Don’t tell me about my moral failures but motivate me to improve: Increasing effectiveness of outgroup criticism by criticizing one’s competence

Criticizing people on their prior moral failures often causes them to react defen-sively, especially when this is done by an outgroup. In the current research, we tested whether people become more receptive to such outgroup criticism when it refers to (failures of) their competence, rather than their morality. We conducted two studies, using a 2: Critic's group-membership (receiving criticism from an ingroup vs. out-group) × 2: Dimension (competence vs. morality as focal concern addressed with the criticism) mixed design. Findings showed that, regardless of source, participants made fewer negative attributions, were more motivated to improve, and more often indicated they changed their behavior after they had been criticized on their competence, instead of on their morality. Thus, criticizing past behavior for failing to show competence instead of morality might be a way to reduce defensive responses and to stimulate behavior change, even for outgroup critics.


| INTRODUC TI ON
In a Dutch interview in 2007, a journalist questioned the former CEO of ABN AMRO for receiving 26 million euros while the bank was taken over by a consortium bank.In the video of this interview, it can be observed that the former CEO had a very defensive response when being accused of acting disloyally towards his former colleagues at ABN.Moreover, he repeatedly referred to agreements that had been made about these types of payments in the financial world during a takeover.According to his group, receiving that much money did not correspond to violating any norms.However, the media and the general public saw this as a moral scandal.In the current article, we will present research findings indicating that a more successful strategy for the journalist (being an outsider) might have been to criticize the former CEO's competencies as a banker, rather than questioning his integrity.This might have stimulated a more open-minded discussion on how things can be changed in the future, rather than the defensive responses from the former CEO dominating the conversation.

| Criticism from an ingroup versus outgroup source
The degree of the banker's defensive reaction might have been influenced by the journalist not being a banker himself.Whereas criticism from our ingroups (e.g., fellow bankers; Tajfel & Turner, 1979) is seen as helpful for the self and the ingroup, criticism from outgroups (e.g., journalists) is often not accepted (Esposo et al., 2013;Hornsey & Imani, 2004;Hornsey et al., 2002).The defensive response towards an outgroup criticizing the ingroup (i.e., group-directed criticism) is often referred to as the intergroup sensitivity effect (e.g., Hornsey & Esposo, 2009).The underlying process here has proven to be related to making negative attributions about both the critic (i.e., the person criticizing) and the criticism (i.e., the message, Hornsey & Imani, 2004).When someone is criticized by an outgroup member, people often attribute negative motives to the critic and perceive the criticism as less constructive.
However, group-directed criticism is only loosely related to any kind of individual-level improvement attempts (e.g., behavioral change).An individual can hardly change the behavior of the whole group and is always presented with only the negative side of criticism (e.g., negative affect, threat) and not the positive side (e.g., opportunity to grow).Thus, to be able to stimulate behavioral change, which is the ultimate goal when criticizing, we should also look at criticism at the individual level.Specifically, the differences between group-and individual-level criticism come from the different paths that a person can follow when responding to outgroup criticism.In group-directed criticism, the relevance of people's own past and future actions is not clear and they are not presented with an opportunity to make up for past failures (van der Toorn et al., 2015).
However, in individual-level criticism situations, this opportunity can make people motivated to improve.In the current research, we examined three underlying processes (i.e., emotional, cognitive, and motivational) that influence people's reactions to being criticized for their individual behavior by an outgroup.Thereby, we offer a novel perspective on why (and how strongly) the intergroup sensitivity effect occurs in individual-level situations.Additionally, we add to the literature by refining the social dimensions that are addressed with the focal concern of a critical message (i.e., morality vs. competence).

| Criticism of someone's morality versus competence
Morality and competence are two basic social dimensions that people use to form impressions of others and themselves.These two dimensions have been labeled in different ways, but there is a consensus about the content.One dimension refers to task ability (e.g., competence, agency) and the other to interpersonal intentions (e.g., morality, communion, warmth, Abele et al., 2016;Brambilla & Leach, 2014;Fiske et al., 2007).In previous research on intergroup criticism these two dimensions were not made explicit, nor directly compared.For example, Hornsey and colleagues (Hornsey & Imani, 2004;Hornsey et al., 2002) investigated group-directed outgroup criticism such as being called a "racist" or "living in an ivory tower".We argue that being called a racist can be considered criticism of one's morality, whereas living in an ivory tower can be considered criticism of one's competence, and that these types of focal concerns cause different reactions to the criticism.Being (seen as) a moral person is important to people, often even more important than being (perceived as) competent (Brambilla & Leach, 2014;Ellemers et al., 2008;Goodwin et al., 2014;Leach et al., 2007).Specifically, morality is more important than competence when it comes to the groups we want to belong to (Leach et al., 2007), the norms we adhere to (Ellemers et al., 2008), and when we form impressions about the character of people (Brambilla & Leach, 2014;Goodwin et al., 2014).Since being moral is so relevant to people, they typically show highly emotional and threatened responses when light is shed on their moral failures (Täuber et al., 2015(Täuber et al., , 2018;;Täuber & van Zomeren, 2013;van der Lee et al., 2016).We argue that this can result in a self-defensive reaction, for instance by making negative attributions about the person voicing this criticism.A self-defensive reaction (discrediting the source of the criticism) can in turn cause people to ignore concerns raised (Gausel & Leach, 2011;Gausel et al., 2012;Giner-Sorolla, 2012).We extend prior work by examining whether this also prevents people from being motivated to improve their own behavior.When one's competence (vs.morality) is criticized, we predict this will result in less self-defensiveness and more motivation towards self-improvement.This prediction builds on prior research showing that people generally feel more capable of coping with their past shortcomings in the competence domain than with past moral failures (Van der Lee et al., 2016).Furthermore, whereas positive (vs.negative) judgments weigh in more when the competence of another person is judged, negative (vs. positive) judgments are more important when the morality of a person is judged (Martijn et al., 1992;Skowronski & Carlston, 1987).Thus, people might get more defensive and are inclined to hide immoral (vs.incompetent) behavior since the former has a higher potential to hurt their image.Even behavior and motivational processes are affected when potential implications of behavior are framed in terms of one's morality versus competence.In past research, participants were being evaluated on a task that could assess their social bias, but the implications of their performance were either framed in terms of their morality (i.e., their moral values concerning egalitarianism) or their competence (i.e., their ability to quickly learn new tasks, Van Nunspeet et al., 2014).Participants showed less negative social bias towards a target group in the morality (vs.competence) condition.
This behavioral effect was complemented by an effect in the same direction for cognitive processes related to unconscious response monitoring (i.e., measured with EEG) which are indicators for the motivation to give the correct responses on a task.Participants were more motivated to respond correctly when they were evaluated on their morality (vs.competence).In other words, in a situation that has a moral frame, people are more concerned about their moral image.This might make people more defensive towards criticism of their morality (vs.competence).
Our research differs from previous studies contrasting morality and competence criticism (Täuber et al., 2018;Täuber & van Zomeren, 2013) by having the criticism delivered by ingroup and outgroup members (i.e., as in the literature on the intergroup sensitivity effect) rather than by a neutral third party.Moreover, the current research investigated cognitive attributions people make about (the motives of) critics and how these impact on their motivation to improve their own behavior, rather than capturing the emotion of "moral outrage" directed against an outgroup that is portrayed as displaying superior morality (Täuber & Van Zomeren, 2013).

| Current research
The reviewed findings suggest that criticism of one's competence (vs.morality) might make people less defensive and more motivated to improve.In turn, this might make people more often accept criticism and change their behavior, even if the criticism comes from the outgroup.
To test this, we designed two studies in which we included the critic's group-membership (i.e., ingroup vs. outgroup) and dimension of criticism (i.e., morality, competence).Balancing the trade-off between higher power and controlling for individual differences (e.g., a priori sensitivity to either ingroup vs. outgroup or morality vs. competence criticism) in within-participant designs and the feasibility of people having experienced a situation for all our four conditions, we varied the design between the two studies.In Study 1, the critic's group-membership was included as a within-participant factor (dimension as a between-factor), since we aimed to test whether the intergroup sensitivity effect also applies in situations where individual behavior is criticized by an outgroup member.Here, we also aimed at getting a full understanding of the underlying (i.e., emotional, motivational, and cognitive) processes in individual-level criticism situations.In Study 2, our main aim was to test whether competence (vs.morality) criticism is a way to decrease defensiveness and increase effectiveness of outgroup criticism (e.g., making people more often change their behavior based on criticism).Thus, we switched the dimension from a between-participants to a within-participant factor (and group-membership to a between-factor).
For both studies, we hypothesized that: Cognitive attributions about the critic (e.g., motives, trustworthiness, credibility) would be more negative for outgroup critics compared to ingroup critics (e.g., Hornsey et al., 2002;Ilgen et al., 1979) and that criticism of the participant's competence, rather than of their morality, would decrease these negative attributions (Martijn et al., 1992;Van der Lee et al., 2016;Van Nunspeet et al., 2014).
Participants would experience a more negative emotional response when being criticized by an outgroup member compared to an ingroup member (Hornsey et al., 2002), and that this emotional reaction would be smaller when criticism concerned their competence rather than their morality (Martijn et al., 1992;Van der Lee et al., 2016;Van Nunspeet et al., 2014).
Participants would be more motivated to improve when criticized by the ingroup (vs.outgroup, Hornsey et al., 2002) and criticism of participants' competence (vs.morality) would make people more motivated to improve (Martijn et al., 1992;Van der Lee et al., 2016;Van Nunspeet et al., 2014).
Criticism would therefore be more effective when given by the ingroup (vs.outgroup) and on participants' competence (vs.morality, see Figure 1).

| Participants and design
A total of 202 participants completed the questionnaire on the online research platform Prolific in exchange for €6.We based our sample size on past (lab) research in which participants recalled evaluations of ingroup members on moral and competence failures (69 participants, Van der Lee et al., 2016) and increased the sample size due to the extended design (i.e., we added a between-participants factor: Critic's group-membership), increased anonymity, and to be able to exclude participants who did not meet attention checks and other controls for serious participation, as usual for online studies.
The study was approved by the local ethics committee.
The study had a 2: Critic's group-membership (in-vs.outgroup, within-participants factor) × 2: Dimension (morality vs. competence, between-participants factor) design.The order of the critic's group-membership (i.e., whether criticism from an ingroup or outgroup member was recalled first) was counterbalanced between participants.Participants answered both quantitative (i.e., closed questions about their experiences being criticized) and qualitative questions (e.g., open-ended questions, in which they could elaborate on their experiences).We focus on the quantitative part in the current article and report on some of the qualitative data for demonstration purposes.
Six participants were excluded from the data analyses because they were unable to recall a situation in which they had received criticism as specified in one of our conditions (i.e., ingroup or outgroup criticism of the participant's morality or competence).The we were able to detect effect sizes as small as η 2 p = 0.04 (small effect) for within-factors using repeated measure analysis of variance (RM ANOVA).
Participants were employed in a broad spectrum of career sectors, holding different types of industry roles, and most participants were from the UK (58.3%, see Supporting Information for more details).

| Procedure
Participants were asked to recall two situations in which they had received criticism at their workplace.Depending on the experimental condition, participants were requested to recall a situation where the criticism had addressed either their competence or morality (i.e., between-participants) and was provided either by an ingroup or an outgroup member (i.e., within-participants).We explained that morality criticism could concern, for example, behavior that was considered to be dishonest, insincere, or untrustworthy, and competence criticism could concern behavior that was considered to be incompetent, unintelligent, or unskilled.To emphasize the group membership of the person providing the criticism, we first explained that "individuals represent different groups, for instance, your work team versus other work teams, colleagues with similar skills or different skills, or individuals who share distinctive features (nationality, education, etc.), or differ from each other in these features" and asked participants to think of different groups at their workplace.
We then instructed participants to think of a specific situation in which they had received criticism from someone at the workplace (excluding their boss/supervisor).Most often mentioned ingroups were colleagues (same position 42%, same work team 32%) and most mentioned outgroups were different work teams (32%) or other departments (27%, also see Supporting Information).

| Measures
To help participants recall the experience, we asked them about the context of the situations, their (negative) attributions about the critic and the criticism, the identification with the group-membership that defined the critic as an ingroup versus outgroup member, their reaction to the criticism (i.e., emotional response, perceived motivation to improve), and asked them to indicate whether they accepted, had the intention to change their behavior based on, and changed their behavior based on, the criticism.Except for guided open-ended questions about the context of the situation recalled, and when otherwise indicated, all items were presented on 7-point scales (1 = not at all, 7 = very much, reliability indices for all measures sufficient, Table 2 in Supporting Information).

Checks
After describing each of the two situations, we asked participants to specify the critic's group-membership (ingroup vs. outgroup) and the primary dimension of the criticism (morality vs. competence).In case participants indicated that the incident they recalled did not meet the criteria, we asked them to think of another situation that did match the experimental condition they were in.Additionally, we asked them to indicate to what extent the criticism referred to their morality (1) or their competence (7, in the actual questionnaire this was labeled "performance") on a 7-point bipolar scale.We used three items to measure group identification (e.g., "I identify with other members of this group") and four items to assess collective self-esteem (e.g., "I feel good about this group") pertaining to the group that defined the critic as an ingroup versus outgroup member.
These items were adapted from the subscales developed by Ellemers et al. (1999).The items could be answered on an 8-point scale, ranging from 1 (not at all) to 8 (very much).
We measured fear of exclusion from the ingroup with seven items (adapted from Cavazza et al., 2014, e.g., "I was afraid that my group would exclude me if they would hear about the criticism") to be able to exclude this as an alternative explanation for our anticipated effects.For other potential confounds, participants were presented with guided questions in which they had to choose between several preset answering options to specify the nature of the situation they had recalled.Participants were asked about their interpersonal relationship with the critic (i.e., "I liked/didn't like this person"), the workrelated status of the critic compared to the participant (i.e., "Lower, Equal, Higher"), the delivery form of the criticism (e.g., "Quick, clear, and friendly"), the circumstances under which the criticism was provided (e.g., "The feedback was offered spontaneously by the other person"), and the time of the event (e.g., "1-3 months ago").

Negative attributions
We measured how participants evaluated the person criticizing them (i.e., the critic) and the critical message itself (i.e., the criticism).We measured perceived intention of the critic (two items, e.g., "I think the person who criticized me did this in my best interest", adapted from Hornsey et al., 2004), trust in the critic (two items, e.g., "I now trust the person"), and credibility of the critic (two items, e.g., "I think the person is credible").To investigate how the message was evaluated by participants, we measured perceived constructiveness of the criticism (two items, adapted from Hornsey et al., 2004, e.g., "The criticism was intended to be constructive"), the credibility of the message (one item, "The criticism was credible"), and perceived threat (two items, e.g., "I perceived the criticism as threatening").
Maximum likelihood factor analysis (MFLA) using varimax rotation revealed that all items loaded on one factor (for excluded items and factor analyses see Data S1), rather than several factors, explaining at least 68.5% (lowest score of the two repeated measures) of the variance with factor loadings ≥0.69.We thus combined the items into a single scale, negative attributions about critic (nine items), and recoded positively framed items so higher scores on this scale indicate more negative attributions.

Emotional response
We measured participants' emotional responses to being criticized with three items from the Positive and Negative Affect Schedule (PANAS, Watson et al., 1988, e.g. "The criticism upset me"), 11 items of more specific emotions related to performance feedback (i.e., achievement success, avoidance failure, adapted from Brockner & Higgins, 2001, e.g., "The criticism made me feel tense") and one item related to perceived control (i.e., "I felt the criticism had to do with something outside of my control, something I am unable to influence").MFLA using varimax rotation yielded a three-factor solution for participants' emotional responses, which together explained 63.9% of the variance in the individual items and was related more to general valence rather than specific emotional responses to the criticism (e.g., avoidance of failure).The first factor comprised negative emotion items (at least 42.4% explained variance with factor loadings ≥0.53), the second factor comprised positive emotion items (at least 16% explained variance, factor loadings ≥0.61), and the third factor comprised the two moral emotions guilt and shame (at least 7.8% explained variance, factor loadings ≥0.70).We thus used three scales to measure the emotional responses to the criticism, negative emotions (eight items), positive emotions (four items), and moral emotions (two items) for our analyses.

Motivation to improve
The motivation to improve based on the criticism was measured with four items (e.g., "The criticism made me feel energized about improving myself", "I felt like I could do something with the criticism").

Effectiveness of criticism
To investigate whether participants accepted and used the criticism to change their behavior, we measured acceptance of criticism (three items, e.g., "I accepted the criticism"), behavioral intention to adapt their behavior based on the criticism (three items, e.g., "I planned to use the criticism to improve"), and actual behavioral change based on the criticism (three items, e.g., "I changed my behavior based on what was criticized").Correlational analysis revealed that all three measures were highly correlated (see Data S1 Table S3).Furthermore, MFLA with varimax rotation revealed that all three measures loaded on one rather than three distinct factors.The one-factor solution explained at least 81.5% of the variance with factor loadings ≥ 0.74.
We thus combined the items to one measure, effectiveness of criticism (nine items).
We tested whether effectiveness was distinctive of motivation to improve by specifying two factors in another MFLA analysis with varimax rotation.Results confirmed this.In the two-factor solution, effectiveness explained at least 73.5% of the variance with factor loadings ≥0.50.Motivation to improve explained at least 6% of the variance with factor loadings ≥0.67.1 Additionally, we specified two factors to test whether effectiveness was distinctive of negative attributions, which was also the case.Effectiveness explained at least 65.4% of the variance, with factor loadings ≥0.62 and negative attributions explained at least 11.6% of the variance with factor loadings ≥0.64.

| Checks
We examined identification with the critic's group-membership with paired sample t tests.As intended, participants reported significantly higher levels of identification with the ingroup (M = 5.61, SD = 1.62) than with the outgroup (M = 3.57, SD = 1.74), t( 195 Thus, our manipulations were successful. We did not find evidence for alternative explanations for our reported effects and criticism situations were comparable (see Data S1).Fear of exclusion from the ingroup, interpersonal liking between the participant and the critic, delivery of the criticism by the critic (e.g., unfriendly), the status of the critic, time of delivery, and context of criticism situations did not vary systematically with our manipulation.Furthermore, criticism concerned participants' (specific) behavior rather than on their (abstract) identity.

| Content coding criticism situations
We content-coded criticism situations participants (N = 392) reported on from one of our open-ended questions (i.e., "First, we would like to ask you to think about the actual criticism.What was the essence of the criticism you received?").We first developed an initial coding scheme with blinded conditions.Then, we formed super-categories including initial codes that were related to each other, separately for the morality and the competence condition.
The coding revealed that criticism situations recalled by participants referred to very common everyday shortcomings people may display in the workplace.The behaviors mentioned most often in the competence condition were related to lack of competence displayed in the "work outcome" (69%), the "work attitude" (11%), and "care in completing work-related tasks" (7%).Examples of work outcomes seen as indicating lack of competence are, for example, "not doing a task good enough" or "working too slow".Work attitudes criticized for lack of competence include behaviors such as "behaving unprofessionally" or "taking too many breaks".Lack of care in completing work-related tasks include behaviors such as "making mistakes" or "forgetting a task".Likewise, the behaviors that were mentioned most often in the morality condition referred to behavior that did not seem to meet moral standards pertaining to "cooperative behavior" (23%), "honesty" (21%), and "work ethic" (18%).Examples of behaviors indicating lack of cooperation are, for example, "letting other people do more work" or "not being part of the team".Examples of failures to meet standards for honesty include behaviors criticized for "not being truthful about having performed a task" or "cheating".Behavior that was criticized as indicating lack of appropriate work ethics includes "accepting gifts from customers" or "not following company guidelines".

| Negative attributions
We predicted that participants would make more negative attributions about outgroup (vs.ingroup) critics.Furthermore, we hypothesized that receiving criticism of one's competence (vs.morality) would decrease negative attributions.To test this, we used a 2: Group-membership Critic (in-vs.outgroup, within-participants factor) × 2: Dimension (morality vs. competence, between-participants factor) mixed RM ANOVA.

| Emotional response
For the participant's emotional responses, we predicted an effect of the critic's group-membership and dimension in the same direction (i.e., a greater negative emotional response to outgroup [vs.ingroup] criticism, lower in case the criticism was about one's competence [vs.morality]).

A mixed RM MANOVA (Critic's group-membership [within],
Dimension [between]) with negative emotions, positive emotions, and moral emotions as dependent factors and using an alpha level of 0.016 (Bonferroni correction, 0.05/3) to correct for multiple comparisons revealed an effect of the critic's group-membership on the emotional response at the multivariate level, Pillai's Trace = 0.07, F(3, 192) = 4.80, p = .003,η 2 p = 0.07.At the univariate level, participants reported stronger negative emotions when the criticism came from the outgroup (M = 4.27, SD = 1.52) as compared to the ingroup

| Mediation
Lastly, we examined mediation effects.We had anticipated that the cognitive (i.e., negative attributions about the critic), the emotional (i.e., affective responses), and the motivational pathway (i.e., the motivation to improve) would mediate the effects of the critic's group-membership and dimension on effectiveness of the criticism.We found direct effects of the critic's group-membership (within-participant factor) for all our DV's (i.e., negative attributions, emotional response, motivation to improve, effectiveness).However, we did not find an effect of dimension on the main dependent variable, effectiveness of the criticism.Therefore, we only tested for mediation of the critic's 2 Significant effects in the same direction were found when using the three initial scales (i.e., acceptance, intention to change, actual behavioral change) independently (acceptance: η 2 p = 0.10, intention to change: η 2 p = 0.04, actual behavioral change: η 2 p = 0.04). 3Interestingly, whereas negative emotions correlated negatively with effectiveness of the criticism, moral emotions (even though also negative in valence) correlated positively with effectiveness.This suggests that reporting more moral emotions was associated with more effectiveness of the criticism (see Data S1 Table S3 [Study 1] and Table S11 [Study 2]).group-membership effect (for indirect effects for dimension see Data S1).We used the SPSS Macro MEMORE (Montoya & Hayes, 2017) for two-condition within-participants mediation.
The relationship between the critic's group-membership and effectiveness of criticism was fully mediated by negative attributions about the critic and motivation to improve (parallel mediation, see Figure 2), with the overall model explaining 71.7% of the variance.The total unstandardized bootstrapped (5,000 samples) indirect effect was significant, 0.78, 95% CI [.45, 1.11].
When adding the emotional responses (i.e., positive, negative, moral) to this mediation model, they did not significantly mediate the relationship between the critic's group-membership and effectiveness of the criticism.

| S TUDY 2
In Study 1, we demonstrated that the intergroup sensitivity effect also applies when individual (vs.group) behavior is criticized.The second goal of the current research was to test whether competence (vs.morality) criticism can increase effectiveness of (outgroup) criticism.We, therefore, switched the within versus between nature of the two main factors in our experimental design.In Study 2 the dimension of the criticism is no longer a between-participants but a within-participant factor (and the critic's group-membership now a between-participants factor).This way, we can control for individual differences and increase statistical power while examining the effects of the dimension of criticism.
A possible alternative explanation for our effect of groupmembership on effectiveness of criticism is that outgroup criticism is perceived as more severe.People usually interact more with ingroup members at the workplace and they might receive more specific and concrete types of criticism from them, which therefore may seem less severe ("I saw you made a mistake on this task") than if it came from outgroup members ("I feel I can't trust you").
To be able to investigate and exclude this alternative explanation, we added a measure to Study 2 to check for perceived severity of the criticism.

| Participants, design, and procedure
A total of 402 participants completed this study on Prolific in exchange for €6.Participants (M age = 34.5 years, SD = 10.94,266 females, one not indicating sex 4 ) were randomly assigned to two of four conditions of the 2: Dimension (morality vs. competence, within-participants factor) × 2: Critic's group-membership (in-vs.outgroup member, between-participants factor) design.Whether participants were first asked to recall criticism of their morality or competence (or vice versa) was counterbalanced between participants.We excluded 17 participants from data analyses because they could not think of a situation that we asked for.The remaining sample size consisted of 385 participants, 197 participants in the ingroup condition, 188 in the outgroup condition.Sensitivity-analyses in G*power (α = 0.05, 1β = 0.80, N = 385) indicated that we can detect effect sizes as small as η 2 p = 0.02 (small effect) for within-factors using repeated measure analysis of variance (RM ANOVA).The study was approved by the local ethics committee.
The sample was similarly diverse as Study 1 (see Data S1).As in Study 1, experiences with ingroup members mostly referred to people from the same work team (60%) or from the same department (33%).Outgroup experiences mostly indicated people from different departments (47%) and other work teams (35%).We used the same instructions and manipulations as in Study 1. 4 Four participants failed to insert their identification number and are not included here since we could not identify their demographic data; however, we included them in the main analyses.

| Effectiveness of criticism
We tested our prediction whether competence criticism can increase the effectiveness of outgroup criticism with a mixed RM ANOVA (Dimension [within], Critic's group-membership [between]).

| Mediation
We expected that negative attributions and motivation to improve would mediate the effects of the critic's group-membership and dimension on effectiveness of the criticism.We found only marginally significant effects of group-membership on effectiveness and thus only investigated mediation for dimension of criticism.As predicted, the relationship between dimension and effectiveness of criticism was fully mediated by the negative attributions about the critic and the motivation to improve.Similar to our results in Study 1, where emotional responses did not mediate the relationship between the critic's group-membership and effectiveness, emotional responses did not mediate the relationship between dimension and effectiveness.However, moral emotions (i.e., shame, guilt) did mediate the relationship between dimension and effectiveness (parallel mediation, see Figure 3)-with competence (vs.morality) criticism being negatively associated with moral emotions, but moral emotions being positively associated with effectiveness.The full model explained 74.8% of the total variance, with the total unstandardized bootstrapped (5,000 samples) indirect effect being significant, 0.46, 95% CI [.25, 0.67].

| Discussion
Past research on intergroup criticism has focused mostly on groupdirected criticism and the negative motives ingroup members attribute to outgroups criticizing the ingroup (intergroup sensitivity effect, Hornsey & Imani, 2004;Hornsey et al., 2002).With the present research, we aimed to extend previous findings by demonstrating that the intergroup sensitivity effect also applies when individual (vs.group) behavior is criticized by an outgroup (Study 1) and that the focal concern addressed with the criticism (i.e., competence instead of morality) can increase effectiveness of outgroup criticism (Study 2).Additionally, we examined which underlying processes (i.e., emotional, cognitive, and motivational) influence people's reactions to 5 Significant effects in the same direction were found when using the three measures independently (acceptance: η 2 p = 0.06; intention to change: η 2 p = 0.05; actual behavioral change: η 2 p = 0.03).
F I G U R E 3 Mediation model Study 2 with unstandardized regression coefficients (b) depicting the relationship between the within-participant factor "dimension" and "effectiveness of the criticism", as mediated by "moral emotions", "negative attributions about critic", and the "motivation to improve".The coefficient for the direct effect of "dimension" on "effectiveness of criticism" is in parentheses.*p < .05,***p < .001In Study 1, we investigated whether a critic's group-membership and the social dimension addressed with a criticism influence its effectiveness.In line with our predictions, participants made more negative attributions about outgroup critics compared to ingroup critics.Additionally, they reported feeling stronger negative and weaker positive emotions and were less motivated to improve when criticism came from the outgroup.Most importantly, ingroup criticism was more effective: Participants accepted, were more likely to change, and actually changed their behavior more often when being criticized by an ingroup (vs.outgroup) member.Additionally, we found a main effect of dimension on motivation to improve in the hypothesized direction.That is, participants reported more motivation to improve when receiving criticism of their competence (vs.morality).This suggests that, even though criticism from an outgroup member is less effective overall, receiving criticism of one's competence (vs.morality) makes a criticism message more effective.
There was no mediating role of emotional response (e.g., negative emotions) between the critic's group-membership and the effectiveness of the criticism.However, negative attributions and the motivation to improve fully mediated the relationship between the critic's group-membership and effectiveness of the criticism.Thus, the underlying process of why outgroup criticism is perceived as less effective does not seem to be a more intense negative emotional reaction, but a more negative evaluation of the critic (i.e., defensiveness).This finding reveals interesting implications.Whereas an intense negative emotional response of the person criticized might be hard to prevent, it might be easier to stimulate a cognitive reevaluation of the situation.For example, the critic could make his/ her good intentions explicit or emphasize the opportunity to grow from the criticism.Decreasing defensiveness towards outgroups criticizing is especially needed considering recent research showing that defensiveness can translate into behavioral reactions, for example defending the ingroup or showing hostile actions towards the outgroup (Thürmer & McCrea, 2018;Thürmer et al., 2019).
In Study 2 we switched dimension from a between-to a withinparticipants factor to control for individual differences.We were able to show strong support for our prediction that criticism of participants' competence (vs.morality) reduces participants' negative attributions about the critic, increases motivation to improve, and is overall more effective (e.g., participants reported having changed their behavior more often).Furthermore, participants reported stronger moral emotions in the morality compared to the competence condition, but not stronger negative emotions.Thus, even though they were not statistically tested here, we can speculate that both types of criticism elicit a similar negative emotional response.Competence criticism, on the other hand, made participants feel stronger positive emotions (e.g., feeling confident or happy) and more motivated to improve (replicating Study 1).Our hypothesized mediation effects were also confirmed.The effect of dimension on effectiveness of criticism was mediated by negative attributions about the critic and motivation to improve (not by negative or positive emotions).
Competence criticism decreased negative attributions about the critic and increased motivation to improve, which, in turn, made the criticism more effective (i.e., more reported behavioral change).The combined results of both studies suggest that competence criticism is more effective than morality criticism, not because it lessens negative affect, but because it changes how people interpret the situation.This finding is promising and offers a relatively easy alternative approach for outgroup critics who aim at evoking behavioral change in the people they are criticizing.To illustrate, a critic could choose to focus the focal concern of a criticism message on the recipient's moral character (e.g., not being loyal to other coworkers by slacking at work) or on situational performance (e.g., being lazy).According to our findings, the latter might evoke more behavioral change by making the recipient less defensive and more motivated to improve.
Please note that moral emotions also mediated the relationship between dimension and effectiveness.People reported stronger moral emotions in the morality (vs.competence) condition (in line with Van der Lee et al., 2016) and stronger moral emotions led to more effectiveness.Thus, in contrast to more motivation to improve and fewer negative attributions in the competence condition, which led to more effectiveness, fewer moral emotions in the competence condition inhibited part of this effectiveness.This might indicate that, in the morality condition, participants felt pressured to change their behavior by their guilt or shame (e.g., stress response).
However, more importantly, in the competence condition, driving the overall effect of dimension on effectiveness, participants made fewer negative attributions about the critic and might have interpreted the criticism more as an opportunity to grow.
We replicated the effect of the critic's group-membership on negative attributions about the critic that we found in Study 1, but not for motivation to improve and effectiveness (only marginally significant).However, we had anticipated that the power for the critic's group-membership might not be sufficient to detect the effects of the between-participants factor in this type of research (i.e., recalling autobiographical events).Since we investigated autobiographical situations there are some limitations to our research.Firstly, we only have self-report measures of effectiveness of criticism.A better approach to this would be to investigate actual behavioral change based on criticism, for example by experimentally manipulating the dimensions.Furthermore, we did not ask participants to reflect on evaluations of moral and competence failures but asked them to reflect on criticism which could have been given unjustly.However, we did this intentionally since people might engage in coping mechanisms (e.g., avoidance, covering up mistakes), restricting the number of situations people can think of as well as narrowing the range of reported criticism situations (e.g., only very small or insignificant failures of morality or competence).Secondly, research has shown that people are often not able to predict how they actually emotionally respond in a situation (Wilson & Gilbert, 2003) or to correctly report on past events (Shiffman et al., 1997).However, even in the absence of these emotional responses, we find consistent effects relating to the cognitive evaluation (i.e., making negative attributions about a source) and motivational tendencies (having seen the ability to improve based on the criticism).In theory, these cognitive responses might have been triggered by immediate emotional responses, and we cannot exclude the possibility that this was the case.Importantly, this does not distract from the fact that the way people interpreted the comments they received-depending on the group-membership of the source and the focal dimension-predicted the effectiveness of the criticism.Future research might expand on this finding by also capturing immediate emotional responses in order to explore whether and how these might contribute to the cognitive effects we observed or have additional independent effects.
Thirdly, we specifically asked participants to reflect on criticism at their workplace-where competence, as well as morality, are relevant dimensions for behavioral judgment.One could argue that this context might highlight the relevance of competence over the morality dimension.However, there is a large body of research from social cognition suggesting that morality generally dominates in person perception and evaluation (Brambilla & Leach, 2014;Goodwin et al., 2014), as well as research on organizations suggesting that morality is also highly relevant (often even more than competence and sociability) at the workplace and in task performance contexts (Ellemers et al., 2011;Leach et al., 2007;Van Prooijen & Ellemers, 2015).Nevertheless, future research might examine whether these effects generalize to other contexts.Perhaps people are generally less concerned, and hence also less responsive, to competence-related criticism in a context where task performance is less relevant-for instance when in a group of friends or a family setting.
A potential confound for our effects of group-membership on our dependent variables could be related to the delivery of the criticism as we found that ingroup (vs.outgroup) criticism was more often (perceived to be) delivered in a "friendly and clear" manner.However, when submitting delivery of criticism as a covariate to our main analyses, the effect of group-membership remained (see Data S1).
Another potential confound for our effects of dimension (i.e., morality vs. competence) on effectiveness of criticism could be the concreteness/abstractness of the criticism that was delivered (Moscatelli et al., 2019).Competence criticism could, for example, be more about concrete behaviors (e.g., working faster), whereas morality criticism could be more abstract (e.g., not being trustworthy in general).To check for this possibility, we investigated whether the criticism referred to participants' concrete behavior or their abstract identity (see Data S1).This investigation confirmed that most of the criticism situations reported by participants were about concrete behaviors (e.g., not working fast enough), rather than about abstract identity (e.g., not being a trustworthy person).Further, we checked and confirmed that the observed effects of criticism dimension hold when controlling for abstractness/concreteness by adding it as a covariate to our analyses.We therefore conclude that it is unlikely that abstractness of criticism received confounds our results.
In conclusion, we show that even in individual-level criticism situations, people make more negative attributions about the critic and they see less motivation to improve when they are criticized by an outgroup member.This, in turn, negatively affects whether people change their behavior based on criticism.However, regardless of this group-membership effect, we demonstrate that being criticized on one's competence, compared to one's morality, makes people less defensive towards the criticism, more motivated to improve, and finally, makes them change their behavior more often in line with the criticism.
Please note that we do not claim that all moral failures can be framed as addressing someone's competence (or should be).Rather, we argue that in situations where both competence and moral aspects of problematic behavior can be made the focal point addressed with a criticism message (e.g., lying about having performed a work task while actually having failed to do so), focusing on lack of competence (failing to do the work task) versus lack of morality (lying about having performed the work task) displayed is more likely to get the critic "a foot in the door" towards implementing change, by making people less defensive.For example, because people are especially motivated to be seen as moral, critics often frame the primary focus of a criticism message as being about someone's morality in order to increase motivation to change.However, this can make people perceive the task of changing their behavior as too difficult and threatening (Van der Lee et al., 2016), which is why they may more easily "give up" this task and get defensive or disengage (Bandura, 1982;Gausel & Leach, 2011;Täuber et al., 2015).Thus, emphasizing the moral dimension of criticism often backfires.In comparison, we show in the current research that the competence dimension of criticism can offer people an opportunity to grow and motivate them to change their own behavior.
Taken together, our findings suggest that criticizing someone's competence rather than someone's morality can make people more open to criticism, even when this criticism comes from the outgroup.
Making people more susceptible to outgroup criticism would be very beneficial in all kinds of intergroup situations, for example, in a political discussion, where people with different political standpoints negotiate agreements that will apply to the whole nation.An open ear to other people's criticism might result in a better outcome for all.
remaining sample consisted of 196 participants, 103 in the competence condition, 93 in the morality condition (M age = 35.5 years, SD = 9.71, 101 females, one not indicating sex).Sensitivityanalyses in G*power (α = 0.05, 1β = 0.80, N = 196) indicated that F I G U R E 1 Conceptual model of hypothesized underlying mechanisms of the effect of group-membership and dimension on the effectiveness of criticism ) = 13.57,p < .001.Analysis of collective self-esteem yielded very similar results (see Data S1).Additionally, we compared the means from both the ingroup and outgroup condition on our bipolar dimension manipulation check (1 = morality, 7 = competence) to the midpoint of the scale (4) with one-sample t tests.Participants in the morality condition reported to have experienced morality criticism (M = 3.14, SD = 1.34), t(92) = −6.17,p < .001.Participants in the competence condition reported to have experienced competence criticism (M = 5.44, SD = 1.33), t(102) = 10.94,p < .001.Additionally, both means were significantly different from each other, t(194) = 12.00, p < .001.
either an ingroup or outgroup member and on either their morality or competence.

| 599
DON'T TELL ME ABOUT MY MORAL FAILURES BUT MOTIVATE ME TO IMPROVE: INCREASING EFFECTIVENESS OF OUTGROUP CRITICISM BY CRITICIZING ONE'S COMPETENCE