feedbackrecipient.

Feedback that confirms the recipient's self-view is preferred andis seen as more informative (Pearce & Porter, 1986; Swann &Read, 1981). Pedler and Boydell (1980) found that feedback thatwas too disconfirming of one's self-view resulted in nondevelop-mental outcomes (i.e., rationalization). The use of 360° feedbackusually includes self-ratings and averaged ratings from others(London & Smither, 1995). Consequently, self-other discrepan-cies are a salient component of the feedback. Overraters, that is,individuals who receive more ratings from others that are lowerthan expected, are predicted to discount the feedback or to perceiveit as less accurate than those who receive fewer ratings from othersthat are lower than expected.' We anticipated that self-other ratingdiscrepancy will be related to perceived accuracy as follows:

Hypothesis 4: Overraters (those who have more ratings from othersthat are lower than expected) will rate feedback as less accurate thanindividuals with fewer overratings.

The saliency of self-feedback versus others' feedback is likelyto focus recipients' attention on themselves rather than on tasks(DeNisi & Kluger, 2000). When attention is focused on the self,affective reactions to the feedback are often strong (e.g., despair oranger). Because self-ratings are often inflated relative to ratingsprovided by others (Mabe & West, 1982), many feedback recipi-ents receive feedback that is more negative than their self-views(overraters). Feedback discrepancy (e.g., over- vs. underrating) isexpected to relate to reactions to feedback. The more feedback thatindicates an individual overrated himself or herself (e.g., lowerratings from others than expected), the more the individual will

1 We refer to overraters and overrating. These terms are comparable towhat others have referred to as overestimators and overestimation (Atwateret al., 1995; Fleenor, McCauley, & Brutus, 1996).

Page 4: 360° Feedback: Accuracy, Reactions, and …eportfolio.lib.ksu.edu.tw/~T093000259/repository/fetch...360 Feedback: Accuracy, Reactions, and Perceptions of Usefulness Joan F. Brett

360° FEEDBACK 933

report negative reactions to the feedback. The more feedback thatindicates an individual underrated himself or herself (e.g., higherratings from others than expected), the more the individual willreport positive reactions to the feedback. We expected that self-other rating discrepancy will be related to feedback reactions asfollows:

Hypothesis 5A: Underratings of self will be related to positivereactions.

Hypothesis SB: Overratings of self will be related to negativereactions.

Accuracy and Reactions

Several studies have indicated that individuals assess the accu-racy of feedback they receive and that perceived accuracy relatesto their reactions to that feedback. O'Reilly and Anderson (1980)reported that employees' perceptions of the accuracy of theirperformance evaluation moderated the relationship between thefeedback received and their satisfaction with the appraisal. In asimilar manner, Cedarbloom and Lounsbury (1980) found that theextent to which peers believed peer evaluations were valid wasdirectly related to their satisfaction with the peer evaluation sys-tem. Perceptions of accuracy may also relate to recipients' affec-tive reactions to feedback (e.g., anger or discouragement), not justtheir satisfaction with the appraisal. We believe that feedback thatis perceived as inaccurate will be associated with negative reac-tions, whereas feedback that is perceived as accurate will berelated to positive reactions. We proposed the following:

Hypothesis 6A: Perceptions of accuracy will be positively related topositive reactions.

Hypothesis 6B: Perceptions of accuracy will be negatively related tonegative reactions.

Reactions and Feedback Usefulness

It seems unlikely that individuals who have negative reactions tofeedback also will report that the feedback is useful. Bernstein andLecomte (1979) found that subjects who received feedback thatwas more negative than expected valued the feedback less than didthose for whom feedback was more positive than expected. Be-cause we expected negative feedback (e.g., low ratings and over-ratings) to contribute to negative reactions, we expected, in turn,that negative reactions will be related to perceptions that feedbackis not useful. It is possible that individuals who react negativelycould, at the same time, see that the feedback provides informationthey did not previously have. According to London (1994), thisinformation should be seen as useful or valuable; however, cog-nitive dissonance theory suggests that feedback that makes oneangry will be seen as having little value. We proposed the follow-ing hypotheses:

Hypothesis 7A: Positive reactions will be positively related to percep-tions of usefulness.

Hypothesis 7B: Negative reactions will be negatively related to per-ceptions of usefulness.

Accuracy and Feedback Usefulness

Individuals who believe feedback from others is inaccurate willlikely discount the value of that feedback. Regarding formal ap-

praisals, Ilgen, Fisher, and Taylor (1979) proposed four stages inunderstanding how feedback results in behavior change. First, thefeedback is perceived. Second, the feedback is accepted as accu-rate. Following acceptance as accurate, the individual expresses adesire to respond or, in our terms, finds the feedback useful forfuture development. The last stage is the actual behavior change.Consistent with these stages, we proposed that feedback that isperceived as an accurate portrayal of the recipient's behavior willbe seen as more useful for development. We anticipated thatindividuals who find the feedback more useful will express moti-vation to change and focus on development in a counseling sessionwith a facilitator. Individuals who find the feedback to be lessuseful are expected to be defensive in the session with the feed-back facilitator. We hypothesized the following:

Hypothesis 8: Perceptions of accuracy will be positively related toperceptions of feedback usefulness.

Hypothesis 9A: Participants' perceptions of usefulness will bepositively related to facilitators' ratings of participants asdevelopment-focused.

Hypothesis 9B: Participants' perceptions of usefulness will be nega-tively related to facilitators' ratings of participants as defensive.

Goal Orientation as a Moderator

Dispositional goal orientation may shed light on why someindividuals view negative feedback as a developmental opportu-nity whereas others discount negative feedback. Learning andperformance goal orientations are associated with two differentinterpretive frameworks. First, learning and performance goal-oriented individuals hold different implicit theories about the con-trollability of personal attributes, such as intellectual ability(Dweck, 1986). Individuals with a performance goal orientationtend to hold an entity theory about their ability, which is viewed asa fixed, uncontrollable personal attribute. Individuals with a learn-ing goal orientation tend to hold an incremental theory about theirability. For them, ability is viewed as a malleable attribute that canbe developed. Second, goal orientation influences how individualsview effort expenditures (Ames, 1992). With a learning goalorientation, effort is believed to be a strategy for developing theability needed for future mastery. With a performance goal orien-tation, ability is perceived as a fixed attribute, and effort expendi-ture indicates that ability is low. That is, only those who areincapable should need to exert extra effort to master a task.

Each goal orientation is associated with different response pat-terns to failure and feedback (Dweck & Leggett, 1988; Elliot &Dweck, 1988). Individuals with a high learning goal orientationpursue an adaptive response pattern to failure in that they persist,escalate effort, and engage in solution-oriented strategies. With aperformance goal orientation, individuals pursue a maladaptiveresponse pattern in that they withdraw from the task and makenegative ability attributions. Such a response pattern is predictablebecause failure calls their ability into question.

A performance goal orientation is defined as the desire both togain favorable judgments and to avoid unfavorable judgmentsabout one's ability (Heyman & Dweck, 1992). Recent research hasconceptualized goal orientation as a three-factor construct: learn-ing, performance-prove, and performance-avoid (VandeWalle,1997). Separating the performance goal orientation into prove and

Page 5: 360° Feedback: Accuracy, Reactions, and …eportfolio.lib.ksu.edu.tw/~T093000259/repository/fetch...360 Feedback: Accuracy, Reactions, and Perceptions of Usefulness Joan F. Brett

934 BRETT AND ATWATER

avoid dimensions captures the distinction between the positiveself-evaluation characteristic of performance-prove and the nega-tive self-evaluation characteristic of performance-avoid.

The different response patterns associated with particular goalorientations moderate individuals' reactions to feedback (Fair etal, 1993). According to Dweck (1986), individuals with a learningorientation view negative feedback as useful information on howto develop mastery, whereas those with a performance orientationview negative feedback as an evaluation of one's competence, tobe proven if positive or avoided if negative. Research suggests thatindividuals with a performance-prove goal orientation may try todemonstrate their ability by attempting to look better than othersdo. Brett and VandeWalle (1999) found that goal orientation wasrelated to the goals that individuals set for a training program.Individuals with a performance-prove orientation were morelikely to set goals for performing better than others, whereasindividuals with a performance-avoid orientation were concernedwith avoiding a negative evaluation during the training. Fair et al.proposed that individuals with a performance goal orientation arelikely to assess their performance by using similar others as areference point and may experience negative reactions when thefeedback does not confirm their competence. Button et al. (1996)suggested that a performance goal orientation might be associatedwith defensive behavior if engaging in a task would risk demon-stration of a low skill level and possible negative evaluation fromothers.

We proposed that a performance-prove goal orientation mod-erates an individual's perceptions of and reactions to feedback.The use of 360° feedback provides an assessment of an individualas others see him or her. In addition, the feedback process providesnormative data for comparison purposes, thereby focusing therecipient's attention on how well he or she did relative to others.Thus, the 360° feedback process is ripe for those with a highperformance-prove goal orientation to react to feedback differ-ently than those with a low performance-prove goal orientation.We hypothesized the following:

Hypothesis 10A: Goal orientation moderates the relationship of rat-ings and perceived accuracy. (Individuals with a higher prove orien-tation will perceive lower ratings as less accurate than individualswith a lower prove orientation.)

Hypothesis WB: Goal orientation moderates the relationship of rat-ings and negative reactions. (Individuals with a higher prove orienta-tion will have more negative reactions to lower ratings than individ-uals with a lower prove orientation.)

Hypothesis IOC: Goal orientation moderates the relationship betweenself-other discrepancy and perceptions of accuracy. (Overraters witha higher prove orientation will perceive feedback as less accurate thanthose with a lower prove orientation.)

Hypothesis 10D: Goal orientation moderates the relationship of self-other discrepant feedback and negative reactions. (Those with a higherprove orientation will report more negative reactions to overratingthan those with a lower prove orientation.)

Rating Source

Managers have different relationships with bosses, peers, anddirect reports; thus, managers may value feedback from certainsources or may expect higher or lower ratings, depending on thesource. Ratings from bosses may be perceived as more important

because superiors control resources and have legitimate powerover recipients. Greller and Herald (1975) found that superiorswere considered to be a more important source of feedback whencompared with coworkers.

Ratings from peers may be perceived as less accurate or valu-able because managers have less direct contact with peers thanwith superiors or direct reports. However, some managers mayexpect higher ratings from peers than from other sources, becausethey are more likely to have informal, nonwork relationships withtheir peers than with bosses or direct reports.

One could also argue that direct reports' ratings may be per-ceived as most relevant to a manager because 360° feedback oftenincludes leadership or management behaviors that are the mostapplicable to that group. In addition, some managers may expectlower ratings from some subordinates, given the various perfor-mance issues they face with their subordinates. Although we didnot hypothesize any specific effects for rating source, we didseparately explore relationships for Hypotheses 1 through 6 byrating source. Support for the preceding hypotheses would extendthe research findings on performance appraisal to 360° develop-mental feedback and suggest that some of the assumptions about360° feedback may be called into question.

Method

Participants and Procedure

The sample consisted of 125 students in a master's of business admin-istration (MBA) program. Participants' average work experience was 5.28years (SD = 2.70 years), and their average age was 27.87 years (SD = 3.06years). Women made up 29% of the sample. Entering MBA students whohad 2 or more years of managerial experience were invited to participate ina 360° feedback process offered by a nonacademic business leadershipcenter at a southwestern university. While still employed full-time, partic-ipants asked their bosses, peers, and direct reports to complete a question-naire that assessed their leadership behaviors. Participants were required tohave a minimum of six total raters. The multirater feedback instrumentused in this study was a customized version of the management skills"Profilor" (Hezlett, Ronnkvist, Holt, & Hazucha, 1997) developed byPersonnel Decision International (PDI). Respondents sent the completedquestionnaires directly to PDI for scoring.

Six to eight weeks into the semester, participants attended a small groupsession conducted by a PDI facilitator. Participants received a feedbackreport that included (a) a summary of self-ratings versus boss ratings on theimportance of 20 skills; (b) self-, boss, direct report, and peer ratings; (c)a graph indicating self-, boss, average direct report, and average peerratings set against a range of normative ratings on the 20 skills; and (d)rankings of the highest and lowest ratings on items across all three ratingsources. The facilitator did not review participants' feedback prior to thegroup session. The content of the group session did not differ on the basisof the participants' feedback, and the facilitator covered the same materialin all of the sessions. The facilitator explained how to interpret theinformation included and how to understand the discrepancies betweenself-ratings and others' ratings. The facilitator emphasized the importanceof feedback and encouraged participants to schedule a one-on-one meetingwith a facilitator to discuss how to use the feedback for developmentalpurposes. At the end of the session, participants completed a questionnairethat assessed their attitudes and reactions to the feedback.

Two to four weeks later, participants attended an hour-long individualmeeting with the facilitator to discuss developmental plans. At the end ofthat session, participants completed a second questionnaire on their reac-tions to the feedback process. The PDI facilitator, who was naive to the

Page 6: 360° Feedback: Accuracy, Reactions, and …eportfolio.lib.ksu.edu.tw/~T093000259/repository/fetch...360 Feedback: Accuracy, Reactions, and Perceptions of Usefulness Joan F. Brett

360° FEEDBACK 935

hypotheses, rated the participants on their receptivity to development. Thevoluntary developmental session was attended by 103 of the 125 partici-pants. There were no significant differences on the study variables (e.g.,accuracy, reactions) on the basis of attendance at the developmentalsessions.

Measures

The measures were collected from multiple sources (three sources ofwork colleagues, facilitator, and self-ratings) across four time periods.Table 1 depicts the time line and data sources. The "Profilor" assessesleadership and managerial behavior on 20 skill dimensions and 6 generalperformance dimensions. Each of the 20 skill dimensions is based on 3 to 6items, resulting in a total of 87 items. For example, the dimensions includelistens to others (3 items), manages disagreements (4 items), and fostersteamwork (4 items). Self, boss, peers, and direct reports rated the extent towhich the participant performed each behavior on a scale ranging from 1(not at all) to 5 (to a great extent). The internal consistency reliability forthe 26 dimensions ranged from .68 to .88 for self-ratings, from .75 to .94for boss ratings, from .85 to .97 for direct report ratings, and from .82 to96 for peer ratings.

Ratings. For the purposes of analysis and because the dimensions arecorrelated, an overall rating for each source (self, boss, direct reports, andpeers) was calculated on the basis of the average ratings for the 26dimensions. The direct report and peer ratings were based on informationfrom multiple raters. An average score across direct reports (and peers) wasobtained for each of the 26 dimensions. PDI collected feedback from themultiple sources and conducted the analysis. Reliability and intraclasscorrelations were based on a PDI "Profilor" database using a randomsample of 1,400 managers with three or more peers and three or more directreports (Hezlett et al., 1997). Interrater reliability was computed usinganalysis of variance (Shrout & Fleiss, 1979). Intraclass correlations rangedfrom .47 to .60 for peers and from .48 to .61 for direct reports, indicatingacceptable interrater reliability.

Accuracy. Participants responded to the following: "Some participantsinitially question the accuracy of the feedback as either too positive or toonegative. Please use the following scale to rate your perceptions of feed-back accuracy." Participants provided separate ratings of the accuracy ofboss, peer, and direct report ratings (accuracy boss, accuracy peer, andaccuracy direct report) using a 7-point scale ranging from 1 (very inaccu-rate) to 7 (very accurate).

Reactions. Participants rated 13 reactions indicating "the extent towhich you feel this way now" by using a 5-point scale ranging from 0 (notat all) to 4 (very much). These items were based on the expected reactionsto feedback from the goal orientation literature and on input from the PDIfacilitator, who routinely conducted 360° feedback sessions and was fa-miliar with the range of reactions experienced by feedback recipients. ThePositive Reaction scale included 7 items (inspired, encouraged, informed,aware, pleased, motivated, and enlightened) with a Cronbach's alphareliability of .81. The Negative Reaction scale included 6 items (angry,judged, confused, examined, criticized, and discouraged) and had a Cron-

bach's alpha of .76. The Positive and Negative Reaction scales were nothighly correlated (r = .02). These measures were the last items on thequestionnaire to minimize any influence on the ratings of accuracy orusefulness.

Usefulness. Participants twice rated their perceptions of the usefulnessof the feedback report: at the group session when they received theirfeedback report and after the one-to-one developmental session with thefacilitator. This scale included 3 items (e.g., "This report was useful to me"and "This report is valuable for helping me diagnose my managementabilities"). The 3 items had a Cronbach's alpha reliability of .74 at Time 1and .67 at Time 2.

Facilitator's ratings. After the developmental session, the PDI facili-tator rated each participant on 12 items using a 7-point Likert scale rangingfrom 1 (strongly disagree) to 7 (strongly agree). The Development Fo-cused scale was based on 7 items (e.g., "This person was open to feedback"and "This person was concerned with understanding what the feedbackmeant for their development"). The Defensiveness scale consisted of 5items (e.g., "This person blamed his/her scores on other people or situa-tions" and "This person came across as defensive"). The two scales hadCronbach's alphas of .95 and .87 for Development Focused and Defen-siveness, respectively.

Goal orientation. A 13-item measure assessed goal orientation(VandeWalle, 1997). The instrument has three subscales: (a) 5 itemsmeasure a learning goal orientation, (b) 4 items measure a performance-prove orientation, and (c) 4 items measure the avoid dimension of aperformance goal orientation. A 7-point Likert-type response scale rangingfrom 1 (strongly disagree) to 7 (strongly agree) was used. This measurewas completed in an academic course during the first week of classes.

Results

Means, standard deviations, and Intel-correlations of the studyvariables are presented in Table 2. Reliability estimates (Cron-bach's alphas) are located along the diagonal in Table 2.

Rating Favombility, Accuracy, and Reactions to theRatings

Hypothesis 1 asserted that the higher the ratings a participantreceived, the more accurate he or she would rate the feedback. Thezero-order correlations shown in Table 2 indicate that this hypoth-esis was supported for boss (r = .31, p < .001) and direct report(r = .57, p < .001) ratings but was not significant for peer ratings.

Hypothesis 2 proposed that higher ratings from boss, peers, anddirect reports would be positively related to positive reactions.This hypothesis was not supported for any rating source. However,in support of Hypothesis 3, lower boss and peer ratings weresignificantly related to negative reactions (r = —.38, p < .001, andr = —.25, p < .01, respectively; see Table 2).

Table 1Time Line, Data Sources, and Measures

Variable Time 1 Time 2 Time 3 Time 4

Time period Employment prior to Week 1 of semester Weeks 6-8 (Feedback returned) Weeks 9-12 (One-on-one with facilitator)MBA program

Measure Self, boss, peer, and Goal orientation Perceptions of accuracy, Time 1 usefulness, Time 2 usefulness and facilitator's ratingsdirect report ratings questionnaire and positive and negative reactions of participant's development and

defensiveness focus

Note. MBA = master's of business administration.

Page 7: 360° Feedback: Accuracy, Reactions, and …eportfolio.lib.ksu.edu.tw/~T093000259/repository/fetch...360 Feedback: Accuracy, Reactions, and Perceptions of Usefulness Joan F. Brett

936 BRETT AND ATWATER

Table 2Means, Standard Deviations, Coefficient Alphas, and Correlations Among Study Variables

Variable

1. Learn"2. Prove"3. Avoid'4. Self-rating5. Boss rating6. Peer rating7. Direct report rating8. Over boss9. Over peer

10. Over direct report11. Under boss12. Under peer13. Under direct report14. Accuracy boss15. Accuracy peer16. Accuracy direct report17. Positive reaction18. Negative reaction19. Useful 120. Useful 221. Develop"22. Defend"MSDN

1

(.69)-.03-.32***

.05**-.04

.02-.22*

.15*

.08

.09-.11-.24***-.22**-.01

.04-.03

.06-.02-.06

.09

.08

.056.220.59

124

2

(.81).26**.06

-.11-.18*-.23*

.15*

.12

.13-.13-.14-.12

.00

.12-.10-.21**-.00

.04-.15-.15

.17*5.041.07

124

3

(.84)-.07-.06-.07

.10-.01-.06-.05-.05-.04

.04-.12-.03

.05-.07

.15

.04-.15-.12-.103.301.14

124

4

(.75).1324**.04.50***.55***.40***

-.46***_ 57***-.34***

.13

.11

.06

.03

.05-.05

.06-.03

.063.870.35

122

5

(.83).20**.19

_7]***-.07-.05

.66***

.08

.09

.31***-.12

.00

.06-.38***

.20**

.13

.01

.023.970.45

117

6

(.88).48***.03

-.50***-.03

.01

.51***

.04

.07-.03

.11

.09-.25**

.08

.04-.07-.034.000.34

117

7

(.90)-.16-.31**_70***

.04

.33***

.57***-.17-.15

.57***

.16

.03

.14

.15-.02-.044.050.46

68

8

.45***

.37***— 54***-.28***-.22**-.16*

.13

.02-.04

.26***-.13-.00-.03

.085.366.67

124

9

.34***-.25**-47***-.22*

.03

.05-.05-.06

.26***-.08

.11-.01

.053.534.99

124

Note. Coefficient alphas appear in parentheses along the diagonal. Over = overrating; Under = underrating; Useful 1 = usefulness at Time 3; Useful2 = usefulness at Time 4.a Learn, prove, and avoid were the three goal orientations. b Develop and defend refer to the facilitator's ratings of the feedback recipient's developmentfocus and defensiveness, respectively.tp< .10 . *p<.05. **p<.0\. ***/><.001.

Hypotheses 4 and 5 examined the relationships of self-otherdiscrepancies and participants' attitudes and reactions. Separatehierarchical regression analyses were conducted for the differentdependent variables and for each of the three rating sources.Predictor variables were entered in the following order: (a) others'ratings, (b) self-ratings, and (c) others' ratings by self-ratingsinteraction. A significant interaction indicated that the self-otherrating discrepancy was relevant (Atwater, Ostroff, Yammarino, &Fleenor, 1998). Hypothesis 4 predicted that self-other discrepan-cies would be negatively related to perceptions of accuracy. Table3 indicates that the regression model explained a significant por-tion of the variance in perceptions of accuracy for direct reports(R2 = .40, p < .001) and for bosses (R2 = .11, p < .05). Theself-other interaction added significantly to the prediction of ac-curacy only for direct reports (A./?2 = .07, p < .05).

To determine whether an over- or underrating discrepancyaccounted for these results, over- and underrating variableswere computed. First, a discrepancy score was created bysubtracting others' ratings from self-ratings for each of the 26dimensions. An overrating variable was computed by summingthe number of discrepancy scores that were positive (self-ratingwas higher than others' rating) across the 26 dimensions. Theunderrating variable was computed by summing the number ofnegative discrepancy scores (self-ratings were lower than oth-ers' ratings) for the 26 dimensions. For example, if a managerrated himself or herself higher than did his or her boss on 13dimensions, the boss overrating score was 13, or if a manager

rated himself or herself lower than peers rated him or her on 19dimensions, the peer underrating score was 19. This measurewas used rather than the more traditional categories of under-and overraters (see Fleenor et al., 1996) for two reasons. First,we believed that given that the recipient saw a graph of his orher self-ratings and others' ratings on each of the 26 dimen-sions, the number of dimensions on which he or she overratedwas likely more salient than the degree of discrepancy on thedimensions. Second, our sample size was too small to createmeaningful cell sizes if we broke the sample into four or sixcategory groupings. As such, six variables were created (overboss, over peer, over direct report, under boss, under peer, andunder direct report). In Table 2, the significant zero-ordercorrelations indicate that when participants rated themselveshigher than their boss and direct reports, they perceived theirboss (r = -.16, p < .05) and direct report (r = -.35,;? < .001)ratings as less accurate. Accuracy and self-peer rating discrep-ancies were not related.

We tested Hypotheses 5A and 5B with multiple regressionanalyses. Specifically, others' ratings, self-ratings, and the inter-action between self-ratings and others' ratings were entered topredict reactions. Hypothesis 5A predicted that underrating wouldbe related to positive reactions to feedback. Table 3 indicates thatself-other discrepancies were not significantly related to positivereactions (i.e., the interactions were not significant). Neither self-ratings, others' ratings, nor the interaction term predicted positivereactions. However, the underrating score was negatively related

Page 8: 360° Feedback: Accuracy, Reactions, and …eportfolio.lib.ksu.edu.tw/~T093000259/repository/fetch...360 Feedback: Accuracy, Reactions, and Perceptions of Usefulness Joan F. Brett

360° FEEDBACK 937

10 11 12 13 14 15 16 17 18 19 20 21 22

-.13 —-.27** .43**-.14 .32***.19* .13.09 -.09

-.35*** -.02.02 .05.04 -.26**.01

-.04-.08.132.385.2268

• 13t.13t.03.096.666.50

124

.40***-.06-.11-.01.01

-.17*.08.01

-.06-.076.306.24

124

-.03-.10.27**.11.01.12.05

-.02-.044.516.6468

.13

.12

.09-.28***.22**.25**.06

-.055.151.30

115

.14

.15*-.08-.02.05.09

-.025.480.81

116

.23**-.03.26**.31**.12.02

5.211.1586

(.81)-.02.54***.38***.33***

-.122.590.63

120

(.76)-.16*-.05-.02-.021.000.67

116

(.74).58***.39***

-.1515.790.75

120

(.67).36***

-.155.990.67

103

(.95)-.525.191.12

103

(.87)2.981.09

103

to negative reactions for boss (r = —.26, p < .01) and peers (r =—.17, p < .05; see Table 2). Individuals who had more underrat-ings did not report more positive reactions to their feedback;however, those who had fewer underratings reported lower nega-tive reactions.

Hypothesis 5B proposed that the more recipients overratedthemselves, the more negative their reactions to feedback would

be. Table 3 shows that this hypothesis was supported for peers andbosses. Specifically, the self-ratings by peer ratings interactionsignificantly predicted negative reactions after self-ratings andpeer ratings were entered into the equation (A/?2 = .04, p < .05).In a similar manner, the self-ratings by boss ratings interaction wasmarginally significant (A/?2 = .03, p < .10). The correlationsbetween overrating and negative reactions (i.e., over boss, over

Table 3Results of the Regression Analysis With Self-Other Ratings

Boss ratings

Variable

Perceived accuracyStep 1: Other ratingStep 2: Self-ratingStep 3: Other X Self

Positive reactionsStep 1 : Other ratingStep 2: Self-ratingStep 3: Other X Self

Negative reactionsStep 1: Other ratingStep 2: Self-ratingStep 3: Other X Self

P

0.70*0.40

-0.51

-1.02-0.80

1.40

-2.54*-l.Slt

2.78f

AR2

.01

.00

.00

.01

.01

.03t

R2

.10**

.11*

.11*

.00

.00

.01

14***.15***.18***

Peer ratings

ft

-0.34-0.19

0.45

-0.27-0.38

0.58

1.572.11*

-2.92*

AR2

.01

.00

.00

.00

.01

.04*

R2

.00

.01

.01

.01

.01

.01

.06*

.08*

.11**

Direct

ft

-2.53-2.40*

3.90*

-2.79t-2.32t

3.70*

0.370.35

-0.42

report ratings

AR2

.01

.07* ,

.00

.06*

.01

.00

R2

.33***

.34***

.40***

.02

.03

.09

.00

.01

.01

Note. Betas are for when all variables were in the equation.tp<.10. *p<.05. **/?<.01. ***/>< .001.

Page 9: 360° Feedback: Accuracy, Reactions, and …eportfolio.lib.ksu.edu.tw/~T093000259/repository/fetch...360 Feedback: Accuracy, Reactions, and Perceptions of Usefulness Joan F. Brett

938 BRETT AND ATWATER

peer, and over direct report with negative reactions) provided inTable 2 showed that self-overratings for boss and peers wererelated to negative reactions (r = .26, p < .001, and r = .26, p <.001, respectively). This was not the case for direct reports. Thus,Hypothesis 5B was supported for two sources.

Accuracy, Reactions, Usefulness, and Facilitator'sRatings

Hypotheses 6 through 10 were tested with zero-order correla-tions (see Table 2). Hypothesis 6A predicted that perceptions ofaccuracy would be positively related to positive reactions. Percep-tions of accuracy were positively related to positive reactions forpeers (r = .15, p < .05) and direct reports (r = .23, p < .01).Hypothesis 6B proposed that accuracy would be negatively relatedto negative reactions. Individuals who believed that feedback wasinaccurate would have more negative reactions. Perceived accu-racy of the boss ratings was negatively related to negative reac-tions (r = -.28, p < .05); that is, ratings that were perceived asless accurate were accompanied by negative reactions. Perceivedaccuracy of peer and direct report ratings was not related tonegative reactions. Partial support was found for Hypotheses 6Aand 6B.

Hypothesis 7A predicted that positive reactions would be relatedto perceptions of usefulness of the feedback. Positive reactionswere positively related to perceptions of usefulness (r = .54, p <.001). Hypothesis 7B predicted that negative reactions would benegatively related to perceptions of usefulness. This hypothesisalso was supported (r = —.I6,p< .05). Therefore, when indi-viduals react more positively, they also believe the feedback ismore useful; when they react more negatively, they see the feed-back as less useful.

Hypothesis 8 asserted that perceptions of accuracy would bepositively related to perceptions of usefulness. As Table 2 indi-cates, the more participants perceived their boss and direct reportratings as accurate, the higher their perceptions of feedback use-fulness (r = .22, p < .01, for boss and r = .26, p < .01, for directreport accuracy). Hypothesis 8 was supported.

Hypothesis 9A proposed that participants' perceptions of theusefulness of the feedback would be positively related to thefacilitator's ratings of the participants as development-focused. Insupport of this hypothesis, participants' ratings of feedback use-fulness were significantly related to the facilitator's ratings of theparticipants as development-focused (r — .39, p < .001). Hypoth-esis 9B was marginally supported in that participants' ratings ofusefulness were marginally negatively correlated with the facili-tator's ratings of defensiveness (r — —.I5,p< .10).

Goal Orientation as a Moderator

Hypotheses 10A through 10D proposed that goal orientationwould serve as a moderator variable. These hypotheses were testedwith separate hierarchical regressions in which goal orientationwas entered after others' ratings and self-ratings were entered topredict each dependent variable. Hypotheses 10A and 10B pro-posed that individuals with a higher performance-prove orienta-tion would perceive lower ratings as less accurate and would havemore negative reactions to lower ratings than those with a lowerperformance-prove orientation. Hypotheses IOC and 10D asserted

that a performance-prove orientation would moderate the relation-ship between self-other rating discrepancies and perceptions ofaccuracy and feedback reactions. Table 4 shows that goal orien-tation did not moderate these relationships.

Discussion

The results of this study, summarized in Table 5, provideevidence that some of the assumptions made by those using 360°feedback should be questioned. Specifically, negative feedback(i.e., ratings that were low or that were lower than expected) wasnot seen as accurate or useful, and it did not result in enlightenmentor awareness but rather in negative reactions such as anger anddiscouragement. In addition, high ratings were not related topositive reactions but merely to the absence of negative reactions.These findings suggest that people who may need feedback themost because they are not performing well or have an inflated viewof their effectiveness (overraters) are least receptive to it and findit least useful. This study extends the research on the detrimentaleffects of negative performance appraisal feedback to 360° devel-opmental feedback.

One striking feature of our findings is that, in contrast to theperformance appraisal feedback studied by Kluger and DeNisi(1996), the feedback in this study was clearly provided as devel-opmental. Participants in this study were no longer working withthe individuals who provided their feedback. Work colleaguesprovided the ratings while the participants were still employed, andparticipants received the feedback 2 to 4 months later after theyhad quit work and begun a full-time MBA program. This 360°process was designed to provide participants with developmentalfeedback to personalize their 2-year program of study. However,rather than perceiving discrepant feedback as an opportunity forawareness and enlightenment (positive reactions), participants re-ported negative reactions to it. One wonders how managers whoreceive negative or discrepant feedback and must continue work-ing with these raters would alter their attitudes and reactionstoward them. A recent study by Atwater, Waldman, Atwater, andCartier (2000) revealed that individuals who received low ratingsfrom their subordinates reduced their level of loyalty and commit-ment to their subordinates after receiving feedback. A secondstudy by Atwater, Brett, Waldman, and Yammarino (2000) sug-gested that supervisors' ratings of leader-member exchange andliking toward their subordinates decreased following negativefeedback from those subordinates. These results, coupled with ourfindings regarding negative reactions, suggest that more research isneeded to identify potential negative reactions and to take steps toprevent or manage them as part of the 360° feedback process.

Goal orientation did not moderate the relationships betweenratings and accuracy and reactions, as we expected. Given that anindividual with a performance-prove orientation views negativefeedback as a statement about his or her ability, we expected thatthese individuals would be more likely to distort the accuracy ofthe ratings and report negative reactions. Participants appeared toreport negative reactions to lower ratings regardless of goal ori-entation. The findings suggest that the 360° process may create an"evaluative" context in which a performance-prove goal orienta-tion has been induced by the situation. Although typically designedfor developmental purposes, 360° feedback programs may rein-force the salience of evaluative and comparative information. In

Page 10: 360° Feedback: Accuracy, Reactions, and …eportfolio.lib.ksu.edu.tw/~T093000259/repository/fetch...360 Feedback: Accuracy, Reactions, and Perceptions of Usefulness Joan F. Brett

360° FEEDBACK 939

Table 4Results of the Regression Analysis With Goal Orientation

Boss ratings

Variable

PerceivedStep 1:Step 2:Step 3:Step 4:Step 5:Step 5:

PerceivedStep 1:Step 2:Step 3:

accuracyOther ratingSelf-ratingOther X SelfProveProve X OtherProve X SelfaccuracyOther ratingProveProve X Other

ft

0.240.14

-0.47-1.33

0.840.64

-0.12-0.76

0.87

A«2

.01

.00

.00

.01

.00

.00

.01

R2

.10***

.11**

.11**

.11*

.12*

.12*

.10***

.10**

.11**

Peer ratings

ft

-0.360.320.320.900.36

-1.26

-0.09-0.06

0.18

AR2

.01

.00

.01

.00

.01

.01

.00

R2

.00

.01

.01

.02

.02

.03

.00

.01

.01

Direct

ft

-2.68-2.78

4.07*-0.70

0.150.78

-0.01-1.04

1.17

report ratings

AS2

.01

.06*

.02

.00

.00

.01

.02

R2

.34***

.34***

.40***

.42***

.42***

.42***

.34***

.35***

.36***Positive reactions

Step 1:Step 2:Step 3:Step 4:Step 5:Step 5:

Other ratingSelf-ratingOther X SelfProveProve X OtherProve X Self

-1.070.661.600.59

-0.26-0.61

.00

.01

.04*

.00

.00

.00

.00

.01

.05

.05

.05

-0.68-0.40

0.85-0.35

0.49-0.38

.00

.00

.05*

.00

.01

.01

.01

.06

.06

-3.22-1.91

4.43*2.11

-0.33-2.08

.00

.08*

.00

.02

.02

.03

.10

.10

.13Positive reactions

Step 1:Step 2:Step 3:

NegativeStep 1:Step 2:Step 3:Step 4:Step 5:Step 5:

NegativeStep 1:Step 2:Step 3:

Other ratingProveProve X Other

reactionsOther ratingSelf-ratingOther X SelfProveProve X OtherProve X Self

reactionsOther ratingProveProve X Other

0.140.01

-0.22

-2.06-1.39

2.841.27

-1.07-0.38

0.120.89

-1.01

.04*

.00

.00

.03tt

.00

.02

.00

.00

.02

.00

.04

.04

.15***

.15***

.18***

.18***

.20**

.20**

.15**

.15**

.16**

-0.16-0.74

0.54

2.00*1.83*

-2.60*1.50

-1.650.16

0.441.78

-1.80

.04*

.00

.01

.04*

.00

.02

.00

.00

.03

.01

.05tt

.05

.07**

.08*

.11**

.11*

.13*

.13*

.07**

.07*

.09*

-0.39-1.04

1.00

0.771.73

-1.77-0.80

1.28-1.36

-0.26-0.57

0.60

.00

.01

.00

.00

.00

.01

.01

.00

.00

.02

.03

.04

.00

.01

.01

.01

.02

.03

.00

.00

.01

t tp<.07. */?<.05. **/?<.01. ***/><.001.

other words, the 360° process itself may induce a "situational"performance orientation (Button et al., 1996), in which individuals,regardless of their dispositional goal orientation, respond with aperformance goal orientation. The feedback format in which self-ratings, others' ratings, and normative data are provided maycreate a situation in which recipients are more concerned abouttheir performance as compared with others than with the develop-mental aspect of the feedback. However, when the "strong" situ-ational cues have subsided, the effects of dispositional goal orien-tation may appear. With time, individuals with a high learningorientation may be more likely to see the value of the feedbackthan individuals with a high performance-prove or avoid orienta-tion. We conducted a post hoc analysis to test whether this rela-tionship changed with time. After we controlled for perceptions ofusefulness at Time 1, the partial correlation between learningorientation and usefulness at Time 2 was significant and positive(r = .17, p < .05). The correlation between performance-proveand usefulness at Time 2 was not significant (r = —.10). Thecorrelation for performance-avoid and usefulness at Time 2 wassignificant and negative (r = —.24, p < .05). Fisher's z testsindicated that the correlation between usefulness and learning goalorientation differed significantly from that for performance-prove

as well as avoid orientation. This finding suggests that individualswith a learning goal orientation may have recovered from thesituationally induced performance context and its concomitantnegative reactions and thus changed their perceptions of the use-fulness of the feedback. Those with performance-prove or avoidorientations did not change their perceptions of the usefulness ofthe feedback with time.

The rating source, although not the focus of this study, showedtwo interesting patterns. First, the findings regarding accuracy thatwere significant for bosses and direct reports did not hold forpeers. For peers, higher ratings were not seen as more accurate. Itis possible that because raters selected the peers who rated them,they were less surprised by their ratings. Or, the participants mayhave had different expectations regarding the ratings they expectedfrom the peers. Perhaps the participants rationalized that lowratings were, in fact, deserved from peers, even though low ratingsfrom peers were related to negative reactions. Nonetheless, theseresults suggest that ratings from different sources are perceiveddifferently. Second, participants did not report negative reactionswhen they received low or discrepant ratings from their directreports. Participants may have expected, given various perfor-mance or interpersonal issues, that some subordinates would rate

Page 11: 360° Feedback: Accuracy, Reactions, and …eportfolio.lib.ksu.edu.tw/~T093000259/repository/fetch...360 Feedback: Accuracy, Reactions, and Perceptions of Usefulness Joan F. Brett

940 BRETT AND ATWATER

Table 5Summary of Results by Rating Source

Finding Hypothesis Boss Direct reports Peers

AccuracyHigher ratings — more accurateMore overratings — less accurate

Positive reactionsHigher ratings — more positive reactionsMore underratings — more positive reactions

Negative reactionsLower ratings — more negative reactionsMore overratings — more negative reactions

Accuracy and reactionsAccuracy (+) related to positive reactionsAccuracy (— ) related to negative reactions

UsefulnessAccuracy ( + ) related to usefulness

Positive reactions (+) with usefulnessNegative reactions (— ) with usefulnessUsefulness and developmental focus (+)Usefulness and defensiveness (-)Goal orientation moderatorExploratory: Goal orientation and usefulness at Time 2

14

25A

35B

6A6B

87A7B9A9B10

YesYes

NoNo

YesYes

NoYes

YesYesYesYesMarginalNoYes

YesYes

NoNo

NoNo

YesNo

Yes

NoNo

NoNo

YesYes

YesNo

No

them lower than or discrepant from their own ratings. Feedbackfrom direct reports appeared not to influence reactions as much asfeedback from bosses and peers. Taken together, these findingssuggest that a better understanding of the value and relevancy offeedback from different sources is needed.

Managerial Implications

The results of this study offer insights for the conduct of 360°programs. This study indicated that negative and discrepant feed-back is seen as less accurate, less useful, and related to negativereactions. It is important for administrators of 360° programs tounderstand that those who receive negative feedback may needdifferent follow-up activities than those who receive positive feed-back. Executive coaches or multiple follow-up sessions may helpthose receiving negative or discrepant feedback to deal with neg-ative reactions and work through them. Indeed, in our study,perceptions of usefulness increased after the one-to-one sessions(but primarily for those with a learning orientation). We can inferthat the facilitator had some impact on this change. It may also beworthwhile to understand specific characteristics of the feedbackrecipients that make them more susceptible to negative reactions(e.g., a performance-prove goal orientation or high self-esteem).

Although it is not surprising that feedback recipients find pos-itive feedback to be most accurate because it often supports theirhigh opinions of themselves, it is surprising that feedback recipi-ents see feedback that they are doing well as most useful when itprovides the least information about development needs. It is alsointeresting that negative feedback is related to negative reactionsbut positive feedback is not related to positive reactions. Facilita-tors may be able to focus some of the feedback recipients' atten-tion on areas in which they received positive ratings and help themfeel good about their high ratings. They also may be able to focussome attention on positive areas to minimize some of the negativereactions to low ratings and suggested development needs. They

also may be able to motivate feedback recipients to continue thebehaviors they are doing successfully.

Given our results, we would caution organizations that areconsidering adopting 360° feedback delivery methods that elimi-nate the costs associated with a facilitator (e.g., mailings, E-mail,or Web). If organizations are to benefit from their investment inthe 360° process, then the costs associated with a facilitator mayensure that those managers who need the feedback the mostreceive special attention.

Limitations

There are a number of limitations that should be consideredwhen one is interpreting the results of this study. First, 360°feedback was conducted as part of an educational experiencerather than as part of the organization in which the recipientworked. Reactions may have been different if recipients weregoing to be expected to continue work-related interactions withtheir feedback providers. In addition, knowledge that the partici-pants were leaving their jobs may have influenced the ratings thatwere provided for the participants. Participants also may have beendifferent from currently employed managers in that part of theirreason for leaving to return to school may have been dissatisfac-tion with their current jobs. Their dissatisfaction may have exac-erbated their negative reactions to ratings from former coworkers.Second, we did not have posttest data to assess who actuallychanged and how perceptions of accuracy, usefulness, and reac-tions were related to subsequent outcomes. We did have facilitatorjudgments about the participants' developmental orientation, butactual outcome data would have certainly added to the interpreta-tion of our findings. Third, we did not collect data on organiza-tional level or organization type. These variables could have in-fluenced the type of feedback that individuals received and,therefore, their reactions to it. Fourth, much of the data that wecollected came from the feedback recipient. Although some mea-

Page 12: 360° Feedback: Accuracy, Reactions, and …eportfolio.lib.ksu.edu.tw/~T093000259/repository/fetch...360 Feedback: Accuracy, Reactions, and Perceptions of Usefulness Joan F. Brett

360° FEEDBACK 941

surements were separated in time, the influence of same-sourcebias on the results should be considered because it could haveartificially raised some of the correlations. Fifth, the facilitatorknew the ratings that each recipient received. There is somepossibility that the facilitator's ratings of the recipient's develop-ment or defensive attitudes could have been affected by thisknowledge. Nonetheless, these results are provocative for futureresearch, suggesting that the outcomes of 360° feedback are com-plex and varied.

Future Research

An assumption underlying many 360° feedback programs is thatas long as feedback is not associated with administrative purposes(e.g., compensation, promotion), it is interpreted as developmentaland not evaluative. Our findings suggest this may not be so. Thedetrimental effects associated with negative performance appraisalresearch were also associated with negative developmental feed-back. If participants are to benefit from the feedback process andif organizations are to recoup their investment in this process,additional research is needed in the following areas. First, a betterunderstanding of the emotional and cognitive reactions to feedbackis needed. Why do higher ratings have no effect on positivereactions? What negative behaviors and repercussions accompanynegative feedback? Do these reactions dissipate over time? If so,how long do negative reactions last? What individual-differencesvariables (e.g., personality, self-esteem) should be considered inattempts to understand reactions to feedback? Second, researchneeds to examine the role of feedback format in recipients' accep-tance of and reactions to feedback. Can feedback format reducenegative reactions and increase perceptions of accuracy and use-fulness? For example, do individuals react less negatively and seefeedback as more useful if normative data are not included? Orperhaps numeric data should be eliminated and only narrativedescriptions of strengths and development needs provided. Whatcan be done in the feedback delivery process to help individualsmanage negative reactions? Do group delivery sessions foster acomparative and evaluative context? How can feedback deliveryfoster a developmental rather than a performance-prove orienta-tion? Are some participants leaving feedback sessions with nega-tive reactions that may later manifest themselves in negativeorganizational attitudes, turnover, retribution, or other negativeconsequences? Third, research needs to consider that those whoreceive positive versus negative feedback may experience a dif-ferent 360° process. Are the positive results found from 360°programs merely capturing the good getting better? Are researchresults on outcomes and consequences of 360° feedback contin-gent on how many recipients in the study received positive versusnegative feedback? The answers to these research questions willhelp organizations understand how to successfully implement 360°programs that will motivate those managers who most need tochange their behaviors to do so.

Conclusion

In sum, our research extends the work done on performanceappraisal feedback by suggesting that even when feedback is usedstrictly for developmental purposes, feedback is perceived as mostuseful when participants get high ratings that confirm their own

self-assessments rather than when they receive information aboutdevelopment needs they did not know they had. Similar to theresearch done with performance appraisal feedback, negative rat-ings are not seen as accurate or useful, and they engender negativereactions. Our results also suggest that recipients may see theprocess as evaluative in nature rather than developmental evenwhen results are confidential and the individuals no longer workwith the feedback providers. Overall, the findings suggest thatresearchers and practitioners need to focus more attention on waysto make the 360° feedback process more developmental and howto minimize negative reactions to negative feedback.

References

Ames, C. (1992). Classrooms: Goals, structures, and student motivation.Journal of Educational Psychology, 84, 261-271.

Antonioni, D. (1996). Designing an effective 360-degree appraisal feed-back process. Organizational Dynamics, 25, 24-38.

Atwater, L. E., Brett, J. F., Waldman, D., & Yammarino, F. (2000).Predictors and outcomes of upward feedback. Unpublished manuscript,Arizona State University West.

Atwater, L. E., Ostroff, C., Yammarino, F., & Fleenor, J. W. (1998).Self-other agreement: Does it really matter? Personnel Psychology, 5],577-598.

Atwater, L. E., Roush, P., & Fischthal, A. (1995). The influence of upwardfeedback on self- and follower ratings of leadership. Personnel Psychol-ogy, 48, 35-60.

Atwater, L. E., Waldman, D., Atwater, D., & Cartier, P. (2000). An upwardfeedback field experiment: Supervisors' cynicism, follow-up and com-mitment to subordinates. Personnel Psychology, 53, 275-297.

Bernstein, B. L., & Lecomte, C. (1979). Supervisory-type feedback effects:Feedback discrepancy level, trainee psychological differentiation, andimmediate responses. Journal of Counseling Psychology, 26, 295-303.

Brett, J. F., & VandeWalle, D. (1999). Goal orientation and goal content aspredictors of performance in a training program. Journal of AppliedPsychology, 84, 863-873.

Button, S., Mathieu, J., & Zajac, D. (1996). Goal orientation in organiza-tional research: A conceptual and empirical foundation. OrganizationalBehavior and Human Decision Processes, 67, 26-48.

Carver, C. S., & Scheier, M. F. (1981). Attention and self-regulation: Acontrol theory approach to human behavior. New York: Springer-Verlag.

Cedarbloom, D., & Lounsbury, J. (1980). An investigation of user accep-tance of peer evaluations. Personnel Psychology, 33, 567-579.

DeNisi, A. S., & Kluger, A. N. (2000). Feedback effectiveness: Can360-degree appraisals be improved? Academy of Management Execu-tive, 14, 129-139.

Dipboye, R. L., & de Pontbriand, R. (1981). Correlates of employeereactions to performance appraisals and appraisal systems. Journal ofApplied Psychology, 66, 248-251.

Dweck, C. S. (1986). Motivational processes affecting learning. AmericanPsychologist, 41, 1040-1048.

Dweck, C. S., & Leggett, E. L. (1988). A social-cognitive approach tomotivation and personality. Psychological Review, 95, 256—273.

Elliot, A. J., & Dweck, C. S. (1988). Goals: An approach to motivation andachievement. Journal of Personality and Social Psychology, 54, 5-12.

Facteau, C. L., Facteau, J. D., Schoel, L. C., Russell, J. A., & Poteet, M.(1998). Reactions of leaders to 360-degree feedback from subordinatesand peers. Leadership Quarterly, 9, 427-448.

Farr, J. L., Hoffmann, D. A., & Ringenbach, K. L. (1993). Goal orientationand action control theory: Duplications for industrial and organizationalpsychology. International Review of Industrial and Organizational Psy-chology, 8, 191-232.

Page 13: 360° Feedback: Accuracy, Reactions, and …eportfolio.lib.ksu.edu.tw/~T093000259/repository/fetch...360 Feedback: Accuracy, Reactions, and Perceptions of Usefulness Joan F. Brett

942 BRETT AND ATWATER

Festinger, L. (1954). A theory of social comparison processes. HumanRelations, 7, 117-140.

Fleenor, J., McCauley, C., & Brutus, S. (1996). Self-other rating agree-ment and leader effectiveness. Leadership Quarterly, 7, 487-506.

Greller, M, & Herold, D. (1975). Sources of feedback: A preliminaryinvestigation. Organizational Behavior and Human Performance, 13,244-256.

Halperin, K., Snyder, C. R., Shenkel, R. J., & Houston, B. K. (1976).Effects of source status and message favorability on acceptance ofpersonality feedback. Journal of Applied Psychology, 61, 142-147.

Heyman, G. D., & Dweck, C. S. (1992). Achievement goals and intrinsicmotivation: Their relation and their role in adaptive motivation. Moti-vation and Emotion, 16, 231-247.

Hezlett, S. A., Ronnkvist, A. M., Holt, K. E., & Hazucha, J. F. (1997). Theprofdor technical summary. Minneapolis, MM: Personnel DecisionsInternational.

Ilgen, D. R., Fisher, C. D., & Taylor, M. S. (1979). Consequences ofindividual feedback on behavior in organizations. Journal of AppliedPsychology, 64, 349-371.

Ilgen, D. R., & Hamstra, B. W. (1972). Performance satisfaction as afunction of the difference between expected and reported performance atfive levels of reported performance. Organizational Behavior and Hu-man Performance, 7, 359-370.

Johnson, J. W., & Ferstl, K. L. (1999). The effects of interrater andself-other agreement on performance improvement following upwardfeedback. Personnel Psychology, 52, 271-303.

Kernis, M., & Johnson, E. (1990). Current and typical self-appraisals:Differential responsiveness to evaluative feedback and implications foremotions. Journal of Research in Personality, 24, 241-257.

Kluger, A. N., & DeNisi, A. (1996). The effects of feedback interventionson performance: A historical review, a meta-analysis, and a preliminaryfeedback intervention theory. Psychological Bulletin, 119, 254-284.

London, M. (1994). Interpersonal insight in organizations: Cognitive mod-els for human resource development. Human Resource ManagementReview, 4, 311-332.

London, M., & Smither, J. (1995). Can multi-source feedback changeperceptions of goal accomplishments, self-evaluation, and performance-related outcomes? Personnel Psychology, 48, 803-839.

Mabe, P., & West, S. (1982). Validity of self-evaluation of ability: Areview and meta-analysis. Journal of Applied Psychology, 67, 280-296.

Meyer, H. (1980). Self-appraisal of job performance. Personnel Psychol-ogy, 33, 291-295.

Morran, K., & Stockton, R. (1980). Effect of self-concept on groupmember reception of positive and negative feedback. Journal of Coun-seling Psychology, 27, 260-267.

O'Reilly, C., & Anderson, J. (1980). Trust and the communication ofperformance appraisal information: The effects of feedback on perfor-mance and job satisfaction. Human Communication Research, 6, 290-298.

Pearce, J. L., & Porter, L. W. (1986). Employee responses to formalperformance appraisal feedback. Journal of Applied Psychology, 71,211-218.

Pedler, M., & Boydell, T. (1980). Is all management development self-development? In J. Beck & C. Cox (Eds.), Advances in managementeducation (pp. 165-196). New York: Wiley.

Podsakoff, P. M., & Farh, J. L. (1989). Effects of feedback sign andcredibility on goal setting and task performance. Organizational Behav-ior and Human Decision Processes, 44, 45—67.

Reilly, R. R., Smither, J. W., & Vasilopoulos, N. L. (1996). A longitudinalstudy of upward feedback. Personnel Psychology, 49, 599-612.

Russell, J. A. (1980). A circumplex model of affect. Journal of Personalityand Social Psychology, 39, 1161-1178.

Schrauger, J. S. (1975). Responses to evaluation as a function of initialself-perceptions. Psychological Bulletin, 82, 581-596.

Shrout, P. E., & Fleiss, J. L. (1979). Intraclass correlations: Uses inassessing rater reliability. Psychological Bulletin, 86, 420-428.

Stone, D., & Stone, E. (1985). The effects of feedback consistency andfeedback favorability on self-perceived task competence and perceivedfeedback accuracy. Organizational Behavior and Human Decision Pro-cesses, 36, 167-185.

Swann, W., & Read, S. (1981). Acquiring self-knowledge: The search forfeedback that fits. Journal of Personality and Social Psychology, 41,1119-1128.

Taylor, M. S., Fisher, C. D., & Ilgen, D. R. (1984). Individuals' reactionsto performance feedback in organizations: A control theory perspective.In K. M. Rowland & G. R. Ferris (Eds.), Research in personnel andhuman resources management (pp. 81-124). Greenwich, CT: JAI Press.

VandeWalle, D. (1997). Development and validation of a work domaingoal orientation instrument. Educational and Psychological Measure-ment, 57, 995-1015.

Waldman, D., Atwater, L. E., Clements, D., & Atwater, D. (1996, August).Attitudinal and behavioral outcomes of an upward feedback process.Paper presented at the National Meeting of the Academy of Manage-ment, Cincinnati, OH.

Received May 4, 2000Revision received November 6, 2000

Accepted November 6, 2000