Does Negative Feedback Benefit (or Harm) Recipient

r Academy of Management Journal2020, Vol. 63, No. 2, 584–612.https://doi.org/10.5465/amj.2016.1196

DOES NEGATIVE FEEDBACK BENEFIT (OR HARM)RECIPIENT CREATIVITY? THE ROLE OF THE DIRECTION OF

FEEDBACK FLOW

YEUN JOON KIMUniversity of Cambridge

JUNHA KIMOhio State University

Negative feedback alerts recipients to a creativity–standard gap, and thus may offer anopportunity to improve creativity. However, existing theories and empirical evidenceare contradictory, with the literature containing evidence of positive, negative, and nullrelationships between negative feedback and recipient creativity. The goal of our re-search is twofold: first, to organize the contradictory theories under a comprehensivetheoretical framework, and, second, to resolve the inconsistency between negativefeedback and recipient creativity. Across two studies—a quasi-field experiment and alaboratory experiment—we find that the direction of feedback flow determines the na-ture of the relationship between negative feedback and recipient creativity, via twodistinct mechanisms: task processes and meta-processes. Negative feedback increasesrecipient creativity in the bottom-up feedback flow (from followers to supervisors), be-cause it heightens the recipients’ focus on task processes, whereby the recipients focuson the generation of better task strategies to close the creativity–standard gap. In con-trast, in the top-down (from supervisors to followers) or lateral (between peers) feedbackflows, negative feedback heightens the recipients’ focus on meta-processes—a psycho-logical state in which recipients feel threatened by negative feedback—and thus hindersrecipient creativity.

Employee “creativity”—defined as the productionof ideas that are both novel and useful (Amabile,1983; Oldham & Cummings, 1996)—is a foundationof organizational success (Anderson, Potocnik, &Zhou, 2014). It allows organizations to continuallyproduce innovative products and keep them com-petitive in the market. Accordingly, understandinghow to improve employee creativity has been alongstanding preoccupation of management scholars(e.g., George, 2007; Perry-Smith, 2006). Since crea-tivity involves a departure from the current ways ofthinking and behaving, employees often attempt to

provide other organizational members with negativefeedback to create dissatisfaction with the status quoor the current levels of creativity (Ilgen, Fisher, &Taylor, 1979). Negative feedback highlights problemswith current creativity, generating awareness of a gapbetween current creativity and the standards. Oncethe gap is recognized, employeesmay bemotivated toclose the gap by improving their current creativity.

However, this argument has received limited em-pirical support. In fact, the evidence is completelyequivocal. Some scholars suggest that negative feed-back has no direct effect on recipient creativity(Fodor & Carver, 2000; George & Zhou, 2001), whileothers suggest that negative feedback inhibits it(Ilies & Judge, 2005; Van Dijk & Kluger, 2011; Zhou,1998). We know of only three studies (Fang, Kim, &Milliken, 2014; Ford & Gioia, 2000; Vuori & Huy,2015) that provide evidence to support that negativefeedback might be positively associated with recipi-ent creativity. Such perplexing empirical evidenceindicates that a basic question remains unanswered:“How and why does negative feedback influencecreativity?”

We wish to thank our editor Jill Perry-Smith and threeanonymous reviewers for their insightful and constructivefeedback. We also thank Angelo DeNisi, Matthew Fein-berg, Soo Min Toh, Jacob Hirsh, Geoffrey Leonardelli,Andrew Hayes, and Sojin Park for their constructivefeedback. Yeun Joon Kim led the entire process of pub-lishing this article. Junha Kim mainly contributed to liter-ature review andStudy 2. Correspondence concerning thisarticle should be addressed to Yeun Joon Kim, JudgeBusiness School, University of Cambridge. e-mail: [email protected].

584

Copyright of the Academy of Management, all rights reserved. Contents may not be copied, emailed, posted to a listserv, or otherwise transmitted without the copyright holder’s expresswritten permission. Users may print, download, or email articles for individual use only.

https://doi.org/10.5465/amj.2016.1196

mailto:[email protected]


To answer this question, the present researchdraws upon feedback intervention theory (Kluger &DeNisi, 1996) to derive a parsimonious and coherenttheoretical account of the link between negativefeedback and creativity. Feedback intervention the-ory argues that negative feedback makes feedbackrecipients awareof thegapbetween their current levelof creativity and the standards (the “creativity–standard gap”) and that such awareness leads the re-cipients to engage in one of two functionally oppositemechanisms in response to the negative feedback.The first mechanism concerns “task processes,”whereby recipientsmake constructive improvementsby engaging in the process of generating better taskstrategies. They identify problems with their currentbehavior in creativity tasks, design more useful andnovel strategies for their creativity tasks, and im-plement those strategies. The second mechanisminvolves “meta-processes,” and refers to the psy-chological state in which recipients feel threatenedby negative feedback. Feedback recipients who en-gage in meta-processes feel that their ego, or self-concept, is threatened by the negative feedback,deterring them from experimentations and creativeattempts to improve their creativity. Despite itsusefulness in illustrating why negative feedback isinconsistently related to recipient creativity, a no-table limitation of feedback intervention theory isthat it does not elaborate onwhen negative feedbackrecipients attend to either of the two processes.

The main objectives of this paper are to organizethe inconsistent theories and empirical findingsunder the two mechanisms—task processes andmeta-processes—and to resolve the inconsistencybetween negative feedback and recipient creativityby introducing an important, but neglected, bound-ary condition: the direction of feedback flow. Thedirection of these flows include bottom up (i.e., fromfollowers to supervisors), top down (i.e., from su-pervisors to followers), and lateral (i.e., from peers topeers). We suggest that, in the bottom-up feedbackflow, negative feedback increases recipient creativ-ity through task processes, whereas, in the top-downand lateral feedback flows, negative feedback de-creases recipient creativity through meta-processes.In organizations, supervisors have asymmetric con-trol over valuable organizational resources, such asmonetary rewards, promotions, training opportuni-ties, and budgets and materials for completing tasks.Followers do not have such control, and their super-visors determine their access to organizational re-sources. This power asymmetry often causes these twoparties to have completely different psychological

mindsets (Keltner, Gruenfeld, & Anderson, 2003;Magee & Galinsky, 2008). Researchers have found thatpower asymmetry leads the powerful (e.g., supervisors)to be approachoriented (vs. inhibition oriented) towardnegative evaluations by the powerless (e.g., followers),while the powerless become inhibited by negativeevaluations by the powerful (Keltner et al., 2003). Inaddition, the powerful tend to maintain high levels oftask focus in the face of task-related criticisms and careless about their social relationships with feedbacksenders, whereas the powerless behave in an op-posite way (Galinsky, Magee, Gruenfeld, Whitson,& Liljenquist, 2008; Smith, Jostmann, Galinsky, &Van Dijk, 2008; Steele, Spencer, & Aronson, 2002).Based on these findings, we expect that negativefeedback increases recipient creativity via taskprocesses in the bottom-up feedback flow and de-creases recipient creativity via meta-processes inthe top-down feedback flow.

The relationship between the feedback sender andthe recipient in the lateral feedback flow is qualita-tively different from that in the bottom-up and top-down feedback flows. A peer relationship does notinvolve differential social power. Instead, this rela-tionship is characterized by rivalry, or competition,given that organizational resources (e.g., promotionsand pay increases) are limited. Thus, employeesstrive to stand out among their peers and becomeincreasingly concerned about the possibility of lag-gingbehind them(Bandura, 1977;DeNisi, Randolph,& Blencoe, 1983; Festinger, 1962). Peer competitionsometimes produces positive organizational out-comes (for a review, see Birkinshaw, 2001). How-ever, regarding negative feedback between peers,evidence has shown that the competitive and non-hierarchical nature of a peer relationship leads em-ployees to interpret lateral negative feedback as anattempt to downplay their ability and an attack ontheir self-esteem. For this reason, those who receivelateral negative feedback have reported that they feelthreatened, distracted, and discouraged (Brett &Atwater, 2001; DeNisi et al., 1983; Druskat & Wolff,1999; Rogers & Feller, 2016). That is, negative feed-back from peers distracts recipients from creativitytasks—low task processes—and causes them to paygreater attention to meta-processes, which reducestheir creativity.

The current research tests these hypotheses in twostudies: Study 1 is a quasi-field experiment at a Ko-rean company and Study 2 is a laboratory experi-ment at a large North American university. Bydemonstrating consistent support for our hypothesesacross the two studies, our researchmakes important

2020 585Kim and Kim

contributions to the literature. We not only re-solve the inconsistency of the relationship be-tween negative feedback and recipient creativitybut also integrate several theoretical arguments,variably applied in past research, under two es-sential mechanisms: task processes and meta-processes. In addition, we push the boundary offeedback intervention theory by providing an im-portant, albeit neglected, boundary condition—thedirection of feedback flow—which determines themechanisms underlying the relationship betweennegative feedback and creativity. Figure 1 depictsour theoretical framework.

THEORETICAL DEVELOPMENT

Defining Negative Feedback

“Feedback” herein refers to information regardingwhether one’s level of creativity meets the orga-nizational standard(s). When feedback providesinformation indicating a “creativity–standard gap”—that is, a discrepancy showing that the demonstratedlevel of creativity is below the accepted standard—thefeedback is considered “negative.” We propose to in-vestigate negative feedback in relation to recipientcreativity for two reasons. First, from a theoreticalperspective, negative feedback has strong potential tocontribute to recipient creativity. If the feedback re-cipients are receptive, negative feedback identifies theinsufficiency in their current level of creativity andcreates dissatisfaction where none previously existed.This dissatisfaction is important because creativity isoften derived from the rejection of previous thoughtpatterns and behavior, and negative feedback informsfeedback recipients that their current levels of crea-tivity should be improved in more novel and usefulways (George, 2007; Zhou & George, 2001). Second,negative feedback is prevalent in organizations and aprimary means by which managers influence em-ployee behavior and performance (Ilgen et al., 1979).For example, according to Zenger and Folkman’s(2017) study of 328 managers and approximately4,200 followers (13 followers permanager on average),the managers perceived themselves as effective lead-ers when they offered criticism or negative feedback,and, thus, they frequently offered negative feedbackto followers to induce meaningful changes in thefollowers’ behavior and performance. This led thefollowers to perceive that negative feedback waspervasive in their organizations. Considering itsutility for recipient creativity and the pervasive-ness in organizations, we believe that a systematic

investigation of negative feedback and its effects onrecipient creativity is important. In the followingsections, we review past research on the relation-ship between negative feedback and recipient cre-ativity and identify both theoretical and empiricalinconsistencies in this relationship.

Inconsistent Theories and Empirical Evidence onNegative Feedback and Creativity

Our comprehensive review of past research on thelink between negative feedback and recipient crea-tivity revealed perplexing results; past studies haveused a variety of contradicting theories and havereported a mix of positive, negative, and null rela-tionships. Table 1 summarizes the contexts and re-sults of past studies.1

Researchers who found a positive relationshipbetween negative feedback and creativity shared asimilar perspective (Fang et al., 2014; Vuori & Huy,2015). These authors argued that negative feedbackcreates dissatisfaction with their current level ofcreativity. This dissatisfaction in turn incentivizesfeedback recipients to look closely at the processesinvolved in creativity tasks in order to identify op-portunities for improvement and fill the gap byimplementing better task strategies. In contrast, theresearchers who found a negative relationship be-tween negative feedback and recipient creativityfocused on the perceived threat resulting from neg-ative feedback (Van Dijk & Kluger, 2011; Zhou,1998). This perspective suggests that negative feed-back threatens recipients’ core beliefs about them-selves, their abilities, and their status in the eyes ofothers and that the experience of a threatened self-concept reduces creativity because threatened re-cipients disengage from any experimentation andcreative attempts to improve their creativity.

Both of these theoretical perspectives reveal fun-damental aspects of the negative feedback–creativityrelationship, and, therefore, the present researchstrives to integrate the two perspectives into abroader framework. To accomplish this integration,we utilize feedback intervention theory (Kluger &DeNisi, 1996), which parsimoniously incorporatesthe two theoretical perspectives.

1 Two studies in this section (Fodor & Carver, 2000;Zhou, 1998) did not report the main effect of negativefeedback on creativity. Thus, we manually calculated theeffect using themeans, standard deviations, and degrees offreedom reported in the papers.

586 AprilAcademy of Management Journal

Reconciling the Inconsistent NegativeFeedback–Creativity Relationship

The primary theoretical innovation of feedbackintervention theory (Kluger & DeNisi, 1996) lies inthe introduction of two separate processes—taskprocesses and meta-processes—that are used to ex-plain how a feedback recipient responds to negativefeedback. “Task processes” refer to the mechanisms

by which feedback recipients attempt to improvetheir current creativity by generating better, diversestrategies for their creativity tasks. In creativity tasks,the most common pitfall is that of using existing orroutinized task strategies. Employees who work oncreativity tasks tend to overly rely on existing taskstrategies, or task routines, and they are less inclinedto experiment with new task strategies. This is be-cause employees routinize existing task strategies

TABLE 1Summary of Studies Linking Negative Feedback and Recipient Creativity

Study and Context Feedback Directions Findings

Laboratory experiment with undergraduate students(Zhou, 1998)

Top down [Negative] Participants who received negativefeedback showed lower creativity than those whoreceived positive feedback.

Laboratory experiment with undergraduate students(Fodor & Carver, 2000)

Top down [Null] Creativity in the negative feedback conditionwas not statistically different from creativity in thepositive feedback and control conditions.

Interviews and survey with managers in the easternUnited States (Ford & Gioia, 2000)

Bottom up [Positive] Negative feedback was positively related todecision creativity.

Field survey with professional employees (George &Zhou, 2001)

Top down [Null] There was a non-significant correlationbetween feedback valence and creative behavior.

Laboratory experiments with undergraduate students(Ilies & Judge, 2005)

Top down [Negative] Participants who received negativefeedback showed downward goal revision inRemote Associates Test.

Laboratory experiments with undergraduate students(Van Dijk & Kluger, 2011)

Not specified [Negative] Negative feedback lead to a reducednumber of ideas generated compared to positivefeedback.

Field survey with professional employees (Hon et al.,2013)

Top down [Null] There was a non-significant correlationbetween negative feedback and creativity.

A simulation model study (Fang et al., 2014) Not specified [Positive]Sugarcoatingordistortingnegative feedbackreduced organizational learning, which in turndecreased innovation.

Qualitative interview study with Nokia’s managersand external experts (Vuori & Huy, 2015)

Bottom up [Positive] Middle managers blocked the flow ofnegative feedback to upper management, whichhampered innovation.

FIGURE 1Theoretical Framework

Direction of Feedback Flow

1.2.3.

Bottom-up Feedback FlowTop-down Feedback FlowLateral Feedback Flow

RecipientCreativity

Negative FeedbackTargeted at Creativity

Recipient ReactionsToward Negative Feedback

Task Processes

Meta-Processes

2020 587Kim and Kim

over time through repeated exposure (Kilduff, 1993;March, 1991; March & Simon, 1958). Employees’tendency to stick to existing task strategies couldincrease their routine performance because it en-sures that employees reliably perform their routinetasks without errors. However, this tendency is det-rimental to creativity because creativity often re-quires deviation from routines and the status quo(Anderson et al., 2014; Zhou & George, 2001). In-deed, the evidence has shown that employees in-creased their creativity by experimenting with newways of performing their tasks, while sticking toexisting task strategies hampered their creativity(Anderson et al., 2014; Kim & Zhong, 2017; March,1991). In other words, task processes are essentialmechanisms for employee creativity. Task processesare highly congruent with the theories used in pastresearch that found a positive relationship betweennegative feedback and creativity (e.g., Fang et al.,2014; Vuori & Huy, 2015).

Alternatively, “meta-processes” refer to the mech-anism whereby negative feedback threatens recipi-ents’ beliefs about their self-concept, their ability toperform creativity tasks, and their social image per-ceived by feedback senders. Responding to negativefeedback by attending to meta-processes, recipientsdirect their attention toward the threatening conse-quences of negative feedback. Specifically, recipientsworry about how negative feedback affects the as-sessments of their own ability (e.g., “Does this feed-back suggest that I’m incompetent?”) or whethernegative feedback implies changes in the recipients’important social relationships (e.g., “Does this meanthat my supervisor does not like me?”). Thisthreatened mindset inhibits recipients from takingrisks by experimenting with creative ideas (Forster,Friedman, & Liberman, 2004; Friedman & Forster,2001; Kluger & DeNisi, 1996) and lowers the mentalresources and cognitive capacity that should be al-located to creative processes (see, for a review,Byron, Khazanchi, & Nazarian, 2010). Thus, meta-processes prevent recipients fromdirectly addressingthe problemswith their current level of creativity andleave them in a threatened mindset, which in turnreduces their creativity. Meta-processes are in linewith the logic presented by the past research thatfound a negative relationship between negativefeedback and recipient creativity (e.g., Van Dijk &Kluger, 2011; Zhou, 1998).

In summary, feedback intervention theory sug-gests that negative feedback has the potential to bothincrease and decrease creativity via task processesand meta-processes. Therefore, it is not surprising

that past researchers have reported contradictoryempirical evidence on the relationship betweennegative feedback and recipient creativity. Despiteits usefulness in understanding the inconsistent re-lationship, feedback intervention theory does notoffer insight into when negative feedback recipientsadopt one of these two processes. Our research aimsto expand feedback intervention theory by investi-gating an important, albeit neglected, boundary con-dition that channels negative feedback to creativitythrough task processes and meta-processes. We sug-gest that the direction of feedback flow is the sine quanon for understanding the relationship between neg-ative feedback and recipient creativity and resolvesthe inconsistency in this relationship.

The Role of the Direction of Feedback Flow

We define the “direction of feedback flow” as thetransfer of feedback from sender(s) to recipient(s)wherein the two parties have the same or differentorganizational ranks. We argue that the direction offeedback flow should not be isolated from feedbackresearch because feedback does not occur sponta-neously in organizations. Rather, it occurswithin thesocial, organizational contexts that create the basis ofthe evaluation standards. Employees try to meet thestandards, and evaluators rate employee creativityby considering the gap between demonstrated crea-tivity and the standards (De Stobbeleir, Ashford, &Buyens, 2011; Zhou, 2008). As a result, in a study offeedback, it is necessary for researchers to specify thesocial and organizational contexts, including theidentities of the feedback sender(s) and recipient(s)(Zhou, 2008). By introducing a novel concept—thedirection of feedback flow—the current researchspecifies the social and organizational context withinwhich negative feedback flows, and examines its in-fluences on the relationship between negative feed-back and recipient creativity.

Specifically, we investigate three directions offeedback flow: “bottom up” (from followers to su-pervisors), “top down” (from supervisors to fol-lowers), and “lateral” (frompeers to peers).Althoughthe top-down feedback flow may be the most fre-quently observed direction, organizations are be-ginning to realize the value of feedback that flowsup the hierarchy and feedback that flows laterallybetween employees at the same organizational rank(Antonioni, 1996; Brett & Atwater, 2001; DeNisi &Kluger, 2000). Nevertheless, a comprehensive andsystematic investigation on the roles of the direc-tion of feedback flow is lacking in organizational


research. The majority of past research has consid-ered only top-down feedback (e.g., Fodor & Carver,2000; George & Zhou, 2001; Zhou, 1998). A smallminority of the research has looked at bottom-upfeedback (e.g., Ford & Gioia, 2000; Vuori & Huy,2015), and other research did not specify thesenders and recipients (e.g., Fang et al., 2014; VanDijk & Kluger, 2011). Interestingly, researchers whoexamined the top-down feedback flow generallyfound a negative relationship between negativefeedback and creativity, while those who investi-gated the bottom-up feedback flow found a positiverelationship. These results show the importance ofthe direction of feedback flow as a boundary con-dition in the relationship between negative feed-back and creativity. We theorize that each directionof feedback flow determines a feedback recipient’sfocus—that is, toward task processes versus meta-processes—and its subsequent effect on creativity.

Bottom-Up and Top-Down Negative FeedbackFlows and Creativity

The bottom-up feedback flow.Research on socialpower provides a basis for understanding howbottom-up and top-down feedback flows influencethe relationship between negative feedback andcreativity. “Social power” refers to asymmetric con-trol over valuable resources, rewards, punishments,and outcomes (Keltner et al., 2003;Magee &Galinsky,2008). Differences in hierarchical rank create a situa-tion in which an employee at a higher rank possessessocial power over other employees at lower ranks;accordingly, manipulations of the hierarchical rankhave been frequently used to study social power andits effects (e.g., DeNisi et al., 1983; Jordan, Sivanathan,& Galinsky, 2011).

Our research relies on the social power literatureto suggest that the bottom-up feedback flow enablesnegative feedback recipients (in this case, supervi-sors) to attend to task processes rather than to meta-processes. Regardless of their findings, the feedbackresearchers listed in Table 1 acknowledged thatnegative feedback hurts recipients’ feelings to a cer-tain degree, as it criticizes some aspects of theircurrent creativity. To utilize negative feedback con-structively, recipients should understand it strictlywithin the boundary of the tasks rather than expand-ing its implications to personal or task-unrelatedmatters (e.g., concerning their image in the eyes ofthe feedback sender). Research on social power hasshown that supervisors may be able to do so betterthan followers, for several reasons. Onemajor reason

for this difference is that supervisors tend to bemoreapproach oriented toward the dissenting, counter-attitudinal opinions provided by their followers.Supervisors are aware that their followers gener-ally cannot engender serious social consequences(e.g., pay decreases and demotions) as followershave neither the formal authority nor social power todo so (Bacharach & Lawler, 1980; Emerson, 1962;Pfeffer & Salancik, 1974; Yukl, 2010). Such aware-ness helps supervisors better cope with the uncom-fortable feelings that negative feedback elicits and bemore approach oriented to the potential positiveoutcomes that could be attained by changing theirbehaviors and correcting the problems. In line withthis argument, Keltner et al. (2003: 268–269) sug-gested that:

The experience of power involves the awareness thatone can act at will without interference or serioussocial consequences . . . Being unconstrained byothers’ evaluations or the consequences of one’s ac-tions, peoplewith elevated power should be disposedto elevated levels of approach-related affect, cogni-tion, and behavior.

In addition, social power tends to increase task-focused and goal-directed behaviors (Karremans &Smith, 2010; Smith et al., 2008; Steele et al., 2002).As mentioned earlier, the main purpose of provid-ing negative feedback is to identify the creativity–standard gap inorder to help recipients close this gapby improving their current level of creativity. How-ever, negative feedback often distracts recipients’focus from their task and directs their attention totask-irrelevant matters, such as concern about theirimage in the eyes of others (Kluger & DeNisi, 1996).Therefore, as long as recipients can maintain theirfocus on their creativity task, they are more likely toutilize negative feedback to improve creativity. Evi-dence from the social power literature has shownthat the powerful, relative to the powerless, canbetter plan andupdate task-relevant information andsuppress their attention to task-irrelevant informa-tion in order to achieve their goals (Smith et al.,2008; Smith & Trope, 2006), even when the infor-mation threatens their self-image (Beilock, Rydell, &McConnell, 2007; Steele et al., 2002).

Finally, social distance theory (Magee & Smith,2013) provides a relational perspective regardingwhy the bottom-up feedback flow may positivelyrelate negative feedback to recipient creativity. Thistheory suggests that high-power employees havea greater sense of social distance; namely, “a sub-jective perception or experience of distance from

2020 589Kim and Kim

another person or other persons” (Magee & Smith,2013: 2).With the heightened level of social distance,high-power employees pay less attention to their so-cial relationships with others; instead, they tend tostrengthen their focus on the achievement of ultimategoals and maintain high levels of self-control in theprocess of goal pursuit. Furthermore, in anticipationof social disapproval, the sense of social distancekeeps high-power employees from feeling sociallyengaging emotions, such as embarrassment and feel-ing threatened, since they value communal, intimatesocial relationships less (Magee&Smith, 2013). Thus,in the bottom-up negative feedback flow from fol-lowers to supervisors, supervisors’ sense of socialdistance is likely to help themmaintain their focus oncreativity tasks and overcome the ego-threateningimplications of negative feedback. In summary, in thebottom-up feedback flow, supervisors likely utilizenegative feedback for their creativity by attending totask processes rather than meta-processes.

The top-down feedback flow. In contrast to thebottom-up feedback flow, the top-down feedbackflow may lead feedback recipients (i.e., followers) torespond to negative feedback by attending to meta-processes instead of task processes, as the asym-metrical social power in the supervisor–followerrelationship likely leads followers to be more vigi-lant to criticism from their supervisor and moreconcerned about their image in the eyes of their su-pervisors. In general, employees have a strong desireto maintain their membership in the organizationand move up the organizational hierarchy. To do so,they need favorable evaluations and support fromtheir supervisors as supervisors have the power tosatisfy or impede their desires. As stated by Keltneret al. (2003: 269):

Less powerful individuals have less access to mate-rial, social, and cultural resources and are more sub-ject to social threats and punishments. Thus, they aremore sensitive to the evaluations and potential con-straints of others.

Furthermore, as suggested above, people withlow social power, relative to those with high socialpower, tend to decrease their task-focused attentionand fail to suppress their attention to task-irrelevantinformation (Smith et al., 2008; Smith&Trope, 2006).This is evenmore pronouncedwhen information thatthreatens the recipients’ self-image (e.g., negativefeedback) is present (Beilock et al., 2007; Steele et al.,2002). Finally, low-power people tend to have alow social distance and value social relationshipswith others, rendering themmore vigilant to negative

feedback and criticisms from others (Magee & Smith,2013). Taken together, we argue that the top-downfeedback flow likely directs feedback recipients’ (i.e.,followers’) focus away from task processes to meta-processes, which is detrimental to their creativity.

Lateral Negative Feedback Flow and Creativity

While the bottom-up and the top-down feedbackflows involve hierarchical relationships in whichsocial power is asymmetrical, the lateral feedbackflow does not involve social power differences be-tween senders and recipients. Instead, it is oftencharacterized as a competitive relationship in whichpeers strive to attain limited organizational resources(e.g., promotions, pay increases, or training opportu-nities; Magee & Galinsky, 2008). Rarely, in organiza-tions with a social hierarchy that does not resemble apyramid, competitionmay not be an essential featurebecause such organizations have several higher po-sitions available, allowing all employees to move upthe hierarchy. “Many real-world situations, by con-trast, offer rewards that depend on an individual’sperformance relative to others” (O’Keeffe, Viscusi, &Zeckhauser, 1984: 27), and the number of people atthe top is significantly smaller than the number ofpeople at the bottom (Magee & Galinsky, 2008). Thus,competition is an inherent characteristic that definesemployee relationships in most organizations.

However, this does not mean that the workplace is“the war of all against all.” Instead, employees mostlikely compete only with their “reference group,” aterm that refers to a group of employees whose be-haviors and performance are compared against eachother (Bandura, 1977; Festinger, 1962). The selectionof a reference group is based on shared similarities inmembers’ experience and ability (Festinger, 1954;Mumford, 1983). In organizations, as peers havesimilar levels of work tenure, experience, ability,and knowledge, peer groups become the referencegroups (Bandura, 1977, 1978, 2001). Most impor-tantly, social comparisons based on appropriatereference groups are related to employee percep-tions of organizational justice. Unfair comparisonsfor performance ratings (e.g., between followers andleaders) significantly undermine employee per-ceptions of both distributive and procedural fairness(Bandura, 1977; Festinger, 1962), which results indetrimental organizational outcomes, such as highlevels of turnover, low commitment, low job satisfac-tion, and low job performance (see, for a review,Colquitt, Conlon, Wesson, Porter, & Ng, 2001). There-fore, organizations assess employee performance


through social comparisons within a peer group andprovide employees with results in the form of per-formance feedback (Mumford, 1983).

Regarding the lateral negative feedback flow, thenonhierarchical and competitive nature of peer rela-tionships is likely to direct recipients’ focus towardmeta-processesandaway fromtaskprocesses. Inapeerrelationship, recipients likely interpret lateral negativefeedback in an unproductive way because they areconcerned about the possibility of lagging behind theircompetitors and because they strive to stand outamong their peers (Bendersky & Shah, 2012; Cho,Overbeck, & Carnevale, 2011; DeNisi et al., 1983;Rogers & Feller, 2016; Tauer & Harackiewicz, 1999).For this reason, lateral negative feedback “could beviewed as an attack on self-esteem” and as an attemptto downplay a competitor’s abilities (DeNisi et al.,1983: 458). Several researchers have provided empir-ical evidence that lateralnegative feedback induces thenegative responses of feedback recipients, particularlyin relation tometa-processes. For instance, DeNisi et al.(1983)designeda longitudinal laboratoryexperiment totest whether lateral negative feedback causes recipientsto exhibit negative reactions. They manipulated thelateral negative feedback by simply indicating that itcame from a peer, and they found that recipients dem-onstrated unproductive reactions. The lateral negativefeedback decreased recipients’ satisfaction with andtrust in (or cohesiveness with) their peers; ultimately, itsignificantly impaired the recipients’ task focus.

Following DeNisi et al. (1983), several other re-searchers have provided further evidence that lateralnegative feedback threatens recipients’ self-conceptand distracts from their task focus. For example,Druskat and Wolff (1999) found that, even thoughlateral negative feedback had a developmental pur-pose, recipients perceived it as detrimental to peercommunication, recipient development, peer cohe-sion, satisfactionwith peers, and the viability of theirpeer group. Brett and Atwater (2001) offered more di-rect evidence that lateral negative feedback threatensrecipient ego and self-image. In their study, recipientsof lateral negative feedback reported that they felt dis-couraged, criticized, and confused by this feedback.More recently, Rogers and Feller (2016) showed that,when people identified a negative gap between theircreative performance (i.e., levels of excellence in essaywriting) and a peer’s creative performance (i.e., thereference point), they felt that their self-image andperceived abilities were threatened. Participants re-ported that they felt discouraged and disappointed bythe gap and that they were not the right person for thecreativity task. In addition, the recipients perceived

that their writing would not attain the same levels ofexcellence as their peers’ essays, and, as a result, theirtask focus was significantly distracted. Relying onthese findings, we suggest that the lateral negativefeedback flow likely leadsnegative feedback recipientsto attend to meta-processes rather than to task pro-cesses, which will ultimately reduce their creativity.

To summarize, we suggest that the direction of feed-back flow is an important boundary condition of thenegative feedback and creativity relationship. Thebottom-up negative feedback helps recipients come upwith better task strategies for their creativity tasks (taskprocesses), which subsequently increases their creativ-ity. In contrast, top-down and lateral negative feedbackdirects recipients’ focus away from creativity tasks totask-irrelevant matters, such as concerns about theirability, their social relationshipswith feedback senders,and their self-concept (i.e., meta-processes), whichprevents them from engaging in experimentation andcreative attempts, and, in turn, decreases their creativ-ity. Thus, we propose the following two hypotheses:

Hypothesis 1. The relationship between negativefeedback and recipient creativity is moderated by thedirection of feedback flow; in the bottom-up feedbackflow, negative feedback is positively associated withrecipient creativity, whereas, in the top-down andlateral feedback flows, negative feedback is nega-tively associated with recipient creativity.

Hypothesis 2. Task processes and meta-processesmediate the relationship between negative feedback,the direction of feedback flow, and recipient creativ-ity; in the bottom-up feedback flow, task processesmediate the positive relationship between negativefeedback and recipient creativity, whereas, in the top-down and lateral feedback flows, meta-processesmediate the negative relationship between negativefeedback and recipient creativity.

OVERVIEW OF THE STUDIES

We conducted two studies—one quasi-field ex-periment and one laboratory experiment—to testour theoretical model. Both studies tested the fullmoderatedmediationmodel. In Study 1,we recruited225 employees who were working in creative jobs—designing new products, researching and developingnew products, and developing marketing plans—in aKorean company. Employees were quasi-randomlyassigned to one of three conditions pertaining to thedirection of feedback flow: bottom up, top down, orlateral. The company employed bottom-up, top-down, and lateral feedback in their quarterly evalua-tions. Relying on this natural organizational setting,

2020 591Kim and Kim

we measured employee perception of the extent towhich their quarterly evaluation was negative. Crea-tivity was measured by focal employees’ superiorstwo months after the quarterly evaluations. Study 2sought to replicate the results of Study 1 with 356undergraduate students in a large North Americanuniversity. We conducted a laboratory experiment toprovide evidence of causality by manipulating bothnegative feedback and the direction of feedback flow.

STUDY 1: QUASI-FIELD EXPERIMENT

Procedure

We conducted the study in the product develop-ment and management department at a health foodcompany in Korea. The main duties of employees inthis department were (a) designing new products, (b)researching and developing new products, and (c)developing marketing plans (e.g., promotion, adver-tisement) for new products. According to our inter-views with the three executives of this company,creativity is the defining job characteristic of em-ployees in thedepartment, and, in formal performanceevaluations, employees are rated regarding how crea-tive they have been in their jobs. The company oper-ates with a team-based structure, and one team leadermanages each team. Each team also has a deputy teamleader, who acts as a team leader when necessary. Theteam leader reports to his/her division head. In ourstudy, we measured the top-down, bottom-up, andlateral feedback received by focal employees in thenatural context of the company’s official quarterlyevaluations. Twomonths later, we collected ratings ofthe focal employees’ creativity from their superiors.

The company provides employees with quarterlyperformance feedback—in March, June, September,and December. In alternating quarters, team mem-bers (or followers in each team) receive lateral feed-back from their peers (in March and September) andtop-down feedback from their team leader (in Juneand December). Team leaders also receive feedbackfrom different sources in alternating quarters. Sinceteam leaders have limited contact with their peers,they receive bottom-up feedback from their teammembers, instead of receiving lateral feedback inMarch and September. In June and December, teamleaders receive feedback from their division heads.The feedback that employees receive (regardless ofits source) includes both a numerical evaluation andwritten feedback. The numerical evaluation rangesfrom 0% to 100%; the higher the score, the better theemployee performance. In the written feedback, a

feedback sender specifies a recipient’s strengths,weaknesses, and any other information that thesender thinks is important for improving the recipi-ent’s task behavior.2 We coordinated our data col-lection with an executive member of the humanresourcesdepartment,whoallowedus to collect datafrom one focal employee per team.

Relying on the unique, natural setting of thiscompany, we designed a quasi-field experiment us-ing March and June quarterly evaluations. We cre-ated the quasi-field experimental conditions,manipulating the direction of feedback flow by se-lectively inviting participants in either the March orthe June evaluations. That is, we created three con-ditions, and each condition contained about onethird of the sample. The bottom-up feedback flowcondition contained team leaders who receivedfeedback from their team members in the Marchevaluation. The lateral feedback flow conditioncontained team members who received feedbackfrom their peers in the March evaluation. The top-down feedback flow condition contained teammembers who received feedback from their super-visors in the June evaluation. To create the threeconditions, in the beginning ofMarch,we sent out aninvitation email to all team members and teamleaders at the product development and manage-ment department. After receiving responses fromemployeeswhowerewilling to takepart,we selectedone participant from each team and assigned them toa condition before they received their quarterlyfeedback at the end of March. When multiple teammembers from the same team volunteered for this

2 Thenumerical evaluations in thebottom-upand lateralfeedback flows are averaged across raters, as there aremultiple raters (i.e., multiple followers for the bottom-upfeedback flow and multiple peers for the lateral feedbackflow), whereas, in the top-down feedback flow, the score isnot averaged, as there is only one rater (i.e., supervisor).Thus, in the numerical evaluations, raters in the bottom-upand lateral feedback flows are anonymous, whereas thetop-down feedback flow reveals the identity of the feed-back sender. However, the written feedback is done non-anonymously—the company has all the feedback sendersspecify their names so that the feedback recipients canreceive further feedback after the quarterly evaluations.Note that it is the top-down evaluation that determines thefinal grade of each employee at the end of the year; thebottom-up and lateral feedback are for informative anddevelopmental purposes. This setting is in linewith that ofthe majority of organizations in which supervisors controlemployee performance evaluation and other importantorganizational resources, as we theorized earlier.


study, we randomly selected one. When a teamleader and team members from the same team vol-unteered, we randomly chose one, if possible, butwere forced to preferentially choose supervisors insome cases in order to ensure equal distribution ofthe number of participants across conditions (amongthe volunteers, there were obviously more teammembers than team leaders, which gave us a largeravailable pool of the former than the latter). For thisreason, this study is a quasi-field experiment inwhich we quasi-randomly, rather than fully ran-domly, invited participants to the three conditions.

Following completion of the recruitment process,researchers visited the company within a week ofemployees receiving their March and June quarterlyevaluations to measure negative feedback, task pro-cesses, andmeta-processes. Twomonths after each ofour visits (i.e., in May and August, respectively), wemeasured creativity froma focal employee’s superior.Thus, thedatacollectionoccurredat fourdistinct timepoints, but the data were collapsed to create twowaves of time-lagged data. Time 1 represented thepoints at which negative feedback, task processes,and meta-processes were measured, and Time 2 (twomonths after Time 1) represented the point at whichsuperiors evaluated the focal employees’ creativ-ity. The final data encompassed 225 employees (re-sponse rate5 54.61%;NBottom-up negative feedback 5 80,NLateral negative feedback5 73,NTop-down negative feedback572). The sample contained 104 females (45.34%) and121 males (54.66%), with an average age of 32.38years (SD 5 4.65). The average tenure of our samplewas6.37years (SD52.69).Themajorityof employeeshad completed a bachelor’s degree (87.6%) and12.4% had completed graduate degrees.

Manipulation and Measures

We created Korean versions of the surveys by fol-lowing the survey translation procedures recom-mended byBrislin (1990). Allmeasures used a Likertscale, with responses ranging from “1” (stronglydisagree) to “7” (strongly agree).

The direction of feedback flow. As previouslymentioned, we manipulated the direction of feedbackflow by collecting data from employees in the differentpositions (either team members or team leaders) in thedifferent quarterly evaluations (in March and June).Thisallowedus tocreate threeconditions: the top-downfeedback flow condition (feedback recipient 5 teammembers, feedback sender 5 team leaders, time offeedback 5 June), the lateral feedback flow condition(feedback recipient5 teammembers, feedback sender5

other teammembers, time of feedback5March), andthe bottom-up feedback flow condition (feedbackrecipient 5 team leaders, feedback sender 5 teammembers, time of feedback 5 March).

Because the conditions are categoricalwithout anyrank order, we represented the three conditions us-ing two dummy variables for our analyses (Hayes,2013). We chose the bottom-up feedback flow con-dition as the reference group, because it is the onlyconditionwhere we expected a positive relationshipbetween negative feedback and recipient creativity;in the other two conditions, we expected negativerelationships between negative feedback and recip-ient creativity. By choosing the bottom-up feedbackflow condition as the reference group, we were ableto investigate differences between the bottom-upfeedback flow condition and the top-down feedbackflow condition using our first dummy variable (D1),and differences between the bottom-up feedbackflow condition and the lateral feedback flow condi-tion using our second dummy variable (D2). Specif-ically, we followed “indicator” dummy coding(Hayes, 2013); the dummy coding in each conditionwas as follows: in the bottom-up feedback flowcondition, D1 5 0 and D2 5 0; in the top-downfeedback flow condition, D1 5 1 and D2 5 0; in thelateral feedback flow condition, D1 5 0 and D2 5 1.

Negative feedback. Participants evaluated theextent to which their quarterly evaluation was nega-tive. To measure negative feedback, we modified the7-item feedback valence scale developed by Georgeand Zhou (2001). The original scale captures bothpositive and negative feedback on the same contin-uum.3 Among the seven items in the original scale,four of them clearly reflect negative feedback and theother three measure positive feedback. We modifiedthe latter three items so that they rated negativefeedback. The items are reported in Appendix A.

Positive feedback. Although the primary pur-poses of our research were to investigate the rela-tionship between negative feedback and creativity

3 The design of the original George and Zhou (2001)feedback scale assumes that negative and positive feed-back are on the same continuum. This assumption can beproblematic, though, as it is hard to interpret the scalemidpoint. The scale midpoint implies two different situa-tions simultaneously: (1) a feedback source providing ahigh volume of both positive andnegative feedback and (2)a feedback source that provides minimal feedback of anytype, positive or negative. We discuss the theoretical andmethodological aspects of this issue in the General Dis-cussion section.

2020 593Kim and Kim

and to resolve the inconsistency in this relationship,we also measured positive feedback, for three rea-sons. First, we tested whether employees differenti-ate negative feedback from positive feedback. Wetested this through our confirmatory factor analysis(detailed below). Second, we examined whether theinfluence of negative feedback is distinctive fromthat of positive feedback. Lastly, we controlled for itwhen examining the relationship between negativefeedback and creativity. The participants rated thedegree to which their quarterly evaluation was pos-itive using another modified version of George andZhou’s (2001) feedback valance scale, with the itemsrephrased to inquire about positive feedback. Wereport all the items in Appendix A.

Task processes and meta-processes. We modi-fied existing scales to measure task processes andmeta-processes, based on the conceptualizations ofboth processes delineated in feedback interventiontheory (Kluger & DeNisi, 1996). To measure taskprocesses (a 5 .93), we created a 4-item scale bymodifying the scales for challenge construal (Elliot &Reis, 2003; McGregor & Elliot, 2002) and primaryappraisal (Tomaka, Palacios, Schneider, Colotla,Concha, & Herrald, 1999). To measure meta-processes(a5 .93), we created a 5-item scale by modifying thescales for threat construal (Elliot & Reis, 2003;McGregor & Elliot, 2002) and secondary appraisal(Tomaka et al., 1999). All the items are reported inAppendix A.

Creativity. Superiors rated the focal employee’screativity using Zhou and George’s (2001) 13-itemscale of creative performance. Specifically, divisionheads evaluated the creativity of team leaders in thebottom-up feedback flow condition, team leadersevaluated the creativity of the focal teammembers inthe lateral feedback flow condition, and deputy teamleaders evaluated the creativity of focal team mem-bers in the top-down feedback flow condition. Thereason we collected creativity evaluations from thedeputy team leaders in the top-down feedback flowcondition, rather than collecting it from the primaryteam leaders, was to avoid artificially inflated cor-relations between negative feedback and creativity.In the top-down feedback flow, team members re-ceived feedback from their team leader. Collectingcreativity ratings from those same team leaders couldhave introduced same-source biases into the data inthis condition. A sample itemwas “This employee isa good source of creative ideas” (a 5 .96).

Control variables. We controlled for five demo-graphic variables: age, gender, education, tenurewith the organization, and tenure with feedback

sender, because there could be individual differ-ences when recipients interpret feedback (Kluger &DeNisi, 1996). Age, tenure with organization, andtenurewith feedback senderweremeasured in years.Gender was measured as a dichotomous variable:“0” for female and “1” formale. Educationwas codedin the following manner: “1” for a high school grad-uate, “2” for a graduate from a two-year communitycollege program, “3” for a person with a degree froma four-year university program, and “4” for partici-pant’s with amaster’s degree or PhD. In addition, wecontrolled for creative role identity (a 5 .93) andcreative self-efficacy (a 5 .89). It is possible that re-cipients who believe that they are creative or effica-cious on creativity tasks aremore likely to seek waysof improving their creativity (Shalley, Zhou, &Oldham, 2004), and thus they may be more recep-tive to negative feedback. We also included positivefeedback to investigate the unique influences ofnegative feedback on creativity.4

Confirmatory Factor Analysis

We conducted confirmatory factor analysis to ex-amine the psychometric validity of our measuresusing a variety of indicators ofmodel fit: a chi-squarestatistic (x2), the Tucker–Lewis index (TLI), thestandardized root mean squared residual (SRMR),the comparative fit index (CFI), and the root meansquared error of approximation (RMSEA). Thisanalysis included the negative feedback, task pro-cesses, meta-processes, and creativity items. Fur-thermore, we added positive feedback to investigatewhether participants differentiated negative feed-back from positive feedback. According to the stan-dards recommended by Hu and Bentler (1998), ourconfirmatory factor analysis indicated a good fit forthe five-factor model.5 The x2 value was 890.794(df 5 550, p , .001), TLI was .957, SRMR was .043,

4 Note that the results remained the samewithorwithoutthe control variables.

5 Although the chi-square test showed a poor fit, all ofother indicators showed good fits. Several scholars havenoted that chi-square indices may mislead the interpreta-tion of model fit because it is too sensitive to sample sizeand the violation of an assumption underlying the test(e.g., multivariate normality of variables). “Hence, usingchi-square to test the hypothesis that the population co-variance matrix matches the model-implied covariancematrix is too strong to be realistic” (Hu & Bentler, 1998:425). Given that the other indices, except for those of thechi-square test, were above the traditional cut-off values,we conclude that our five-factor model has a good fit.


CFI was .960, and the RMSEA was .050. The aver-age levels of the standardized loadings on the fiveconstructs were as follows: creativity 5 .83, negativefeedback5 .86,meta-processes5 .85, taskprocesses5.88, and positive feedback5 .90.

In addition, we compared our five-factor modelwith potential alternative models. The alternativemodel showing the highest fit was a four-factormodel, which combined task processes and meta-processes, x2(554) 5 1549.98, p , .001; TLI 5 .875;SRMR5 .073; CFI5 .884; RMSEA5 .085. However,a comparison analysis of our original five-factormodel and the four-factor alternative model showedthat our model has a significantly better fit than thealternative model, Dx2(4) 5 659.19, p , .001. Theseresults show strong evidence that our variables werevalid, and that employees differentiated all of theimportant variables from one another. Importantly,employees differently perceived negative and posi-tive feedback according to the confirmatory factoranalysis, which suggests that each captures differentaspect of feedback.

Results of Study 1

The means, standard deviations, and correlationsfor all variables in Study 1 are presented in Table 2.To test our hypotheses, we used multiple hierarchi-cal regression, combined with conditional indirecteffect analysis—moderated mediation analysis—using the PROCESS macro developed by Hayes(2013).

Hypothesis 1 predicted that, in the bottom-upfeedback flow condition, the relationship betweennegative feedback and recipient creativity would bepositive, whereas, in the top-down and lateral feed-back flow conditions, the relationship would benegative. The results revealed that the interactionsbetween negative feedback and the two dummyvariables were significant. Specifically, Table 3shows significant interactions between negativefeedback and D1 (bottom up 5 0, top down 5 1;b 5 21.03, SE 5 0.15, t 5 26.94, p , .001), andbetween negative feedback and D2 (bottom up 5 0,lateral5 1; b520.97,SE5 0.15, t526.61,p, .001)in predicting creativity. The result of the former in-teraction indicates that there is a significant slopedifference between the bottom-up and top-downfeedback flow conditions. The result of the latterinteraction indicates that there is a significant slopedifference between the bottom-up and the lateralfeedback flow conditions. We further investigatedthese interactions by analyzing the simple slopes at

each value (0 or 1) of each moderator (D1 and D2).The results showed that, in the bottom-up feedbackflowcondition (from followers to supervisors; D150and D2 5 0) the relationship between negativefeedback and recipient creativity was significantlypositive (b 5 0.48, SE 5 0.11, t 5 4.34, p , .001,LLCI 5 0.264, ULCI 5 0.705). In the top-down feed-back flow condition (from supervisors to followers;D1 5 1 and D2 5 0), the relationship was signifi-cantly negative (b520.54, SE5 0.10, t525.39, p,.001; LLCI 5 20.739, ULCI 5 20.343). In the lateralfeedback flow condition (from peers to peers; D15 0and D2 5 1) the relationship was significantly neg-ative (b 5 20.49, SE 5 0.10, t 5 24.94, p , .001,LLCI 5 20.681, ULCI 5 20.293). Figure 2 depictsthese relationships. Hypothesis 1 was supported.

Hypothesis 2 posited that the moderated rela-tionships between negative feedback, the directionof feedback flow, and recipient creativity would bemediated by task processes and meta-processes.Before testing the moderated mediation hypothesis,we conducted two separate analyses to examine theinteraction of negative feedback and the direction offeedback flow in predicting task processes andmeta-processes. With regard to task processes, we foundsignificant interactions of negative feedback and D1(b 5 20.94, SE 5 0.18, t 5 25.37, p , .001) and D2(b5 21.26, SE 5 0.18, t 5 27.22, p , .001). Table 3summarizes the full results. The simple slope testsshowed that the bottom-up negative feedback en-abled supervisors to focus on task processes (b 50.52, SE 5 0.13, t 5 3.89, p , .001, LLCI 5 0.255,ULCI 5 0.778). However, the top-down negativefeedback (b520.43, SE5 0.12, t523.58, p, .001,LLCI520.662, ULCI520.192) and lateral negativefeedback (b520.74, SE5 0.12, t526.36, p, .001,LLCI 5 20.975, ULCI 5 20.513) significantly de-creased employees’ focus on task processes. Figure 3depicts these relationships.

Regarding meta-processes, the interactions thatnegative feedback had with D1 (b5 0.36, SE5 0.17,t 5 2.17, p 5 .031) and D2 (b 5 0.57, SE 5 0.17, t 53.47, p 5 .001) were both significant (see Table 3).The simple slope tests revealed that the relationshipbetween negative feedback and meta-processes wasnot significant in the bottom-up feedback flow con-dition (b 5 0.07, SE 5 0.13, t 5 0.59, p 5 .553,LLCI 5 20.173, ULCI 5 0.322), but the relationshipwas significantly positive in the top-down (b5 0.43,SE 5 0.11, t 5 3.86, p , .001, LLCI 5 0.212, ULCI 50.657) and lateral (b5 0.65, SE5 0.11, t5 5.85, p,.001, LLCI 5 0.429, ULCI 5 0.865) feedback flowconditions. Figure 4 illustrates these relationships.

2020 595Kim and Kim

TABLE2

Mea

ns,StandardDev

iation

s,an

dCorrelation

sof

Variables

inStudy1

Variables

MSD

12

34

56

78

910

1112

13

1D1(bottom

up5

0;top

dow

n5

1)

0.32

0.47

2D2(bottom

up5

0;lateral5

1)

0.32

0.47

2.48*

*

3Neg

ative

Fee

dba

ck3.98

1.09

2.02

2.19*

*

4TaskProcesses

4.86

1.35

.02

2.18*

*2.12†

5Meta-Processes

3.35

1.16

.10

.09

.29*

*2.44*

*6

Creativity

5.14

1.07

2.12†

2.05

2.17*

.37*

*2.46*

*7

Gen

der

0.54

0.50

2.01

2.04

2.03

.05

2.11

2.01

8Age

32.38

4.65

2.41*

*2.04

.13†

.03

2.07

.02

.02

9Education

3.08

0.51

.05

2.03

2.05

.02

2.08

2.05

.12†

.04

10Ten

ure

with

Com

pan

y6.37

2.69

2.25*

*.00

.02

.07

2.08

2.05

.03

.33*

*2.04

11Ten

ure

with

Tea

m2.15

0.84

2.04

.07

2.02

2.07

.06

.00

2.20*

*.08

.03

.04

12CreativeRole

Iden

tity

5.44

1.08

.02

2.14*

.07

.27*

*2.10

.16*

.14*

2.01

2.01

2.09

2.14*

13Creative

Self-efficacy

5.28

1.15

2.04

2.09

.09

.37*

*2.18*

*.14*

.13*

.00

.03

2.05

2.19*

*.71*

*

14Positive

Fee

dba

ck4.46

1.21

2.04

.09

2.26*

*2.10

.05

2.03

2.03

2.02

.04

2.05

.07

.04

.05

Note:

n5

225.

†p,

.10

*p,

.05

**p,

.01(allteststw

o-tailed

)


Lastly, we tested the full moderated mediationmodel using a conditional indirect effect analysiswith 10,000 bias-corrected bootstrap resampling(Hayes, 2013). The results showed that, in the bottom-up feedback flow condition, the positive relationshipbetween negative feedback and creativity was medi-ated by task processes (estimated size of the indirecteffect5 0.08,SE5 0.04, LLCI5 0.011,ULCI5 0.181),but not by meta-processes (estimated size of the in-direct effect 5 20.03, SE 5 0.04, LLCI 5 20.114,ULCI 5 0.046). In the top-down feedback flow con-dition, the negative relationship between negativefeedback and creativity was mediated by bothtask processes (estimated size of the indirecteffect 5 20.07, SE 5 0.04, LLCI 5 20.157,ULCI 5 20.010) and meta-processes (estimated sizeof the indirect effect 5 20.15, SE 5 0.08,LLCI 5 20.356, ULCI 5 20.037). In the lateralfeedback flow condition, the negative relationshipbetween negative feedback and creativity wasalso mediated by both task processes (estimatedsize of the indirect effect 5 20.12, SE 5 0.06,LLCI520.246, ULCI520.021) andmeta-processes

(estimated size of the indirect effect 5 20.23, SE 50.09, LLCI 5 20.433, ULCI 5 20.082). These resultssupport Hypothesis 2.

Discussion of Study 1

Using a quasi-field experimental design, Study 1found that the relationship between negative feed-back and creativity depends on the direction offeedback flow. Bottom-up feedback flow was theonly case in which negative feedback increased re-cipient creativity; top-down and lateral feedbackflows of negative feedback were detrimental to re-cipient creativity. The two mechanisms—task pro-cesses and meta-processes—accounted for theseinteractional relationships. In the bottom-up feed-back flow condition, negative feedback increasedcreativity because recipients attempted to improvethe ways of performing their creativity tasks, andthey did not feel threatened by negative feedback.However, in the top-down and lateral feedback flowconditions, recipients failed to generate diversestrategies to perform their creativity tasks. Instead,

TABLE 3Multiple Hierarchical Regression on Creativity, Task Processes, and Meta-Processes in Study 1

Variables

Creativity Task Processes Meta-Processes

Model 1 Model 2 Model 1 Model 2 Model 1 Model 2

b (SE) b (SE) b (SE) b (SE) b (SE) b (SE)

Gender 20.09 (0.14) 20.08 (0.13) 20.05 (0.17) 20.09 (0.15) 20.11 (0.15) 20.09 (0.15)Age 20.01 (0.02) 20.02 (0.02) 0.00 (0.02) 20.01 (0.02) 20.00 (0.02) 0.00 (0.02)Education 20.11 (0.14) 20.10 (0.12) 0.01 (0.16) 0.04 (0.15) 20.13 (0.14) 20.14 (0.14)Tenure with Company 20.04 (0.03) 20.02 (0.02) 0.04 (0.03) 0.06 (0.03)† 20.01 (0.03) 20.02 (0.03)Tenure with Team 0.04 (0.09) 0.06 (0.08) 0.02 (0.10) 0.02 (0.09) 0.01 (0.09) 0.01 (0.09)Creative Role Identity 0.10 (0.09) 0.08 (0.08) 0.01 (0.11) 20.02 (0.10) 0.08 (0.09) 0.10 (0.09)Creative Self-Efficacy 0.08 (0.09) 0.07 (0.08) 0.44 (0.10)** 0.41 (0.09)** 20.25 (0.09)** 20.24 (0.09)**Positive Feedback 20.09 (0.06) 20.03 (0.05) 20.18 (0.07)* 20.12 (0.06)† 0.14 (0.06)* 0.11 (0.06)†

Negative Feedback 20.24 (0.07)** 0.48 (0.11)** 20.29 (0.08)** 0.52 (0.13)** 0.42 (0.07)** 0.08 (0.13)D1 (bottom up5 0;

top down 5 1)20.57 (0.20)** 3.66 (0.63)** 20.16 (0.23) 3.78 (0.75)** 0.50 (0.20)* 21.01 (0.71)

D2 (bottom up5 0;lateral5 1)

20.43 (0.18)* 3.52 (0.61)** 20.57 (0.21)** 4.49 (0.72)** 0.57 (0.19)** 21.71 (0.68)*

Negative Feedback3 D1 21.03 (0.15)** 20.94 (0.18) ** 0.36 (0.17)*Negative Feedback3 D2 20.97 (0.15)** 21.26 (0.18)** 0.57 (0.17)**F 2.61** 7.41** 5.52** 10.09** 5.24** 5.58**DF 2.61** 29.96** 5.52** 27.65** 5.24** 6.08**R2 .12 .31 .22 .38 .21 .26DR2 .12 .19 .22 .16 .21 .05

Note: n 5 225.†p , .10*p , .05

**p , .01 (all tests two-tailed)

2020 597Kim and Kim

they felt threatened by negative feedback, which inturn decreased their creativity. Although our primaryfocus was to investigate the relationship betweennegative feedback and creativity, we also examinedhow positive feedback relates to creativity. The re-sult shown in Table 3 reveals that positive feedbackhad a nonsignificant relationship with creativity(b 5 20.04, SE 5 0.06, t 5 20.63, p 5 .528). Wediscuss this relationship in the General Discussion.

Although the results of Study 1 supported ourhypotheses, some limitations should be noted. First,Study 1 quasi-randomly assigned employees to thethree feedback flow conditions using the naturalfield setting, and we did not manipulate negativefeedback. This raises a question of the potential in-fluences that spurious third variables might have onour findings. Second, we measured the recipients’subjective impression of feedback. In the quarterly

FIGURE 3Regression Slopes for the Interaction of Negative Feedback and the Direction of Feedback Flow on Task

Processes (Study 1)

1

2

3

4

5

6

7

LowNegative Feedback

HighNegative Feedback

Tas

k P

roce

sses

Bottom-up Flow

Top-down Flow

Lateral Flow

FIGURE 2Regression Slopes for the Interaction of Negative Feedback and the Direction of Feedback Flow on Creativity

(Study 1)

1

2

3

4

5

6

7



Em

plo

yee

Cre

ativ

ity

Bottom-up Flow

Top-down Flow

Lateral Flow


performance appraisal, employees received bothnumerical and written feedback. Rather than usingthe numerical and written feedback directly, weasked the employees to rate their subjective evalua-tion of the performance appraisal with an assump-tion that the same feedback could be perceiveddifferently by employees. For example, two em-ployees who receive 60% in the numerical appraisalmay interpret the same number differently; one mayinterpret it as negative feedback, while another mayperceive it as neutral feedback. In addition, meth-odologically speaking, it was difficult to quantify thenegativity of the written feedback objectively. Third,there could be a confounding effect of feedbackrecipient positions. We collected data from super-visors in thebottom-up feedback flow, but, in the top-down and lateral feedback flows, our data collectionwas done with team members. Ideally, the datashould have been collected from employees at thesame position across the three conditions (e.g., mid-dle managers who receive feedback from their fol-lowers, supervisors, and peers). This limitation is lesslikely tohurt the validity of the top-downandbottom-up feedback flows because those two flows presup-pose positional differences between feedback giversand recipients. However, there could be an unex-pected confounding effect of position in the lateralfeedback flow. For example, the underlying psy-chology of middle managers receiving feedback from

peers could be different from that of team membersreceiving feedback from peers.

To address these limitations, we conducted Study2, in which we manipulated both negative feed-back and the direction of feedback flow and ran-domly assigned participants to conditions in awell-controlled laboratory environment. This ex-periment directly manipulates negative feedbackrather than measuring the subjective evaluation ofthe feedback. Additionally, we manipulate the lat-eral feedback flow without disclosing the positionsof feedback giver and recipient.

STUDY 2: LABORATORY EXPERIMENT

Participants and Procedures

Three hundred and fifty-six undergraduate stu-dents at a large North American university (217males, 136 females, three participants did not reporttheir gender; MAge 5 19.80, SDAge 5 1.51, four par-ticipants did not report their age) participated in thisstudy for one credit point. The sample was 57.02%East Asian, 23.31% Caucasian, 8.99% South Asian,3.37% African/African Canadian, 0.84% Hispanic,0.28% Native American/Canadian, and 7.58%Other. Two independent variables (feedback and thedirection of feedback flow) weremanipulated in thisexperiment based on a 2 (feedback: negative vs.neutral) by 3 (the direction of feedback flow: bottom

FIGURE 4Regression Slopes for the Interaction of Negative Feedback and the Direction of Feedback Flow on Meta-

Processes (Study 1)

1

2

3

4

5

6

7



Met

a-P

roce

sses

Bottom-up Flow

Top-down Flow

Lateral Flow

2020 599Kim and Kim

up vs. top down vs. lateral) experimental design. Werandomly assigned the participants to one of the sixconditions.

Upon arrival, the participants were led to a largeroomwhere they completed a short survey regardingtheir demographic information.After completing thesurvey, the participants were randomly assigned toone of three roles (i.e., supervisor, follower, or peer)and led to four different rooms. Participants playingthe same role were grouped together and shared thesame room, except for those playing the peer role.The peers were divided into two groups, and eachgroup used one room. Then, the participants wereinformed that they would complete a two-roundstudy that would involve exchanging feedback withanother participant in a different room through anonline platform. Both rounds involved an in-baskettask, which is an idea generation task that requirescreative solutions (Shalley, 1991). In this task, par-ticipants assume the role of an employee in a com-pany and are asked to generate creative ideasregarding 22 organizational issues. We used one ofthe 22 issues in the first round and three issues in thesecond round, as explained below. The participantswere informed that they would be paired with apartner in a different room. During the first round,the participants individually generated one idea re-garding an issue, and then they were paired withanother participant. After the pairing, the twopartiesexchanged feedback regarding the counterpart’sidea, including a numerical rating and written feed-back in which they could provide additional infor-mation to qualify their rating. Subsequently, theparticipants responded to the measures of task pro-cesses and meta-processes. Then, the participantsproceeded to the second round. The participantsagain individually generated ideas regarding threeadditional issues in the in-basket tasks,which servedas our dependent variable, exchanged their ideas,and evaluated each other’s ideas. In reality, the par-ticipants did not have a real partner, and all feedbackprovided was prepared by the experimenter, whichallowed us to manipulate the negative feedback.

Negative Feedback Manipulation

After the participants completed the first idea-generation task, they sent their idea to their partnerand received their partner’s answer to the samequestion. The idea thatwas supposedly generated bytheir partner was in fact prepared by the experi-menter and was identical for all participants. Theparticipants were asked to offer feedback regarding

the level of creativity of their partner’s idea, includ-ing a percentile score and written feedback. Whilethe participants completed this task, they believedthat their partner also evaluated their idea and pro-vided feedback. After completing their feedback, theparticipants received their partner’s feedback re-garding their ownwork. In reality, this feedback wasprepared by the experimenter and included thenegative feedback manipulation. The participantsreceived feedback in the same format they had usedto provide feedback to their partner: they received apercentile score and additionalwritten feedback.Wemanipulated the negative feedback using the per-centile score. In the negative feedback condition, weinformed the participants that their creativity was inthe bottom 20th percentile, and, in the neutral feed-back condition, the participants were informed thattheir creativity was average, as they were in the40th;60th percentile. All participants received thesame written feedback, which contained advice re-garding generating creative ideas. The advice in-cluded useful tips on how to increase cognitiveflexibility and cognitive persistence in creativitytasks, which were derived from findings reported inthe creativity literature (Nijstad, DeDreu, Rietzschel,& Baas, 2010; Shalley, 1991; see Appendix A). Thesetips would likely contribute to the generation ofcreative ideas if the participants were open to usingthem. To ensure that the manipulation worked asintended, we asked the participants to report howthey felt about the feedback they received on a scalefrom “1” (extremely negative) to “7” (extremelypositive).

Direction of Feedback Flow Manipulation

The “direction of feedback flow” refers to thetransfer of feedback from sender(s) to recipient(s) inwhich the two parties have different or the same or-ganizational ranks. As observed above, the bottom-up and top-down feedback flows are characterizedby power asymmetry between the two parties,whereas the lateral feedback flow involves a nonhi-erarchical and competitive relationship between thetwo parties who strive to attain limited resources.Reflecting upon these characteristics, we manipu-lated the direction of feedback flow in two ways: (1)by assigning the participants to the role of supervi-sor, follower, or peer, and (2) by offering differentlevels of resource controls (Anderson & Berdahl,2002; Jordan et al., 2011; Tost, Gino, &Larrick, 2013).The role assignment manipulation has been popularamong previous researchers because evidence has


shown that participants primed with a certain rolepossess the mindset corresponding to the role andbehave accordingly (Anderson & Berdahl, 2002;Galinsky, Gruenfeld, & Magee, 2003). The role as-signment was performed at the beginning of the ex-periment by randomly assigning participants to therole of supervisor, follower, or peer. In the bottom-upfeedback flow condition, the participants wereassigned to the role of supervisor and paired with afollower. In the top-down feedback flow condition,the participants were assigned to the role of followerand paired with a supervisor. In the lateral feedbackflow condition, both the participants and partnerswere assigned to the role of peer without indicatingany information regarding the positions of bothparties.

To strengthen our manipulation, we also manip-ulated the different levels of resource controls usinga lottery system. This manipulation intended tocreate the approximate experience of having socialpower in the real world and a sense of competition toattain a limited resource (e.g., Anderson & Berdahl,2002; Jordan et al., 2011; Tost et al., 2013). The ex-perimenter provided instructions regarding a lotteryas follows: the participants would be entered into alottery for nine $100 Amazon gift cards (three perrole) at the end of the semester and compete for thelottery only with those who played the same role. Inthe bottom-up feedback flow condition, we gave thesupervisors unique control over the desired reward(i.e., chance of winning the lottery). The participantsin this condition learned that their partner’s chanceof winning the lottery would depend on how theyevaluated their partner’s creativity in the secondround. In other words, by considering the follower’s(or partner’s) levels of creativity in the idea-generation task, the supervisor (or the participant)could decide how many lottery tickets the followerwould receive (from 1 to 10 tickets), and each addi-tional ticket increased the chance of winning thelottery. Furthermore, the participantswere informedthat their partner’s evaluation would not influencetheir chance of winning the lottery. Instead, at theend of the semester, the researchers would evaluatethe creativity of their ideas and determine the num-ber of tickets they would receive. In the top-downfeedback flow condition, we provided the same in-structions, but, this time, the participant–partnerrelationship was the opposite, as follows: the par-ticipants played the role of follower, and their part-ner played the role of supervisor. Thus, the partnercould determine the number of tickets that the par-ticipantswould receive,while their evaluation of the

partner’s creativity could not influence the partner’schance of winning the lottery. In the lateral feedbackflow condition, the instructions stated that the part-nerswere at the same organizational rank, indicatingthat they were peers. Therefore, the partner’s evalu-ation did not influence the participant’s chance ofwinning the lottery and vice versa. The participantswere also informed that, at the end of the semester,the researchers would evaluate their ideas and de-termine the number of tickets they would receive forthe lottery.

To ensure that the participants understood themanipulation of the direction of feedback flow, weemployed two types of manipulation checks. First,we asked the participants to report their and theirpartners’ roles. Second, we checked the level of au-thority and social power each participant felt overtheir partner by asking the following two questions:(1) “Do you have authority and social power to con-trol your partner’s chance of winning the lottery?”and (2) “Towhat extent do you feel social power overyour partner?” The participants responded on aLikert-type scale ranging from “1” (none) to “7” (alot). The scores of the two items of the second ma-nipulation check scale were then averaged. Finally,we used the same “indicator” coding scheme as wasused in Study 1 to analyze the direction of feedbackflow: the bottom-up feedback flow was coded asD15 0 and D25 0; the top-down feedback flow wascoded as D15 1 and D25 0; and the lateral feedbackflow was coded as D1 5 0 and D2 5 1.

Task processes and meta-processes. Task pro-cesses (a 5 .96) and meta-processes (a 5 .96) weremeasured by the same scales that were used inStudy 1. The participants rated the two scales be-tween the first and second rounds of their ideageneration and evaluation tasks. All items are listedinAppendixA. Theparticipants reported the extentto which they agreed with the items on a 7-pointscale ranging from “1” (not at all) to “7” (verymuch).

Creativity. The three ideas generated in the sec-ond round of the idea generation and evaluationtasks served as our dependent variable: recipientcreativity. Two independent judges were trained toevaluate the extent to which the three ideas werecreative using a scale ranging from “1” (not at allcreative) to “7” (very creative). The judges showedsubstantial agreement in the evaluation of the levelsof creativity of the ideas: ICC 1 5 .80 (p , .001) andICC 2 5 .89, (p , .001). Thus, the final creativityscores were obtained by averaging the scores of thetwo judges on each idea and then averaging thescores of the three ideas.

2020 601Kim and Kim

Results of Study 2

Feedback manipulation check. A 2 (feedback:negative vs. neutral) by 3 (direction of feedback flow:bottom up vs. top down vs. lateral) analysis of vari-ance (ANOVA) with the feedback manipulationcheck item as the dependent variable showed a sig-nificant main effect of the feedback condition. Theparticipants who received negative feedback re-ported that the feedback they received was morenegative (M 5 3.34, SD 5 1.85) than those who re-ceived the neutral feedback (M 5 4.20, SD 5 1.38),F(1, 350) 5 27.71, p , .001, hp

2 5 .07. Thus, ourfeedback manipulation was successful.

Direction of feedback flow manipulation check.We checked this manipulation in two ways, as pre-viously described. First, we checked whether theparticipants accurately identified their and theirpartners’ roles. Three hundred and forty-nine of the356 participants (98.03%) accurately identified theroles. Second, we conducted a 2 (feedback: negativevs. neutral) by 3 (direction of feedback flow: bottomup vs. top down vs. lateral) ANOVA with the ma-nipulation check scale as the dependent variable.The results showed that the participants in thebottom-up feedback flow condition reported ithigher (M5 6.32, SD5 1.21) than those in the lateralfeedback flow condition (M 5 2.01, SD 5 2.02) andthose in the top-down feedback flow condition (M51.88, SD5 1.55), F(2, 350)5 298.93, p, .001, hp

2 5.63. The participants in the lateral feedback flowcondition and top-down feedback flowconditiondidnot differ, t(231) 5 0.54, p 5 .587. Thus, our manip-ulation was successful.

Hypothesis testing. The means, standard devia-tions, and correlations of all variables in Study 2 arepresented in Table 4. To test our first hypothesis, wecreated two dummy variables for the direction offeedback flow, following the same procedure used inStudy 1, and performed a regression analysis.

Hypothesis 1 predicted that, in the bottom-upfeedback flow condition, the effect of negative feed-back on recipient creativity would be positive,whereas, in the top-down and lateral feedback flowconditions, the effect would be negative. The resultsshowed that the interaction between the feedbackcondition (neutral 5 0, negative 5 1) and D1 wassignificant (bottom up5 0, top down5 1; b521.55,SE 5 0.30, t 5 25.15, p , .001; see Table 5), as wasthe interaction between the feedback condition andD2 (bottom up5 0, lateral5 1; b522.08, SE5 0.32,t 5 26.48, p , .001; see Table 5). The former inter-action indicates that there is a significant slope dif-ference between the bottom-up and top-downfeedback flow conditions; the latter interactionshows a significant slope difference between thebottom-up and lateral feedback flow conditions.Furthermore, we conducted simple slope tests ateach value (0 or 1) of eachmoderator (D1 andD2) andfound that negative feedback (vs. neutral feedback)led to higher levels of recipient creativity in thebottom-up feedback flow condition (MNeutral 5 3.56,SDNeutral 5 1.07; MNegative 5 4.52, SDNegative 5 1.47;b5 0.96, SE5 0.22, t5 4.45, p, .001, LLCI5 0.536,ULCI 5 1.385). However, negative feedback led tolower levels of creativity in the top-down feedbackflow condition (MNeutral 5 3.77, SDNeutral 5 1.11;MNegative 5 3.19, SDNegative 5 1.04; b 5 20.59, SE 50.21, t 5 22.81, p 5 .005, LLCI 5 20.995,ULCI520.176) and lateral feedback flow condition(MNeutral 5 4.34, SDNeutral 5 1.40; MNegative 5 3.22,SDNegative5 .94; b521.12, SE5 0.24, t524.71, p,.001, LLCI 5 21.589, ULCI 5 20.652). These rela-tionships are depicted in Figure 5. Thus, Hypothesis1 is supported.

Hypothesis 2 stated that the interactional effectsbetweennegative feedback anddirection of feedbackflow on recipient creativity are mediated by taskprocesses and meta-processes. Before we tested the

TABLE 4Means, Standard Deviations, and Correlations of Variables in Study 2

Variables M SD 1 2 3 4 5

1 D1 (bottom up5 0; top down5 1) 0.37 0.482 D2 (bottom up5 0; lateral5 1) 0.28 0.45 2.48**3 Negative Feedback 0.50 0.50 2.03 2.284 Task Processes 4.88 1.65 2.05 2.11* 2.045 Meta-Processes 3.61 1.82 .03 .02 .16** 2.32**6 Creativity 3.79 1.29 2.17** .01 2.07 .53** 2.43**

Notes: n 5 356. Negative Feedback (neutral feedback 5 0, negative feedback5 1).*p , .05



full moderated mediation model, we tested the in-teractions between negative feedback and directionof feedback flow in predicting task processes andmeta-processes. In predicting task processes, theinteractions between the feedback condition and D1(b 5 21.46, SE 5 0.40, t 5 23.64, p , .001) andbetween the feedback condition and D2 (b 5 21.61,SE5 0.43, t523.75, p, .001) were significant (seeTable 5). Specifically, the results of the simple slopetests showed that, in the bottom-up feedback flow

condition, negative feedback led to greater attentionto task processes (MNeutral 5 4.79, SDNeutral 5 1.50;MNegative 5 5.61, SDNegative 5 1.46; b 5 0.82, SE 50.29, t5 2.86, p5 .005, LLCI5 0.257, ULCI5 1.392).However, negative feedback reduced the partici-pants’ attention to task processes in both the top-down feedback flow condition (MNeutral 5 5.08,SDNeutral 5 1.27; MNegative 5 4.45, SDNegative 5 1.71;b 5 20.63, SE 5 0.28, t 5 22.28, p 5 .023,LLCI521.182,ULCI520.088) and lateral feedback

TABLE 5Multiple Hierarchical Regression on Creativity, Task Processes, and Meta-Processes in Study 2

Variables

Creativity Task Processes Meta-Processes

Model 1 Model 2 Model 1 Model 2 Model 1 Model 2

b (SE) b (SE) b (SE) b (SE) b (SE) b (SE)

Negative Feedback 20.20 (0.14) 0.96 (0.22)** 20.17 (0.17) 0.82 (0.29)** 0.60 (0.19)** 20.04 (0.32)D1 (bottom up5 0; top down5 1) 20.60 (0.16)** 0.21 (0.21) 20.46 (0.21)* 0.29 (0.29) 0.22 (0.23) 20.25 (0.32)D2 (bottom up5 0; lateral 5 1) 20.28 (0.17)† 0.78 (0.23)** 20.64 (0.22)** 0.18 (0.30) 0.23 (0.24) 20.31 (0.34)Negative Feedback3 D1 21.55 (0.30)** 21.46 (0.40)** 0.90 (0.45)*Negative Feedback3 D2 22.08 (0.32)** 21.61 (0.43)** 1.06 (0.48)*F 5.31** 13.04** 3.40* 5.82** 3.60* 3.38**DF 5.31** 23.62** 3.40* 9.20** 3.60* 3.00†

R2 .04 .16 .03 .08 .03 .05DR2 .04 .11 .03 .05 .03 .02

Notes: n 5 356. Negative Feedback (0 5 neutral feedback; 15 negative feedback).†p , .10*p , .05


FIGURE 5Regression Slopes for the Interaction of Negative Feedback and the Direction of Feedback Flow on Recipient

Creativity (Study 2)

1

2

3

4

5

6

7

Neutral Feedback Negative Feedback

Rec

ipie

nt

Cre

ativ

ity

Bottom-up Flow

Top-down Flow

Lateral Flow

2020 603Kim and Kim

flow condition (MNeutral 5 4.97, SDNeutral 5 1.89;MNegative 5 4.19, SDNegative 5 1.79; b 5 20.78,SE 5 0.32, t 5 22.47, p 5 .014, LLCI 5 21.410,ULCI 5 20.159). These effects are depicted inFigure 6.

In predicting meta-processes, the interactionsbetween the feedback condition and D1 (b 5 0.90,SE 5 0.45, t 5 2.01, p 5 .045) and between thefeedback condition and D2 (b5 1.06, SE5 0.48, t52.20, p 5 .029) were significant (see Table 5).6 Thesimple slope tests showed that negative feedbackdid not influence attention to meta-processes in thebottom-up feedback flow condition (MNeutral 5 3.50,SDNeutral 5 1.61; MNegative 5 3.47, SDNegative 5 1.53;b 5 20.04, SE 5 0.32, t 5 20.11, p 5 .914,LLCI 5 20.672, ULCI 5 0.601). However, negativefeedback directed the participants’ attention to meta-processes in the top-down feedback flow condition(MNeutral 5 3.26, SDNeutral 5 1.49; MNegative 5 4.12,SDNegative 5 1.80; b 5 0.87, SE 5 0.31, t 5 2.79, p 5.006, LLCI 5 0.255, ULCI 5 1.482) and lateral feed-back flow condition (MNeutral5 3.19, SDNeutral5 1.77;MNegative54.21,SDNegative52.56;b51.02,SE50.36,t5 2.87,p5 .004, LLCI5 0.322,ULCI5 1.725). Theseeffects are depicted in Figure 7.

Finally, to test the full moderated mediationmodel, we used Hayes’s (2013) PROCESS macro

with 10,000 bias-corrected bootstrapping resamples.In the bottom-up feedback flowcondition, consistentwith the results of Study 1, the positive effect ofnegative feedback on recipient creativity was medi-ated by task processes (estimated size of the indirecteffect5 .28, SE5 0.10, LLCI5 0.093, ULCI5 0.484)but not by meta-processes (estimated size of theindirect effect 5 0.01, SE 5 0.06, LLCI 5 20.104,ULCI 5 0.128). In the top-down feedback flow con-dition, the negative effect of negative feedback oncreativity was mediated by both task processes (esti-mated size of the indirect effect 5 20.22, SE 5 0.10,LLCI5 20.421, ULCI5 20.039) and meta-processes(estimated size of the indirect effect 5 20.18, SE 50.07, LLCI 5 20.336, ULCI 5 20.066). In the lateralfeedback flow condition, the negative effect of nega-tive feedback on creativity was alsomediated by bothtask processes (estimated size of the indirecteffect 5 20.27, SE 5 0.13, LLCI 5 20.545,ULCI 5 20.026) and meta-processes (estimatedsize of the indirect effect 5 20.21, SE 5 0.10,LLCI520.427, ULCI520.040). Thus, the resultssupported Hypothesis 2.

Discussion of Study 2

In the laboratory experiment, we again found thatthe direction of feedback flow moderates the re-lationship between negative feedback and recip-ient creativity. Specifically, negative feedbackincreased recipient creativity when the feedback

FIGURE 6Regression Slopes for the Interaction of Negative Feedback and the Direction of Feedback Flow on Task

Processes (Study 2)

1

2

3

4

5

6

7


Tas

k P

roce

sses

Bottom-up Flow

Top-down Flow

Lateral Flow

6 In the F test of overall significance, the DF of theseinteractions was marginally significant, p 5 .051 (seeTable 5).


flowed upward, but negative feedback reducedrecipient creativity when the feedback floweddownward or laterally. We also found support forthe hypothesized mediating roles of task pro-cesses and meta-processes, replicating the find-ings from Study 1. Study 2 demonstrates causalityby manipulating both negative feedback and thedirection of feedback flow and randomly assign-ing participants into one of the six conditions in acontrolled laboratory environment.

GENERAL DISCUSSION

The main objective of our research was to re-solve the theoretical and empirical inconsistency inthe relationship between negative feedback and re-cipient creativity. The results of our studies—onequasi-field experiment and one fully randomizedlaboratory experiment—demonstrate that this in-consistency can be resolved by considering the di-rection of feedback flow. We found that bottom-upnegative feedback (which flows from followers to asupervisor) enabled recipients to utilize the feedbackby directing their focus to task processes rather than tometa-processes, and their creativitywas enhanced as aresult. In contrast, top-down negative feedback (whichflows from a supervisor to a follower) and lateral neg-ative feedback (which flows from peers to peers) di-rected recipients’ focus away from task processes andtoward meta-processes, which hindered creativity. By

investigating our theoretical model in the field and inthe laboratory, we were able to establish both thegeneralizability and causality of the hypothesizedrelationships.

Theoretical and Managerial Contributions

Our research contributes to the creativity and feed-back literature by integrating contradictory findingsregarding the relationship between negative feedbackand creativity into a coherent theoretical model.Through our comprehensive review of the literature,we were able to categorize the past findings into twogroups based on their theoretical commonalities. Thefirst group (e.g., Fang et al., 2014; Ford & Gioia,2000; Vuori & Huy, 2015) relied on theories em-phasizing the fact that negative feedback identifiesthe creativity–standard gap, which creates dissat-isfaction with current creativity and encouragesfeedback recipients to close the gap by generatingbetter task strategies for their creativity tasks. Thesecond group (e.g., Van Dijk & Kluger, 2011; Zhou,1998) argued that negative feedback elicits feelingsof insecurity or threat and diverts attention awayfrom creativity tasks, resulting in reduced creativ-ity. By theorizing the two essential mechanisms—task processes and meta-processes—underlyingthe relationship between negative feedback and re-cipient creativity (Kluger & DeNisi, 1996), our re-search provides a comprehensive, parsimonious

FIGURE 7Regression Slopes for the Interaction of Negative Feedback and the Direction of Feedback Flow on Meta-

Processes (Study 2)

1

2

3

4

5

6

7


Met

a-P

roce

sses

Bottom-up Flow

Top-down Flow

Lateral Flow

2020 605Kim and Kim

theoretical framework that incorporates these twocontradictory perspectives.

Our research not only develops a coherent theo-retical model for organizing the conflicting findingsin the literature but also seeks to resolve the incon-sistency by introducing a novel concept: the direc-tion of feedback flow. Past research has examinednegative feedback and recipient creativity with littleunderstanding of how the social contexts surround-ing this relationship play a role. Zhou (2008: 130)identified the limitation of the extant feedback re-search as follows:

How effectively we can use feedback in promotingcreativity depends on the nature and components ofthe feedback itself, on the characteristics of the feed-back recipient, and on the characteristics of thefeedback giver.

Addressing this limitation, our research highlightsthe importance of conceptualizing feedback as a flowthat occurs between two (or more) social actors. Byinvestigating feedback flows, we could identify thefeedback senders and recipients and examine theirsocial hierarchical relationships, which are all es-sential for understanding the effect of negativefeedback on creativity. Using this novel concept, wewere able to investigate the question of how the dy-namics of the hierarchical relationship betweenfeedback senders and recipients might resolve theinconsistent association between negative feedbackand recipient creativity. We believe that feedbackresearchers will benefit by considering the directionof feedback flow in their research.

The investigation of the direction of feedback flowalso extends feedback intervention theory (Kluger &DeNisi, 1996). To the best of our knowledge, empir-ical studies directly testing the two psychologicalprocesses of feedback intervention theory (i.e., taskprocesses and meta-processes) are lacking in thecurrent literature. Because the key value of this the-ory is that it proposes two contradictory processesthat transmit the opposite effects of negative feed-backonemployeebehavior andoutcomes, the lackofempirical tests of the two processes could under-mine its usefulness. Our research, conducted in botha professional organization and a laboratory setting,provides empirical evidence supporting the validityof these two psychological processes. By doing so,we help future researchers confidently utilize feed-back intervention theory. In addition, our researchadds the direction of feedback flow to feedback in-tervention theory as a critical boundary condi-tion within which the theory operates. Despite its

parsimonious theoretical framework organizing thecontradicting influences of negative feedback onorganizational outcomes, this theory does not offerinsights into boundary conditions in which a recip-ient responds to negative feedback by attending toeither task processes or meta-processes. Given theprevalence of negative feedback in organizations(Zenger & Folkman, 2017) and its potential for bothbenefiting and hampering employee outcomes,feedback intervention theory can increase its use-fulness and value by offering insights about whenorganizations can reap benefits from negative feed-back. Our research provides an initial investigationinto an important, albeit largely neglected, boundarycondition: the direction of feedback flow.We believethat our findings could benefit future researcherswho utilize feedback intervention theory by em-phasizing the importance of considering the direc-tion of feedback flow in their research.

Our research also emphasizes the theoretical valueof separating negative feedback from positive feed-back. Some past works have investigated both pos-itive and negative feedback simultaneously andtreated the two as opposite ends of a single contin-uum. However, in real organizations, the assump-tion that negative and positive feedback are on thesame continuum may be problematic. Consideringthat a job comprises a set of many different tasks inorganizations (Morgeson, Garza, & Campion, 2012),a feedback sender may simultaneously provide afeedback recipient with negative feedback on sometasks and positive feedback on other tasks. Evenwithin one task, some behaviors may receive posi-tive feedback, while other behaviors may be thetargets of criticism. Therefore, positive andnegativefeedback can be simultaneously high and low; thus,the single-continuum assumption may not suc-cessfully reflect the setting in real organizations.Supporting this, Table 2 showsonly aweaknegativecorrelation between the two (r 5 2.26, p , .01). Inaddition, our confirmatory factor analysis con-firmed that these constructs differed. We hope futurescholars will build separate theoretical frameworksfor negative feedback and positive feedback andmeasure the two independently.

Our research also has implications for practi-tioners. First, it is important to emphasize thatbottom-up negative feedback creates an opportunityfor increasing recipient creativity. In organizations,followers tend to have low motivation to send criti-cism and negative feedback to their supervisors be-cause such feedback may be considered a challengeagainst people who have formal authority and social


power over important organizational resources (Brett& Atwater, 2001; DeNisi & Kluger, 2000; London &Beatty, 1993). In contrast to such a notion, our re-search revealed that supervisors tend to not takenegative feedback personally. Instead, supervisorspay heightened attention to addressing the problemsindicated by the bottom-up negative feedback in or-der to improve their behavior. Therefore, organiza-tions should consider instituting formal processesthat encourage followers to provide thoughtful, crit-ical feedback to their superiors. We believe that theevaluation system used by the company in Study 1can encourage bottom-up negative feedback. Thiscompany provided formal opportunities (i.e., thequarterly evaluations) for followers to provide nega-tive feedback to their supervisors. Even though thequarterly evaluations were not anonymous and thebottom-up feedback was used only for informativeand developmental purposes, the followers activelyutilized such formal opportunities to offer negativefeedback to their supervisors in both numerical andwritten forms.

Our research also demonstrated that the top-downand lateral negative feedback flows reduce recipientcreativity because they direct recipients’ attention tometa-processes and away from task processes.However, this conclusion does not imply that orga-nizations should prohibit supervisors from givingnegative feedback to their followers or that peersshould avoid giving negative feedback to one an-other. Instead, supervisors and peers might want tolimit the flow of negative feedback in the middle ofcreativity tasks, as it decreases recipient creativity.Perhaps negative feedback can be offered after crea-tivity tasks have been completed so that the recipi-ents can take time to cope with their threatenedmindset and think about the ways in which they canimprove their current creativity. This suggestion is inline with the findings in the brainstorming literaturethat participants should not criticize one another’sideas during group idea generation but should focusonly on generating asmany diverse ideas as possible(Esser, 1998). Another way of increasing receptivityof feedback recipients in the top-down and lateralfeedback flows could be that organizations offerseveral follow-up sessions to employees. Brett andAtwater (2001: 940) suggested that “executivecoaches or multiple follow-up sessions may helpthose receiving negative or discrepant feedback todeal with negative reactions and work throughthem.” Through multiple follow-up sessions, feed-back recipients may have a deeper understanding oftheir current creativity and evaluation standards,

which could help them to focus more on ways ofaddressing the creativity–standard gap instead ofattending to meta-processes.

Finally, our theory suggests that competitionmight be the main cause of feedback recipients at-tending to meta-processes when they receive nega-tive feedback from peers (i.e., lateral feedback flow).Competition is inevitable between peers becausesocial hierarchies in organizations often resemble apyramid with peers competing for limited resources(Bandura, 1977; Festinger, 1962; O’Keeffe et al.,1984). The sense of competition makes peers nega-tively react to lateral negative feedback by havingthemattend tometa-processes. It is thenpossible thatcarefully crafted negative feedback that does not re-mind of competition between peers could enhancepeers’ receptivity to lateral negative feedback. Onepossible way to achieve this goal is that peers mayneed to use temporal feedback instead of using socialcomparison feedback when they provide negativefeedback to other peers. “Temporal feedback” com-pares past performance with current performancewithin one employee, whereas “social comparisonfeedback” compares performance between em-ployees. Feedback researchers have shown that so-cial comparison feedback is the most prevalent typeof feedback in organizations, and it increases a senseof competition (Brown, Cron, & Slocum, 1998). Al-though empirical evidence concerning temporalfeedback is lacking in the literature, one group ofresearchers recently showed that people tend to bemore receptive to temporal feedback than socialcomparison feedback (Chun, Brockner, &DeCremer,2018) because such feedback can reduce the sense ofcompetition. Thus, organizations, such as the com-pany in Study 1, may need to consider asking em-ployees to use temporal feedback in their quarterlyperformance appraisal of peers.

Limitations and Future Research

As with any research, the current research haslimitations that can be fruitfully addressed by futureresearch. First, national cultures might have influ-enced our findings. Study 1was conducted in Korea,where cultures of shame and power distance are sa-lient (House, Hanges, Javidan, Dorfman, & Gupta,2004). In Korean culture, recipients of negativefeedback may feel shameful and believe that feed-back senders intended to humiliate them. This cul-tural background might have made the Koreanemployees in Study 1 more likely to attend to meta-processes rather than to task processes when they

2020 607Kim and Kim

received negative feedback, particularly in the top-down and lateral feedback flows. In addition, a highpower distance might influence supervisors’ atten-tion to task processes in the bottom-up feedbackflow, because, in a high power-distance culture, bothsupervisors and followers are unlikely to believe thatfollowers can threaten supervisors. Thus, rather thaninterpreting negative feedback as a threat to them-selves, supervisors might be able to increase theirfocus on task processes. Nevertheless, we believethat such cultural influences are unlikely becauseStudy 2 replicated the same findings in a NorthAmerican cultural context. However, the sample ofStudy 2 was undergraduate students, which lowersthe external validity of our findings. Therefore, wecall for future research to investigate our model witha sample of employees in other cultural contexts.

Our findings showed that the only feedback flowthat benefits recipient creativity is the bottom-upfeedback flow. However, it is possible that supervi-sors may also take bottom-up negative feedbackpersonally and be upset by it when their power isunstable or illegitimate (e.g., new leaders who haverecently joined). Research has shown that, whenpower holders perceive that their power is unstable,they become anxious about others’ evaluations,more vigilant to potential threats to their power,and motivated to protect their power (Lammers,Galinsky, Gordijn, &Otten, 2008). Thus, in situationsin which supervisors perceive their power as unsta-ble and illegitimate, negative feedback from fol-lowers may decrease their creativity because theymay feel threatened by negative feedback (meta-processes), and their task focus may be significantlydistracted by such feedback (task processes). Thus,future research could explore situations wherebottom-up negative feedback hampers supervisorcreativity.

Another limitation of our research is that we in-vestigated only the effects of negative feedback.Whether positive feedback positively or negativelyinfluences creativity remains an open question. Ourreview of past research on positive feedback andcreativity revealed inconsistencies in the literature:researchers have found positive (e.g., Hon, Chan, &Lu, 2013;VanDijk&Kluger, 2011) or null (e.g., Fodor& Carver, 2000) relationships between positivefeedback and creativity. This inconsistency could beexplained by completely different mechanisms fromours (task processes and meta-processes). For ex-ample, positive feedback may either increase or de-crease task motivation, which tends to have apositive relationship with creativity. On the one

hand, positive feedback may increase task motiva-tion because such recognition likely encourages re-cipients to put more effort into their tasks (Amabile,Conti, Coon, Lazenby, & Herron, 1996). On the otherhand, as positive feedback highlights the sufficiencyof recipients’ current behavior, it may provide asense of satisfaction with their current behavior,which can reduce task motivation (Locke & Latham,2002). Addressing this inconsistency was beyondthe scope of our research. It would beworthwhile forfuture researchers to shed further light on the rela-tionship between positive feedback and creativity.

The focus of this researchwas limited to creativityas an outcome of negative feedback. Thus, we didnot investigate another important organizationaloutcome, “routine performance,” which refers toemployee accomplishment on well-learned and fre-quently practiced tasks (Kilduff, 1993;March, 1991).Researchers have suggested that routine perfor-mance (or exploitation) and creativity (or explora-tion) are different, independent dimensions of jobperformance (Cyert & March, 1963; March & Simon,1958). According to our review, our theoreticalmodel is less likely to be applicable to routine per-formance. In particular, we believe that task pro-cesses in our model could be unrelated or evennegatively related to routine performance. By defi-nition, task processes refer to processes wherebyemployees attempt to generate different, diverse, andnovel strategies for their tasks. Such experimenta-tion with task strategies could be detrimental toroutine performance because routine performancerequires that employees strictly follow existing waysof doing their jobs (i.e., task routine; Cyert & March,1963; March & Simon, 1958). Therefore, even in thecase in which negative feedback increases recipientattention to task processes (e.g., bottom-up feedbackflow), we are not certain whether negative feedbackis beneficial for routine performance. As the mainpurpose of our research was to resolve the inconsis-tency between negative feedback and creativity, theexamination of routine performancewas beyond ourresearch scope. Thus, we call for future research onthis topic.

Finally, our research investigated only unsoughtnegative feedback. It is possible that sought negativefeedback, or negative feedback that follows feedbackseeking, may have completely different effects onrecipient creativity. Feedback seeking is a type ofproactive behavior that often represents the readi-ness to receive criticism and negative feedback andthe willingness to correct behaviors accordingly(Ashford, 1986; Ashford, Blatt, & VandeWalle, 2003;


Ashford & Tsui, 1991). Therefore, negative feedbackthat follows feedback seeking should be beneficialfor constructive changes in task behaviors and crea-tivity; negative feedback indicates the shortcomingsof one’s creativity, and feedback seekers are ready toadmit and correct their shortcomings. We believethis is a promising research area to which future re-searchers may be able to contribute.

Notwithstanding these limitations, this paperdeepens our understanding of the relationship be-tween negative feedback and recipient creativity. Byproposing a comprehensive theoretical frameworkthat incorporates inconsistent theories and empiri-cal findings in past research, our research contrib-utes to the creativity and feedback literature. Acrosstwo studies, we show that the effects of negativefeedback on recipient creativity depend on the di-rection of feedback flow.We hope to motivate futureresearchers to explore the differential effects offeedback on creativity by considering top-down,bottom-up, and lateral feedback flows.

REFERENCES

Amabile, T.M. 1983.The social psychology of creativity:Acomponential conceptualization. Journal of Person-ality and Social Psychology, 45: 357–376.

Amabile, T. M., Conti, R., Coon, H., Lazenby, J., & Herron,M. 1996. Assessing the work environment for crea-tivity. Academy of Management Journal, 39:1154–1184.

Anderson, C., & Berdahl, J. L. 2002. The experience ofpower: Examining the effects of power on approachand inhibition tendencies. Journal ofPersonality andSocial Psychology, 83: 1362–1377.

Anderson, N., Potocnik, K., & Zhou, J. 2014. Innovationand creativity in organizations: A state-of-the-sciencereview, prospective commentary, and guiding frame-work. Journal of Management, 40: 1297–1333.

Antonioni, D. 1996. Designing an effective 360-degree ap-praisal feedback process. Organizational Dynamics,25: 24–38.

Ashford, S. J. 1986. Feedback-seeking in individual adap-tation: A resource perspective. Academy of Man-agement Journal, 29: 465–487.

Ashford, S. J., Blatt, R., & VandeWalle, D. 2003. Reflectionson the looking glass: A reviewof research on feedback-seeking behavior in organizations. Journal of Man-agement, 29: 773–799.

Ashford, S. J., & Tsui, A. S. 1991. Self-regulation for man-agerial effectiveness: The role of active feedback seek-ing. Academy of Management Journal, 34: 251–280.

Bacharach, S.B., &Lawler, E. J. 1980.Powerandpolitics inorganizations. San Francisco, CA: Jossey-Bass.

Bandura, A. 1977. Social learning theory. EnglewoodCliffs, NJ: Prentice Hall.

Bandura, A. 1978. Social learning theory of aggression.Journal of Communication, 28: 12–29.

Bandura, A. 2001. Social cognitive theory: An agenticperspective.Annual Review of Psychology, 52: 1–26.

Beilock, S. L., Rydell, R. J., & McConnell, A. R. 2007. Ste-reotype threat and working memory: Mechanisms,alleviation, and spillover. Journal of ExperimentalPsychology. General, 136: 256–276.

Bendersky, C., & Shah, N. P. 2012. The cost of status en-hancement: Performance effects of individuals’ statusmobility in task groups. Organization Science, 23:308–322.

Birkinshaw, J. 2001. Strategies for managing internalcompetition. California Management Review, 44:21–38.

Brett, J. F., & Atwater, L. E. 2001. 360� feedback: Accuracy,reactions, and perceptions of usefulness. Journal ofApplied Psychology, 86: 930–942.

Brislin, R. W. 1990. Applied cross-cultural psychology:An introduction. Thousand Oaks, CA: SAGE.

Brown, S. P., Cron, W. L., & Slocum, J. W., Jr. 1998. Effectsof trait competitiveness and perceived intraorganiza-tional competition on salesperson goal setting andperformance. Journal of Marketing, 62: 88–98.

Byron, K., Khazanchi, S., & Nazarian, D. 2010. The rela-tionship between stressors and creativity: A meta-analysis examining competing theoretical models.Journal of Applied Psychology, 95: 201–212.

Cho, Y., Overbeck, J. R., & Carnevale, P. J. 2011. Statusconflict in negotiation. In J. R.Overbeck, E.A.Mannix,& M. A. Neale (Eds.), Negotiation and groups:111–136. Bingley, U.K.: Emerald.

Chun, J. S., Brockner, J., & De Cremer, D. 2018. How tem-poral and social comparisons in performance evalu-ation affect fairness perceptions. OrganizationalBehavior and Human Decision Processes, 145:1–15.

Colquitt, J. A., Conlon, D. E., Wesson, M. J., Porter,C. O. L. H., &Ng, K. Y. 2001. Justice at themillennium:A meta-analytic review of 25 years of organizationaljustice research. Journal of Applied Psychology, 86:425–445.

Cyert, R. M., & March, J. G. 1963. A behavioral theory ofthe firm. Englewood Cliffs, NJ: Prentice Hall.

De Stobbeleir, K. E., Ashford, S. J., & Buyens, D. 2011. Self-regulation of creativity at work: The role of feedback-seeking behavior in creative performance. Academyof Management Journal, 54: 811–831.

2020 609Kim and Kim

DeNisi, A. S., &Kluger, A.N. 2000. Feedback effectiveness:Can 360-degree appraisals be improved?Academy ofManagement Executive, 14: 129–139.

DeNisi, A. S., Randolph, W. A., & Blencoe, A. G. 1983.Potential problems with peer ratings. Academy ofManagement Journal, 26: 457–464.

Druskat, V. U., & Wolff, S. B. 1999. Effects and timingof developmental peer appraisals in self-managingwork groups. Journal of Applied Psychology, 84:58–74.

Elliot, A. J., & Reis, H. T. 2003. Attachment and explorationin adulthood. Journal of Personality and SocialPsychology, 85: 317–331.

Emerson, R. M. 1962. Power-dependence relations.American Sociological Review, 27: 31–41.

Esser, J. K. 1998. Alive and well after 25 years: A review ofgroupthink research. Organizational Behavior andHuman Decision Processes, 73: 116–141.

Fang, C., Kim, J. H. J., &Milliken, F. J. 2014.Whenbadnewsis sugarcoated: Information distortion, organizationalsearch and the behavioral theory of the firm.StrategicManagement Journal, 35: 1186–1201.

Festinger, L. 1954. A theory of social comparison pro-cesses. Human Relations, 7: 117–140.

Festinger, L. 1962. A theory of cognitive dissonance.Stanford, CA: Stanford University Press.

Fodor, E.M., &Carver, R.A. 2000.Achievement andpowermotives, performance feedback, and creativity. Jour-nal of Research in Personality, 34: 380–396.

Ford, C. M., & Gioia, D. A. 2000. Factors influencing crea-tivity in the domain of managerial decision making.Journal of Management, 26: 705–732.

Forster, J., Friedman, R. S., & Liberman, N. 2004. Temporalconstrual effects on abstract and concrete thinking:Consequences for insight and creative cognition.Journal of Personality and Social Psychology, 87:177–189.

Friedman, R. S., & Forster, J. 2001. The effects of pro-motion and prevention cues on creativity. Journalof Personality and Social Psychology, 81: 1001–1013.

Galinsky, A. D., Gruenfeld, D. H., &Magee, J. C. 2003. Frompower to action. Journal of Personality and SocialPsychology, 85: 453–466.

Galinsky, A. D., Magee, J. C., Gruenfeld, D. H., Whitson,J.A., &Liljenquist, K.A. 2008. Power reduces thepressof the situation: Implications for creativity, confor-mity, and dissonance. Journal of Personality andSocial Psychology, 95: 1450–1466.

George, J. M. 2007. Creativity in organizations. In J. P.Walsh & A. P. Brief (Eds.), Academy of Managementannals, 1: 439–477. New York, NY: Erlbaum.

George, J. M., & Zhou, J. 2001. When openness to experi-ence and conscientiousness are related to creativebehavior: An interactional approach. Journal of Ap-plied Psychology, 86: 513–524.

Hayes,A. F. 2013. Introduction tomediation,moderation,and conditional process analysis: A regression-based approach. New York, NY: Guilford Press.

Hon, A. H., Chan, W.W., & Lu, L. 2013. Overcoming work-related stress and promoting employee creativity inhotel industry: The role of task feedback from super-visor. International Journal of Hospitality Manage-ment, 33: 416–424.

House, R. J., Hanges, P. J., Javidan, M., Dorfman, P. W., &Gupta, V. 2004. Culture, leadership, and organiza-tions: The GLOBE study of 62 societies. ThousandOaks, CA: SAGE.

Hu, L.-T., & Bentler, P. M. 1998. Fit indices in covariancestructure modeling: Sensitivity to underparameterizedmodel misspecification. Psychological Methods, 3:424–453.

Ilgen, D. R., Fisher, C. D., & Taylor, M. S. 1979. Conse-quences of individual feedback on behavior in orga-nizations. Journal of Applied Psychology, 64:349–371.

Ilies, R., & Judge, T. A. 2005. Goal regulation across time:The effects of feedback and affect. Journal of AppliedPsychology, 90: 453–467.

Jordan, J., Sivanathan, N., & Galinsky, A. D. 2011. Some-thing to lose and nothing to gain: The role of stressin the interactive effect of power and stability onrisk taking. Administrative Science Quarterly, 56:530–558.

Karremans, J. C., & Smith, P. K. 2010. Having the power toforgive: When the experience of power increases in-terpersonal forgiveness. Personality and Social Psy-chology Bulletin, 36: 1010–1023.

Keltner, D., Gruenfeld, D. H., & Anderson, C. 2003. Power,approach, and inhibition.Psychological Review, 110:265–284.

Kilduff, M. 1993. Deconstructing organizations.Academyof Management Review, 18: 13–31.

Kim, Y. J., & Zhong, C.-B. 2017. Ideas rise from chaos: In-formation structure and creativity. OrganizationalBehavior and Human Decision Processes, 138:15–27.

Kluger, A. N., & DeNisi, A. 1996. The effects of feedbackinterventions on performance: A historical review,a meta-analysis, and a preliminary feedback inter-vention theory. Psychological Bulletin, 119: 254–284.

Lammers, J., Galinsky, A. D., Gordijn, E. H., & Otten, S.2008. Illegitimacy moderates the effects of power onapproach. Psychological Science, 19: 558–564.


Locke, E. A., & Latham, G. P. 2002. Building a practicallyuseful theory of goal setting and task motivation:A 35-year odyssey. American Psychologist, 57:705–717.

London,M., & Beatty, R.W. 1993. 360-degree feedback as acompetitive advantage. Human Resource Manage-ment, 32: 353–372.

Magee, J. C., & Galinsky, A. D. 2008. Social hierarchy: Theself‐reinforcing nature of power and status. Academyof Management Annals, 2: 351–398.

Magee, J. C., & Smith, P. K. 2013. The social distance theoryof power. Personality and Social Psychology Re-view, 17: 158–186.

March, J. G. 1991. Exploration and exploitation in orga-nizational learning. Organization Science, 2: 71–87.

March, J. G. & Simon, H. A. 1958. Organizations. NewYork, NY: Wiley.

McGregor, H. A., & Elliot, A. J. 2002. Achievement goals aspredictors of achievement-relevant processes prior totask engagement. Journal ofEducationalPsychology,94: 381–395.

Morgeson, F. P., Garza, A. S., &Campion,M.A. 2012.Workdesign. In I. B.Weiner (Ed.),Handbook of psychology(2nd ed.): 525–559. Hoboken, NJ: Wiley.

Mumford, M. D. 1983. Social comparison theory and theevaluation of peer evaluations: A review and someapplied implications. Personnel Psychology, 36:867–881.

Nijstad, B. A., De Dreu, C. K., Rietzschel, E. F., & Baas, M.2010. The dual pathway to creativity model: Crea-tive ideation as a function of flexibility and persis-tence. European Review of Social Psychology, 21:34–77.

O’Keeffe, M., Viscusi, W. K., & Zeckhauser, R. J. 1984.Economic contests: Comparative reward schemes.Journal of Labor Economics, 2: 27–56.

Oldham,G. R., & Cummings, A. 1996. Employee creativity:Personal and contextual factors at work. Academy ofManagement Journal, 39: 607–634.

Perry-Smith, J. E. 2006. Social yet creative: The role of so-cial relationships in facilitating individual creativity.Academy of Management Journal, 49: 85–101.

Pfeffer, J., & Salancik, G. R. 1974. Organizational decisionmaking as a political process: The case of a universitybudget. Administrative Science Quarterly, 19:135–151.

Rogers, T., & Feller, A. 2016. Discouraged by peer excel-lence: Exposure to exemplary peer performance cau-ses quitting. Psychological Science, 27: 365–374.

Shalley, C. E. 1991. Effects of productivity goals, crea-tivity goals, and personal discretion on individual

creativity. Journal of Applied Psychology, 76:179–185.

Shalley, C. E., Zhou, J., &Oldham,G. R. 2004. The effects ofpersonal and contextual characteristics on creativity:Where should we go from here? Journal of Manage-ment, 30: 933–958.

Smith, P. K., Jostmann, N. B., Galinsky, A. D., & Van Dijk,W. W. 2008. Lacking power impairs executive func-tions. Psychological Science, 19: 441–447.

Smith, P.K., &Trope,Y. 2006.You focuson the forestwhenyou’re in charge of the trees: Power priming and ab-stract information processing. Journal of Personalityand Social Psychology, 90: 578.

Steele, C.M., Spencer, S. J., &Aronson, J. 2002. Contendingwith group image: The psychology of stereotype andsocial identity threat. In M. P. Zanna (Ed.), Advancesin experimental social psychology, vol. 34: 379–440.San Diego, CA: Elsevier.

Tauer, J. M., & Harackiewicz, J. M. 1999. Winning isn’teverything: Competition, achievement orientation,and intrinsic motivation. Journal of ExperimentalSocial Psychology, 35: 209–238.

Tomaka, J., Palacios, R., Schneider, K. T., Colotla, M.,Concha, J. B., & Herrald, M. M. 1999. Assertivenesspredicts threat and challenge reactions to potentialstress among women. Journal of Personality andSocial Psychology, 76: 1008–1021.

Tost, L. P., Gino, F., & Larrick, R. P. 2013. When powermakes others speechless: The negative impact ofleader power on team performance. Academy ofManagement Journal, 56: 1465–1486.

Van Dijk, D., & Kluger, A. N. 2011. Task type as a mod-erator of positive/negative feedback effects onmotivation and performance: A regulatory focusperspective. Journal of Organizational Behavior,32: 1084–1105.

Vuori, T. O., & Huy, Q. N. 2015. Distributed attention andshared emotions in the innovation process: HowNokia lost the smartphone battle. AdministrativeScience Quarterly, 61: 9–51.

Yukl, G. A. 2010. Leadership in organizations (7th ed.).Upper Saddle River, NJ: Prentice Hall.

Zenger, J., & Folkman, J. 2017, May 2. Why do so manymanagers avoid giving praise? Harvard Business Re-view. Retrieved from https://hbr.org/2017/05/why-do-so-many-managers-avoid-giving-praise

Zhou, J. 1998. Feedback valence, feedback style, task au-tonomy, and achievement orientation: Interactive ef-fects on creative performance. Journal of AppliedPsychology, 83: 261–276.

Zhou, J. 2008. Promoting creativity through feedback. In J.Zhou & C. Shalley (Eds.),Handbook of organizational

2020 611Kim and Kim

https://hbr.org/2017/05/why-do-so-many-managers-avoid-giving-praise

https://hbr.org/2017/05/why-do-so-many-managers-avoid-giving-praise

creativity: 125–145.NewYork,NY:LawrenceErlbaumAssociates.

Zhou, J., &George, J.M. 2001.When jobdissatisfaction leadsto creativity: Encouraging the expression of voice.Academy of Management Journal, 44: 682–696.

Yeun Joon Kim ([email protected]) is assistant profes-sor of organizational behavior at Judge Business School,University of Cambridge. He received his PhD from theUniversity of Toronto, hisMS and BA from Seoul NationalUniversity, and another BA from Yonsei University. Hisresearch interests include creativity, leadership, andculture.

Junha Kim ([email protected]) is a doctoral student inmarketing at Fisher College of Business, Ohio State Uni-versity. He received his BA and MS from KAIST. His re-search interests generally focus on consumer judgmentand decision-making.

APPENDIX A

Negative (Positive) Feedback

Note: “reference” 5 supervisor/peers/followers.

(1) In the performance appraisal, [reference] toldmethat I didn’t (did) do a good job.

(2) In the performance appraisal, [reference] criti-cized (praised) my work.

(3) In the performance appraisal, [reference] gaveme negative (positive) feedback.

(4) In the performance appraisal, [reference] toldmethat my performance is not (is) up to thestandard.

(5) In the performance appraisal, [reference] toldmethat my performance is poor (excellent).

(6) In theperformance appraisal, [reference] indicatedthat [reference] is not (is) happy with my work.

(7) In the performance appraisal, [reference] gaveme many criticisms (compliments).

Task Processes

(1) This feedback helped me pay more attention tohow I conduct my tasks.

(2) This feedback helped me think about strategiesthat I could use to improvemy task performance.

(3) This feedback made me wonder whether therewere different approaches I coulduse to dobetteron my tasks.

(4) This feedback made me improve the processesinvolved in completing my tasks.

Meta-Processes

Note: “reference” 5 supervisor/peers/followers.

(1) This feedbackmademe care about how I presentmyself to my [reference].

(2) This feedback made me be more self-consciousabout the way I look to my [reference].

(3) This feedback made me worry about the im-pression my [reference] has of me.

(4) This feedback made me think about how my[reference] might perceive me.

(5) This feedbackmademe concerned about howmy[reference] evaluates my abilities or weaknesses.

Feedback Text Used in Study 2

Note: In this text, grammatical errors wereintended.

“I am actually just finishing reading a book aboutcreativity and it gave some good advice. If you want tothink ‘outside the box,’ you may want to think aboutbroad categories first, rather than focusing on detailedideas. For example, on the coffee break topic you maystart thinking w ideas related to at least three broadcategories (e.g., employees, HR system, or supervisors).Then focusononecategory&generate ideaswithin thatcategory. For example: for the HR system, think aboutsome rules/regulations that prevent the problem. Ifyou cannot come up with creative ideas within thecategory, you can go on to the next category.

“Another thing they said is that good ideassometimes combining two or more categories. E.g.:initiate a contest where employees generate ideasabout effective HR regulations to solve the prob-lem. This provides an opportunity that employeesthink about their behaviors as well as HR policy forpreventing such behaviors. (This would be relatedto both employee and HR system categories.)

“The book also talks about the importance of per-sistence I mentioned that you can jump into other cat-egories when you feel you cannot generate creativeideaswithinacertaincategory,pleasedonotgiveup tooeasily. Sometimes, you can suddenly come up withgood ideas when you put more efforts within thecategory.”




Documents

Does Negative Feedback Benefit (or Harm) Recipient