The effect of others’ attributions in feedback

Abstract

Purpose

Research has supported both feedback’s variable effects on performance and the effect of attributions on subsequent behavior. Managers’ attributions for subordinates’ performance affect how they react to those subordinates and the feedback they give, and subordinates’ own attributions affect their subsequent behavior. It is unclear whether (or how) a manager’s attributions for subordinate behavior affect subordinate behavior. Building on research that shows emotional reactions in response to attributions in feedback, this study aims to examine how recipients’ perceptions and subsequent effort and performance are affected when others’ attributions are shared through feedback.

Design/methodology/approach

Drawing on attribution theory and feedback intervention theory, this study conducts a lab experiment using manipulated performance feedback to test the effects of feedback sign and attributions in the feedback. Perceptions of the attribution are also measured to test their effects. The data were analyzed using analysis of variance and regression in SPSS 27.

Findings

Results show that perceptions of the attribution communicated in feedback, rather than feedback sign alone, affect perceived valence of the feedback (e.g. feedback with an attribution to luck is generally perceived as negative). These perceptions also affect feedback acceptance and impact subsequent effort and performance more than the “objective” attribution, underscoring the importance of recipient reactions and perceptions in the feedback process.

Originality/value

This paper shows that recipients’ perceptions of others’ attributions included in feedback impact feedback reactions, effort and performance. This is valuable to scholars researching feedback and to practitioners to better understand how feedback they deliver may be interpreted.

Citation

Sockbeson, C.E.S. and DeNisi, A.S. (2023), "The effect of others’ attributions in feedback", Organization Management Journal , Vol. 20 No. 2, pp. 75-85. https://doi.org/10.1108/OMJ-02-2022-1486

Introduction

The present study explores how attributions, or causal explanations, for observed performance, communicated through feedback, might affect recipient reactions and subsequent behavior. Feedback, or information communicated about performance, is lauded as an important motivational resource in the workplace (Locke et al., 1981) and critical for performance management. Yet, feedback may not always work the way we think it should. Given the importance of feedback in the workplace, it is important to understand factors that might influence feedback effectiveness. One such factor might be the feedback provider’s attributions for observed performance, if communicated to recipients.

As we detail below, prior research indicates that feedback effects are variable, employees’ attributions affect their performance and managers’ attributions for subordinate performance affect how they treat those subordinates, but how attributions in feedback affect recipient reactions and outcomes has not been sufficiently researched. Our study’s central research question needs more exploration: Do attributions for employee behavior, transmitted through feedback, affect employees’ reactions and behavior? If so, how? Our starting point is Kluger and DeNisi’s (1996) feedback intervention theory (FIT), but our study’s hypotheses are mainly based on attribution theory, specifically research on attributions at work and their effects on behavior. Our study has practical implications, as managers need to understand when feedback will and will not be effective, and theoretical implications, as there is some evidence of the impact of others’ attributions on emotional reactions (Hareli & Weiner, 2002), but little on effort or performance.

Attributions and feedback

Attribution theory (Heider, 1958; Kelley, 1973; Weiner, 1985) focuses on the drive to explain observed behaviors and outcomes. Individuals are compelled to explain their own and others’ successes and failures; these attributions affect subsequent behavior. Particularly, personally relevant work outcomes trigger attributions, which impact behavioral motivation (Harvey & Martinko, 2009). Typically, four attributions for performance are considered: luck, task difficulty, effort and ability; these encompass the dimensions of locus of causality (internal–external) and stability (stable–unstable) (see review by Martinko et al., 2006). Attribution theories, divided into those that focus on antecedents and those that focus on consequences (Kelley & Michela, 1980), originated in the realm of psychology but increasingly have been applied in management research.

For example, the attributions employees make for their own performance can affect their self-efficacy, goals and subsequent performance (Donovan & Williams, 2003; Stajkovic & Sommer, 2000). Other research has shown that the attributions managers make for subordinate performance can affect both how the manager behaves toward the subordinate (Green & Mitchell, 1979) and the content of feedback given to the subordinate (Ilgen & Knowlton, 1980; Moss & Martinko, 1998). Specifically, managers treat subordinates differently when they attribute their poor performance to lack of effort, and they adjust their feedback accordingly. Overall, research in management demonstrates that attributions help to explain affect, subsequent performance, leader-member exchange and punishment and reward intentions (Harvey et al., 2014, for a meta-analysis).

These studies show that attribution theory is useful in management research, though they overlook the case where managers’ attributions for employee performance are communicated to employees. An exception was a study by Bannister (1986) where experimental subjects received performance feedback, were asked to develop their own attributions and subsequently received supervisor attributions for their performance. Discrepancies between self- and others’ attributions affected recipients’ reactions (credibility and satisfaction), but not behavioral intentions. Hareli and Weiner (2002) explicate how others’ attributions affect a myriad of emotional reactions, but effects on performance are left an open question.

Performance and attributions do not occur in isolation, and responses to feedback, including behavioral intentions, are sensitive to others’ reactions, possibly including attributions (Hareli & Weiner, 2002). Furthermore, Kluger and DeNisi’s (1996) meta-analysis established that feedback is not always effective in improving performance. FIT argues that, because cognitive resources are limited, feedback that directs attention away from the task at hand can reduce performance (Kluger & DeNisi, 1996). Others have found that recipients’ reactions to feedback can also influence feedback effectiveness (Jawahar, 2010), and attributions in feedback affect emotional reactions (Hareli & Hess, 2008). Thus, managers’ attributions for performance, communicated through feedback, could affect recipient reactions, including their willingness to accept the feedback and view it as credible, and whether subordinates’ attention remains on task. Therefore, whether managers’ attributions for subordinate behavior, transmitted through feedback, affect subordinate reactions and behavior is important to explore. Knowing this can inform organizations about how to make feedback more effective.

Past research leaves some questions unanswered. For example, Hareli and Hess (2008) did not report effects on subsequent performance and Bannister’s (1986) study did not consider effects of feedback sign (positive or negative). Recipients react better to positive feedback (Young et al., 2017), and Anseel and Lievens (2009) found that feedback sign affects feedback acceptance, which is critical for effects on behavior. Research on managerial attributions found that there was an interaction between the attributions made and the sign of the feedback (Green & Mitchell, 1979). Thus, it is likely that any effects of attributions communicated in feedback will depend on the sign of the feedback as well.

The “self-serving bias” in attributions (Martinko et al., 2006; Zuckerman, 2009), a predisposition to attribute one’s success to internal causes (such as high ability) and one’s failures to external causes (such as a difficult task), may play a role. Given this tendency, attributions that align with those beliefs should be more acceptable and so more likely to improve subsequent performance. But feedback messages which include attributions that contradict the self-serving bias (e.g. attributing poor performance to an internal cause) are likely to induce some type of cognitive dissonance (Festinger, 1957). Attempts to reduce that dissonance may divert cognitive resources away from the task at hand and trigger distracting ruminations about the self, which FIT posits should have a negative effect on performance. Thus, we suggest that attributions, both by themselves and especially in combination with the sign of the feedback, will affect reactions to that feedback as well as subsequent performance such that these two factors will help determine feedback effectiveness. Specifically, we hypothesize that:

H1.

Positive feedback attributed to internal causes and negative feedback attributed to external causes will be more acceptable and more credible.

Performance attributions communicated in feedback might also have effects independent of the sign of the feedback. Another potentially important aspect of performance attributions is stability; performance may be attributed to stable causes (e.g. ability) or unstable causes (e.g. luck; Weiner, 1985). One of the tenets of attribution theory is that “outcomes ascribed to stable causes will be anticipated to be repeated in the future with a greater degree of certainty than are outcomes ascribed to uncertain causes” (Heider, 1958, p. 559). Thus, stable attributions for performance are likely to have a smaller impact on subsequent performance than unstable attributions, which provide hope (or uncertainty) about a different level of performance in the future. Thus, we propose:

H2.

In feedback, attributions to unstable causes will affect effort and performance more positively than attributions to stable causes.

Attributions also vary in their locus (internal or external), and the interaction of stability and locus of causality could also be important. For example, an attribution for poor performance to lack of ability (a stable, internal attribution) is less likely to result in additional effort being exerted in the future, because it is unlikely to have any effect (cf., Bandura, 2001). However, an attribution to task difficulty (stable, external) may result in extra effort to overcome the difficulty. Attributing performance to effort (unstable, internal) is also likely to encourage persistence (Weiner, 1985, for a brief review). Such attributions also have the advantage of not leading to questions about self-worth, which (according to FIT) could distract attention from the task at hand. With luck (external, unstable), on the other hand, feedback recipients may be unsure how to respond or change tactics to improve. Thus, expanding on H2, we propose:

H3.

Attributions to effort, in feedback messages, will have the greatest positive effect on reactions and subsequent performance, as opposed to any other attribution.

Although attributions have long been categorized as internal or external, stable or unstable, recipient perceptions vary along a continuum. For example, some may interpret ability as more stable than others (e.g. implicit theory of intelligence; Dweck & Leggett, 1988). Thus, recipients may categorize the same attribution differently based on individual perception. In addition, perceptions of valence (how positive or negative recipients interpret the feedback to be) may also affect recipient perceptions and reactions, such as whether the recipients accept the feedback and whether they believe the attribution (credibility). Thus, recipients may interpret feedback differently than the feedback provider intended, and these perceptions are likely to have a larger effect on reactions and subsequent performance than the objective categorization of feedback sign and attribution, because the impact of feedback is largely because of recipient reactions (Jawahar, 2010). Thus, recipients’ perception of the attribution and feedback characteristics is likely to be as important as the actual feedback, perhaps even more so. We propose:

H4.

Recipients’ perception of feedback, in terms of valence, locus of causality and stability of the attribution, will affect (a) feedback acceptance, (b) feedback credibility, (c) subsequent effort and (d) subsequent performance independent of the objective sign and attributions in feedback.

Method

Participants

Participants were 269 undergraduate business students from a mid-sized university in the Southern United States. Mean age was 20 years, and the sample was 49% female. The sample was 85.5% white. Students received extra credit in one class in exchange for participation.

Procedure

We told participants they were helping to pretest materials for future studies and presented an anagram task featuring scrambled letters, similar to Tolli and Schmidt (2008; anagrams from Dyczewski & Markman, 2012). Each problem might have no solution, one solution or multiple solutions (e.g. “Ipsil” can be “spill” or “pills”), making performance ambiguous. Participants were shown a sample problem and solutions. Discovering 50% of the possible solutions was acceptable performance. Participants went through two trials of ten problems each. After the first trial, participants provided demographic information and were told their first round of problems would be graded during this time. Regardless of actual performance, half of the participants were told they answered 70% correctly (positive feedback) and half were told they answered 30% correctly (negative feedback). The feedback also contained attributions for their performance (e.g. “You must have an inherent aptitude for these types of problems” for a positive attribution to ability). Participants rated their perceptions regarding the feedback before completing the second set of problems, after which they were thanked, debriefed and dismissed.

Measures

Feedback valence: Participants rated the feedback they received on a seven-point Likert scale, ranging from −3 (very negative) to 3 (very positive).

Causal attributions: Causal attributions were measured with the Revised Causal Dimension Scale (McAuley et al., 1992), on a scale from 1 to 9. Locus of causality (α = 0.854) and stability (α = 0.808) were measured with three items each, centered, with higher scores indicating more internal and more stable attributions, respectively.

Attribution credibility: Participants were asked to rate how likely the attribution they received as feedback (i.e. ability, effort, task difficulty or luck) was to be the main cause of their level of performance, on a scale from 1 to 5.

Feedback acceptance: Feedback acceptance was measured with three items from Nease et al. (1999), on a scale from 1 to 5 (α = 0.743), centered.

Effort: Effort was measured as the time spent on task. Time on the first trial was entered as a control, with time on the second trial serving as the dependent variable.

Performance: Performance was measured as the score (from 0 to 100) on the second trial of problems. Performance on the first trial was used as a control.

Results

Manipulation checks

Means and standard deviations by condition are presented in Table 1. Analysis of variance (ANOVA) confirmed that subjects in the positive and negative feedback conditions perceived the feedback significantly differently [M_negative = −1.267; M_positive = 0.913; F(1, 267) = 146.40, p < 0.001, partial η² = 0.35]. Participants in the internal conditions (ability and effort) perceived the locus of the attribution to be more internal than participants in the external conditions [luck and task difficulty; F(1, 267) = 56.04, p < 0.001, partial η² = 0.21]. Subjects in the stable conditions (ability and task difficulty) rated the attributions as more stable than those in the unstable conditions [effort and luck; F(1, 267) = 14.77, p < 0.001, partial η² = 0.05].

Objective attribution effects on recipient reactions, acceptance and performance

H1 predicted that feedback sign would interact with attributions for performance in impacting feedback acceptance and credibility. The sign of feedback and the attribution in feedback impacted credibility [F(1, 261) = 6.28, p = 0.01, partial η² = 0.02; F(1, 261) = 10.27, p < 0.001, partial η² = 0.11] and acceptance [F(1, 261) = 10.69, p = 0.001, partial η² = 0.04; F(1, 261) = 6.54, p < 0.001, partial η² = 0.07]. Positive feedback was more acceptable and more credible than negative feedback, in line with prior research. Luck drove the effect of attributions on credibility, being significantly less credible than all and significantly less acceptable than task or ability attributions. In addition, sign × attribution impacted acceptance [F(1, 261) = 8.73, p < 0.001, partial η² = 0.09]. Additional analyses coding the attributions by locus and stability revealed that stability positively affected acceptance [F(1, 261) = 11.52, p < 0.001, partial η² = 0.04] and credibility [F(1,261) = 14.44, p < 0.001, partial η² = 0.05] and that sign × locus affected acceptance [F(1,261) = 17.96, p < 0.001, partial η² = 0.06] and credibility [F(1,261) = 4.75, p = 0.03, partial η² = 0.02] such that positive feedback with internal attributions was significantly more acceptable than any other condition and significantly more credible than negative feedback. Negative attributions to external causes were rated as more acceptable and more credible than negative attributions to internal causes, but not significantly. Thus, H1 is partially supported.

To test H2 and H3, we first ran ANOVAs to determine how the feedback conditions directly affected effort and performance. The attribution communicated in feedback did not affect subsequent effort [F(3, 260) = 0.24, p = 0.87, partial η² = 0.003], but the sign of the feedback did [F(1, 260) = 4.12, p = 0.04, partial η² = 0.02], with participants in the negative feedback condition spending more time on the task following feedback (M_negative = 320.86; M_positive= 286.63). The attribution did not affect subsequent performance [F(3, 260) = 0.15, p = 0.93, partial η² = 0.002]. Sign also did not have a significant effect, although it approached significance [F(1, 260) = 3.64, p = 0.058, partial η² = 0.01], with participants in the negative feedback condition tending to perform higher. Thus, H2 and H3 were not supported.

Perceptions of feedback

The manipulation checks indicated that the conditions were perceived correctly; however, initial analyses indicated that the attributions given and the sign of the feedback often interacted to affect recipient perceptions, as predicted. Positive feedback was perceived as more stable [F(1, 261) = 20.55, p < 0.001, partial η² = 0.07], but not more internal [F(1, 261) = 3.43, p = 0.07, partial η² = 0.01]. Likewise, the attribution given has a significant relationship with perceived locus of causality [F(3, 261) = 23.82, p < 0.001, partial η² = 0.22], stability [F(3, 261) = 7.4, p < 0.001, partial η² = 0.08] and valence [F(3, 261) = 4.14, p = 0.01, partial η² = 0.05].

The interaction between attribution and feedback sign was also significant for stability [F(3,261) = 5.64, p = 0.001, partial η² = 0.06], locus of causality [F(3,261) = 4.22, p = 0.01, partial η² = 0.05] and valence [F(3,261) = 35.34, p < 0.001, partial η² = 0.29]. Generally, attributions to luck tend to be interpreted differently than other attributions. Most attributions (especially ability) are seen as more stable as feedback becomes more positive, but a positive attribution to luck is less stable than a negative one. Ability and effort are seen as more internal as feedback becomes more positive; task difficulty’s perceived locus does not vary much, but again, luck is seen as more external when feedback is positive. Negative feedback with internal attributions is interpreted as more negative (and positive feedback with those attributions more positive) than luck or task attributions, and a positive attribution to luck is not perceived as positive feedback at all. Positive feedback attributed to luck was less acceptable than negative feedback with that attribution; the relationship was the opposite for all other attributions.

Effects of perceptions on reactions

Supporting H4, perceptions of the feedback also affected reactions in terms of credibility and feedback acceptance. Perceived locus of causality (t = 2.81, p = 0.004) and perceived valence (t = 2.47, p = 0.01), but not perceived stability (t = 0.72, p = 0.471), affected credibility. Feedback that was perceived as more positive or more internal was more believable as the main cause of participants’ performance. Perceived locus of causality (t = 2.93, p = 0.004) and perceived valence (t = 5.54, p < 0.001), but not perceived stability (t = −0.17, p = 0.87), also affected feedback acceptance, but these main effects were qualified by significant valence × locus (t = 2.09, p = 0.04) and locus × stability (t = −3.79, p < 0.001) interactions. Positive feedback is more acceptable than negative feedback, but this is especially true when paired with an internal rather than external attribution. Feedback with stable attributions remains at the same level of acceptance regardless of locus, but feedback with unstable attributions is more acceptable when attributions are perceived to be internal rather than external.

Effects of perceptions on effort and performance

Next, we ran regressions to determine how perceptions affected effort and performance. The only significant effect of perceptions on effort was the interaction between perceived valence, perceived locus and perceived stability (t = 2.04, p = 0.043). Effort was higher after feedback perceived as positive when attributions were perceived to be external, particularly if the attribution was perceived to be unstable. After negative feedback, effort increased for unstable attributions perceived to be more internal and decreased for stable attributions perceived to be more internal (see Figure 1). Feedback acceptance positively impacted subsequent performance (t = 2.35, p = 0.02), in line with prior research. There were no other significant main effects on performance, but the interaction of perceived locus × stability (t = −3.40, p = 0.001) and perceived valence × locus × stability both significantly affected performance (t = 2.459, p = 0.015; see Figure 1). When participants perceived that the attribution given was to an internal source, performance decreased as that attribution was perceived as more stable, particularly when the feedback was perceived as negative. When participants perceived the feedback was negative and external, perceptions of stability positively affected performance. The highest performance and the highest effort resulted when feedback was perceived as negative, internal and unstable.

Discussion

Feedback sign interacted with attributions for performance communicated in feedback to affect recipient perceptions and reactions. Importantly, feedback messages that include others’ attributions can influence what recipients do about the feedback received. For example, positive feedback containing an attribution to luck was perceived as negative feedback. We also found that perceptions, rather than the actual feedback, predicted subsequent effort and performance. Generally, positive feedback was more acceptable than negative (except when good performance was attributed to luck), and attributions to luck were seen as the least credible. Furthermore, while neither the perceived sign, locus of causality or stability affected subsequent performance on their own, the interaction of the three did. Finally, feedback acceptance also positively impacted performance.

This study adds to the body of knowledge on attributions at work and feedback. Feedback recipients did not always perceive attributions as they were intended. This is potentially important because recipients typically believe they have more control over performance attributed to internal causes and so are more likely to work to improve when performance is attributed to those causes. If recipients perceive attributions contained in feedback messages differently than they were intended, it is difficult to predict how they will react. Likewise, feedback recipients did not always perceive the sign of the feedback as it was intended, and the interaction of feedback sign and attributions for performance led to recipients perceiving some positive feedback as negative. All of this seems to reinforce prior research suggesting that how feedback is delivered is important for how it is perceived and how it will affect subsequent performance (Baur et al., 2014).

Thus, managers should remember that what they intend to communicate may not be what comes across to employees and should be careful when providing attributions for employees’ level of performance, implicitly or explicitly. The combination of whether feedback is interpreted as positive or negative, as having stable or unstable causes and as having internal or external causes affects subsequent effort and performance. Feedback is important, but its effects are complex.

Limitations and future research

Like all studies, this one has limitations. As a lab study conducted to determine whether attributions in feedback are capable of impacting recipient reactions and subsequent performance, it did not involve real employees receiving feedback on their actual jobs by real supervisors, so there is always some question about external validity. Laboratory studies allow control over feedback messages that cannot be controlled in the field. Thus, they provide insight into what might occur and what subsequent field research should focus upon. One of the findings of the present study was that recipients’ perceptions of feedback and attribution messages were somewhat different from the actual messages. This suggests that future field research must focus on how feedback messages are perceived and how to ensure that they are perceived as intended. Now that we have established that attributions contained in feedback messages can influence reactions, further research in the field about what types of attributions are likely to be contained in feedback and how recipients react when they have an ongoing relationship with the feedback source is warranted to gauge the frequency and generalizability of the effects. Future research should also focus upon the nature of the job or tasks involved, as this is an important determinant of recipients’ reactions to the feedback they receive (Van Dijk & Kluger, 2011). Feedback is critical in organizations and anything that sheds light on how to make that feedback more effective is an important direction for future research.

Conclusion

Feedback recipients interpret positive and negative feedback differently depending on the attribution for performance communicated in the feedback. These perceptions impact recipient reactions, effort and performance, with feedback perceived as negative, internal and unstable increasing effort and performance. Managers are cautioned to consider how attributions will be interpreted when constructing feedback messages.

Figures

Figure 1.

Effects of recipients’ perceptions on performance

Table 1.

Feedback perceptions and reactions by condition

	Ability				Effort				Task difficulty				Luck
	Positive		Negative		Positive		Negative		Positive		Negative		Positive		Negative
N	30		30		34		32		36		36		38		33
	M	SD	M	SD	M	SD	M	SD	M	SD	M	SD	M	SD	M	SD
Perceived valence	1.77	1.22	−1.73	1.08	1.82	0.94	−1.94	0.95	1.00	1.27	−0.94	1.17	−0.66	1.60	−0.55	1.42
Locus of causality	20.27	4.53	17.43	5.49	19.74	4.25	16.44	5.38	15.47	6.129	14.83	5.759	11.16	4.36	13.36	3.90
Stability	16.40	5.45	10.67	4.62	12.76	4.84	9.25	5.09	13.44	5.88	10.75	4.94	9.11	4.64	10.21	3.11
Acceptance	3.37	0.64	2.74	0.80	3.31	0.67	2.62	0.82	3.28	0.68	3.03	0.68	2.43	0.68	2.86	0.76
Credibility	3.87	0.82	3.10	1.32	3.71	1.03	3.19	1.49	3.78	0.96	3.75	0.84	2.79	1.12	2.73	1.26

Note:

Raw means are presented here, but mean-centered variables for locus, stability, acceptance and credibility are used in the analyses