The Harmful Side of Thanks: Thankful Responses to High-Power Group Help Undermine Low-Power Groups' Protest.

Giving thanks has multiple psychological benefits. However, within intergroup contexts, thankful responses from low-power to high-power group members could solidify the power hierarchy. The other-oriented nature of grateful expressions could mask power differences and discourage low-power group members from advocating for their ingroup interests. In five studies (N = 825), we examine the novel idea of a potentially harmful side of "thanks," using correlational and experimental designs and a follow-up. Across different contexts, expressing thanks to a high-power group member who transgressed and then helped undermined low-power group members' protest intentions and actual protest. Thus, the expression of thanks can pacify members of low-power groups. We offer insights into the underlying process by showing that forgiveness of the high-power benefactor and system justification mediate this effect. Our findings provide evidence for a problematic side of gratitude within intergroup relations. We discuss social implications.

How can you thank a man for giving you what’s already yours? How then can you thank him for giving you only part of what’s already yours?

Not a single day goes by without us expressing thanks. Saying “thank you” seems indisputably positive, universal, and multifunctional. Among others, it can constitute an expression of appreciation of someone else’s investment in our well-being, an act of courtesy, or simply reciprocation (Carr, 2015; Watkins, 2014). Psychological research has largely documented the intra- and inter-personal benefits of giving thanks. In a nutshell, it makes us feel better and brings us closer together (Watkins, 2014; Wood, Froh, & Geraghty, 2010). However, lately, there has been rising doubt in the psychological and philosophical literature that expressions of thanks, or gratitude, are always beneficial (Carr, 2015; Eibach, Wilmot, & Libby, 2015; Wood et al., 2016). Expressions of thanks in abusive relationships (Wood et al., 2016) or for benefits provided by otherwise exploitative institutions (Eibach et al., 2015; for example, welfare capitalism) are noted illustrations of how giving thanks might be misplaced. It has been argued that in contexts of unequal power, the reciprocal and harmonious nature of giving thanks could backfire for those holding lesser power in that it could unintentionally perpetuate their dependency on a controlling benefactor. In spite of these considerations, research has largely neglected the study of possible negative effects of expressing thanks.

To our knowledge, the present research is the first to empirically examine the harmful consequences of expressing thanks within an intergroup context marked by unequal power relations. We approach this investigation by combining gratitude literature with the literature on the problematic effects of intergroup contact and helping for the disadvantaged. Our analysis centers on the question of whether expressions of gratitude by low-power group members for favors given to them by the high-power group can demobilize low-power groups to challenge the status quo. In addition, we investigate the underlying psychological processes. In focusing on everyday behavior, our empirical account of harmful effects of expressing thanks in the intergroup context helps us to understand how and why disadvantaged groups often tolerate their unjust social standing (e.g., Jost, Banaji, & Nosek, 2004).

The Normativity of Thanks

Expressions of thanks can yield beneficial psychological outcomes both in the giver and recipient of thanks (for an overview, see Watkins, 2014). Giving thanks can increase happiness and decrease depressive symptoms (Seligman, Steen, Park, & Peterson, 2005), strengthen social bonds (Algoe, Fredrickson, & Gable, 2013), and motivate recipients of thanks to show prosocial behavior (Grant & Gino, 2010).

Within the research above, giving thanks is largely considered a behavioral outcome of gratitude, which is a positive and other-oriented moral emotion (cf. Watkins, 2014). However, people do not have to experience gratitude when saying “thank you” (Visser, 2009). This becomes apparent when observing the difficulty young children have in expressing thanks, despite persistent prompting from their parents (Greif & Gleason, 1980). Thanks-saying not only transmits authentic gratitude but is also socially expected (Mills, 2005). For one, expressions of thanks are prescribed to enable social reciprocation. According to politeness theory (Brown & Levinson, 1978/1987), expressing thanks signals acceptance of a gift or a favor. This limits the beneficiary’s freedom of action because it implies that they have taken on debt and will have to reciprocate. Therefore, giving thanks can cause indebtedness (Watkins, Scheer, Ovnicek, & Kolts, 2006), which can, in turn, be dissolved through the beneficiary paying off their debt with another benefit.

Second, failure to express thanks is socially undesirable. While displays of thankfulness when receiving benefits are encouraged, displays of dissatisfaction, no matter how unattractive the benefit is, signify rudeness and moral defect (Carr, 2015; Eibach et al., 2015). This might explain why people are more likely to express thanks when there is an audience (Baumeister & Ilko, 1995).

In contexts of inequality, gratitude norms can be especially restrictive. A German federal minister labeled refugees who left shelters because of unbearable conditions as “ungrateful” and demanded “a culture of adaptation” (“De Maizière wirft Flüchtlingen,” 2015). In the United States, with the Civil Rights achievements of the 1960s, many Whites viewed racism as eroded and Black activists who continued fighting for racial equality were stereotyped as “ungrateful” (T. J. Davis, 2016, p. 220). Gratitude norms can accordingly be applied to justify restrictions on disadvantaged groups’ autonomy. Empirical research underlines that failure to express thanks to a high-power benefactor is a punishable offense: Insecure high-power holders denigrated low-power holders more when they were not thanked than when they received thanks (Cho & Fast, 2012). Therefore, members of low-power groups could be motivated to express thanks to even unfair high-power benefactors to escape penalties. Situational demands to express gratitude are prevalent on a daily basis through gratitude norms which of course also affect disadvantaged group members. But above that, disadvantaged group members face demands that target them in particular—such as when women and LGBTI (lesbian, gay, bisexual, transgender, and intersex) are encouraged to be grateful for their rights because they could “have it worse” (Klein, 2018; Valenti, 2014).

In the context of social inequality, communicating gratitude could negatively affect members of disadvantaged groups: Findings on the other-oriented nature of gratitude show that being grateful encourages yielding to the benefactor (Watkins et al., 2006), fosters cooperation, and increases forgiveness (McCullough, Emmons, & Tsang, 2002). Joined with the imposition to reciprocate a benefit, expressing thanks could lead low-power group members to put their resentments aside and demonstrate their appreciation through engaging in behavior that may be contrary to their group’s interests.

In line with our reasoning, Eibach and colleagues (2015) argue that gratitude norms motivate system justification—the rationalization of unfair sociopolitical arrangements (Jost et al., 2004). In an extensive literature review, the authors propose (but do not empirically test) that interpersonal gratitude norms overgeneralize to a system level and oblige citizens to display gratitude for benefits provided by sociopolitical institutions through statements that approve of the system, and by refraining from voicing dissent about its injustices.

We extend this reasoning to the intergroup level and test these ideas with empirical data. We predict that when members of low-power groups express thanks for help from high-power groups, they will self-censor their criticism. Because system justification comes at the expense of status improvement for the low-power group, which could potentially be achieved through protest, the status quo remains unchallenged, and thus, solidified. The empirical test of this pacifying effect of expressing thanks is the core objective of our research. Next, we illustrate how two intergroup theories lead to our reasoning.

The Pacifying Effect of Thanks

Research shows that social hierarchies have been stabilized not only through intergroup conflict and hostile practices by groups in power, but through collaborative intergroup relations as well (Dixon, Levine, Reicher, & Durrheim, 2012; Halabi & Nadler, 2010; Jackman, 1994; Jost et al., 2004; Wright & Lubensky, 2009). Theories of intergroup contact and intergroup helping have captured how intergroup hierarchies can be upheld under an appearance of benevolence and fairness.

Intergroup Contact

Mounting evidence points to the status-maintaining side effects of specific forms of positive intergroup contact for disadvantaged groups (cf. Dixon et al., 2012; Dixon, Tropp, Durrheim, & Tredoux, 2010; Wright & Lubensky, 2009). It has been found that positive contact with the respective advantaged group can be accompanied with members of disadvantaged groups’ lower support for governmental measures aimed toward social change (Dixon, Durrheim, & Tredoux, 2007; Sengupta & Sibley, 2013), lower efforts to engage in collective action on behalf of one’s group (Becker, Wright, Lubensky, & Zhou, 2013; Wright & Lubensky, 2009), and decreased awareness for inequality and discrimination (Saguy, Tausch, Dovidio, & Pratto, 2009; Sengupta & Sibley, 2013). In that they create an illusion of harmony (Saguy et al., 2009), and positive encounters with advantaged group members, which leave power relations unaddressed (Becker et al., 2013), can undermine disadvantaged groups’ efforts toward social change. We suggest that the pacifying effect of thanks might be one of the mechanisms explaining these findings. Within positive intergroup encounters, disadvantaged group members might feel grateful because the advantaged group members behave kindly, and thus feel motivated to express their gratitude. Integrating the literature on intergroup contact with the literature on the benefits of expressing gratitude, we can expect that thankful responses for benefits from high-power group members might create a harmonious atmosphere and divert members of low-power groups from self-serving protest to other-oriented, cooperative reciprocation.

Intergroup Helping

Exchanges of help between groups can also be generally viewed as positive. Yet, besides its caring and redistributive nature, helping can create a power disparity (Halabi & Nadler, 2010): While helping is associated with independence and competence, receipt of help signals dependence and inferiority. Helping can therefore sometimes serve more the helper’s needs, at the expense of the one helped (van Leeuwen & Täuber, 2010). Helping relations can hence be viewed as unequal relations, which is especially insidious when they occur between groups of socially unequal status. With more resources at hand, advantaged groups can provide help to disadvantaged groups as a means to maintain dominance and foster cooperation while upholding an image of generosity (Halabi & Nadler, 2010). As an illustration, it has been found that when status relations were identity threatening for members of a high-power group, they increasingly provided help to members of a low-power group to protect their group’s superiority (Nadler, Harpaz-Gorodeisky, & Ben-David, 2009). At the same time, recipients can be punished when they reject help (Rosen, Mickler, & Collins, 1987), and this still holds when the help is patronizing (Becker, Glick, Ilic, & Bohner, 2011; Wang, Silverman, Gwinn, & Dovidio, 2015). It is therefore not unlikely that members of disadvantaged groups express thanks for even patronizing help to escape negative consequences.

It follows that when low-power groups thank high-power groups for help, they might signal acceptance and, figuratively, agree to dependency (Nadler et al., 2009). This could result in reciprocal behavior that feeds the interests of the high-power group, for example, in the form of self-censorship of protest (Eibach et al., 2015).

From Thanks to Silence: Mediators

Which psychological processes may underlie the relation between expressions of thanks and the curbing of protest? We propose that expressions of thanks function as acts of forgiveness of the transgressions that elicit protest. By transgressions, we mean single discriminatory actions by high-power group members or social inequality in a broad sense, as a set of chronic, structural transgressions. Benefits provided by a high-power group member could represent compensation, and expressing thanks could imply its acceptance and communicate forgiveness of the former transgression.

Previous research suggests that forgiveness is a crucial step for the restoration of justice in victims of interpersonal or intergroup transgressions. Forgiveness ameliorates the symbolic threats caused by transgressions, such as the fact that the transgressor illegitimately harmed and disempowered the victim. Accepting compensation alone does not resolve concerns around status and power (Wenzel & Okimoto, 2010). Forgiveness, however, helps to restore power because the victim can determine their own and the transgressor’s moral image (Shnabel & Nadler, 2015). Moreover, forgiveness indicates morality, and by forgiving, victims can (temporarily) elevate their status (Wenzel & Okimoto, 2010, 2015). Communicating forgiveness through expressions of thanks might induce the perception among low-power group members that power differences are straightened and justice has been restored. If there is justice, there should be no need to protest. Consequently, we predict that forgiveness mediates the inhibiting effect of thanking on protest. This assumption is informed by research which found that gratitude and forgiveness were positively associated (McCullough et al., 2002) and that forgiving caused a sense of power and reduced perceptions of injustice in victims and members of low-power groups, increasing their willingness to reconcile (Wenzel & Okimoto, 2010, 2015)

In addition, we propose that this perception of justice reflects in system justification because it creates the impression among low-power group members that “everyone benefits” from the interaction, and therefore, the system seems fair (Jost & Kay, 2005). If the system seems fair, protest will be less likely. In support of our reasoning, previous research found that balancing of group disadvantage with ostensibly positive group stereotypes or benefits has a palliative effect on members of disadvantaged groups’ protest (Becker & Wright, 2011; Jost & Kay, 2005). Because the sense of justice is contingent on forgiveness, as illustrated above, we propose system justification as a subsequent mediator to forgiveness.

The Present Research

Our research program delivers the first empirical test of the harmful effects of expressions of thanks in an intergroup context that is characterized by power relations. We integrate the literature on interpersonal benefits of expressing thanks with problematic effects of positive intergroup contact and help for low-power groups while following recent theorizing on the system-justifying function of gratitude norms (Eibach et al., 2015). We outlined that members of disadvantaged groups are encouraged to express thanks for benefits from members of advantaged groups through restrictive gratitude norms and intra- and inter-personal benefits of expressing thanks. While helping can perpetuate the high-power groups’ dominance, acceptance of help through the expression of thanks can affirm the low-power groups’ subordination. The seeming benevolence of helping and the harmonizing, reciprocal nature of thankful responses should lower perceptions of inequality. Forgiveness and justification of the status relations should move members of low-power groups away from voicing dissatisfaction.

In five studies, we tested the hypothesis that expressing thanks to a member of a high-power group for their help undermines low-power group members’ protest against them (H1, Studies 1-4). We expected that the negative effect of the expression of thanks on protest intentions is mediated by forgiveness (H2, Studies 2a, 2b, and 3) and system justification (H3, Study 3).[1]

Study 1

First, we conducted a conservative test of our main hypothesis. We designed Study 1 to test whether expressions of thanks to a high-power group member would inhibit low-power group members’ protest in a minimal group-type manner before extending our findings to more naturalistic settings. In a laboratory experiment, we induced an intergroup context and assigned participants to a low-power group position (employee) in a simulated organizational scenario.

Method

Participants

Participants were recruited on campuses of two German universities. The final sample consisted of 95 participants (63 women, 32 men; Mage = 23.07, SDage = 2.95, 98.9% German).[2] Because this was a lengthy lab study and we tested a novel effect, we aimed to recruit at least 50 participants per cell. Using G*Power (Faul, Erdfelder, Lang, & Buchner, 2007), we determined with a sensitivity analysis that to reach 80% power, a sample of N = 95 would require an effect of gHedges = 0.58.

Design and procedure

Detailed information on the method and analyses can be found in the supplemental online material (SOM). Participants were supposedly working with another employee and a manager (high-power group member) through the computer. All members of the team had to individually complete problem-solving tasks, and the difficulties of the tasks were assigned by the manager. For every solved task, participants would collect tickets for a lottery where they could win 5€. The manager behaved unfairly by assigning all easy tasks to himself, whereas the employees failed to complete the difficult tasks. Then, the manager contacted the participant and offered help, saying that he will give them more easy tasks in the next round.

Participants were then randomly assigned to either the experimental condition (n = 53), in which they could choose between three rated expressions of thanks (“Thank you,” “Thank you very much,” “Great, thank you very much”) or the control condition (n = 42) where they could choose between neutral responses (“I have received the message,” “I have read the message,” “I have received the information”). We next measured with nine items how willing participants will be to protest against the manager on behalf of the employees (e.g., “demanding that the manager hands over lottery tickets to the employees”). In addition, we measured whether participants will directly confront the manager in a message.

Results and Discussion

Results showed that the induction of an intergroup context and relative power, as well as the manipulation, were effective. In line with expectations, protest intentions were lower for participants who expressed thanks (M = 3.89, SD = 1.24) than for participants who did not express thanks (M = 4.35, SD = 0.97), t(92.99) = 2.03, p = .045, 95% confidence interval (CI) [0.01, 0.91], gHedges = 0.42, 95% CI [0.01, 0.83]. Moreover, expressions of thanks indirectly affected protest behavior through protest intentions (b = −0.38, SE = 0.25, 95% CI [–1.06, −0.04]).

Study 1 provides the first evidence that expressing thanks to a high-power group member for their help inhibits low-power group members’ intentions to protest on behalf of their group. This study tested our hypotheses strictly because participants in the experimental condition had to express thanks, which might have caused reactance. However, we chose this manipulation to circumvent self-selection effects. We expected that the sedating effect will be even stronger when participants express thanks voluntarily, as in Study 2.

Study 2

Study 2 is divided into Study 2a and its conceptual replication with increased power, Study 2b. We tested whether voluntary expressions of thanks to a member of a high-power group would be negatively associated with members of low-power groups’ protest intentions and extended our investigation to forgiveness as a mediator.

In an online vignette study, students (low-power group members) imagined interacting with a professor (high-power group member) who transgressed and then helped. We then assessed whether participants expressed thanks, forgave the professor, and how much they were willing to protest.

Study 2a

Method

Participants

Participants were recruited through student mailing lists and postings on bulletin boards of two German universities, in exchange for taking part in a voucher raffle. The final sample consisted of 125 students (81 women, 41 men, one other, two not indicated; Mage = 22.82, SDage = 2.58, 93.6% German; see Note 2). An a priori power analysis (power ≥ 80%, α = .05, two-tailed) using a medium-sized effect size showed that we needed 128 participants.

Procedure

The study ostensibly assessed how students cope with feedback, which they would receive for presentations in seminars. Participants read a vignette about a student’s experience and imagined being a part of the scenario. In this scenario, they had intensively prepared for a very important group presentation with two other students. In a meeting, their professor advised the group on how to make their presentation even better. Although the students followed his recommendation, the professor graded them much lower than expected, apparently because of the changes they had made. When the group reminded the professor that these exact changes were his idea, he said that he strongly doubted that and could not remember giving that advice. However, he gave them a slightly higher grade because he knew of its impact on their bachelor’s degrees. This formed our operationalization of help. Participants could then choose the one out of two answers which they would most likely give to the professor in that situation (expression of thanks, see below).

Moving on in the scenario, participants read that outside one student from their group said that the professor acted unfairly and suggested not letting his behavior pass without comment. The other student said that the professor’s behavior was appropriate and that they do not want to take any action against him. Both students asked for the participants’ opinion. At this point, we administered our protest intentions measure. Next, participants were asked to explain their stated behavioral intentions, supposedly based on statements provided by previous participants. The statements contained the forgiveness measure embedded in distractor items. Specifically, forgiveness referred to excusing the professor’s failure to fully compensate the students for the fact that it was his advice that had put them at a disadvantage. Participants then responded to a power perception check and demographic questions. They were debriefed and compensated.

Measures

Expression of thanks

Participants could pick one out of two answers to the professor, aligned side by side in a randomized order, which either contained an expression of thanks (in italics) or not: “You respond: ‘Yes, this grade is very important for the Bachelor’s degree. So thank you very much for your favor!’ and say goodbye: ‘Bye, see you next week.’”

Forgiveness

Participants indicated forgiveness on two items (“I forgive the professor, no matter whether his behavior was right or wrong” and “I excuse the professor’s behavior,” r = .68, p < .001; 1 = not at all to 7 = very much).

Protest intentions

Protest intentions were assessed with eight items adapted from Study 1 (1 = would definitely not participate in, 7 = would definitely participate in). Actions were, for example, “calling out the professor together with the presentation group” or “jointly complaining about the professor to the student council” (α = .80; for the complete scale, see the SOM). We adapted the power perception check from Study 1.

Results

Power perception check

Participants attributed more power to the professor (M = 4.36, SD = 0.97) than to themselves, M = 1.70, SD = 0.66, t(124) = 29.91, p < .001.

Expression of thanks and protest intentions

Forty-two (33.6%) participants expressed thanks, while 83 participants did not express thanks. An independent-samples t test revealed a significant effect of thanking on protest intentions, t(123) = 2.04, p = .043, 95% CI [0.01, 0.86], gHedges = 0.39, 95% CI [0.01, 0.76]. In line with expectations, protest intentions were lower for participants who expressed thanks (M = 3.78, SD = 0.94) than for participants who did not express thanks (M = 4.22, SD = 1.22).

Mediation

To examine the indirect effect of the expression of thanks on protest intentions through forgiveness, we conducted a mediation analysis.[3] As expected, expressing thanks positively predicted forgiveness (b = 0.85, SE = 0.23, p < .001, 95% CI [0.39, 1.30]), which in turn negatively predicted protest intentions, approaching significance (b = −0.15, SE = 0.08, p = .075, 95% CI [−0.32, 0.02]. The mediation was confirmed by a significant indirect effect (b = −0.13, SE = 0.08, 95% CI [−0.33, −0.01]). The direct effect was not significant (b = −0.31, SE = 0.22, p = .168, 95% CI [−0.75, 0.13]).

Study 2b

Participants were recruited on social media platforms and at three German universities, in exchange for course credit or participation in a voucher raffle. The final sample consisted of 264 students (187 women, 74 men, three participants did not indicate their gender; Mage = 22.86, SDage = 3.84, 95.5% German; see Note 2). An a priori power analysis (power ≥ 80%, α = .05, two-tailed) using the effect size found in Study 2a (gHedges = 0.39) showed that we needed a sample of 210 participants.

The procedure was identical to Study 2a except that, now, we measured forgiveness before protest intentions.

Measures

Measures were identical to Study 2a, except for an additional item in forgiveness (“I forgive the professor,” α = .85). Reliability for protest intentions was α = .81.

Participants attributed more power to the professor (M = 4.35, SD = 0.74) than to themselves, M = 1.89, SD = 0.77, t(259) = 34.72, p < .001. Four participants had missing values.

Sixty-four (24.2 %) participants expressed thanks, while 200 participants did not express thanks. An independent-samples t test revealed a significant effect of the expression of thanks on protest intentions, t(262) = 2.40, p = .017, 95% CI [0.07, 0.75], gHedges = 0.41, 95% CI [0.13, 0.69]. Again, protest intentions were lower for participants who expressed thanks (M = 3.62, SD = 1.34) than for participants who did not express thanks (M = 4.03, SD = 1.14).

As expected, expressing thanks positively predicted forgiveness (b = 1.04, SE = .18, p < .001, 95% CI [0.68, 1.40]), which in turn negatively predicted protest intentions (b = −0.28, SE = 0.06, p < .001, 95% CI [−0.39, −0.17]). The mediation was confirmed by a significant indirect effect (b = −0.29, SE = 0.08, 95% CI [−0.47, −0.16]). The direct effect was nonsignificant (b = −0.12, SE = 0.17, p = .225, 95% CI [−0.46, 0.22]).

Discussion

We confirmed our hypotheses in two different samples: Expressing thanks to a high-power helper was associated with lower willingness to protest against unjust treatment among low-power group members. In line with our assumptions, forgiveness mediated this relation: thanking increased forgiveness, which in turn reduced protest intentions. It could be argued that this evidence is correlational and that the effect is driven by third variables, for instance, agreeableness.[4] To address potential self-selection bias regarding the effect of thanking on forgiveness, we conducted another experiment. Finally, we extended our investigation to a socially disadvantaged group.

Study 3

In Study 3, we tested our hypotheses in an online experiment with social groups of relative power: All participants were women (as members of a socially disadvantaged group) and imagined an interaction with a male colleague (as member of a socially privileged group), who behaved in a sexist way. We manipulated whether participants thanked him for subsequent help and measured how that affected forgiveness and protest intentions. In addition, we tested system justification as a subsequent mediator to forgiveness.

The final sample consisted of 248 female MTurk workers who reside in the United States (Mage = 36.33, SDage = 11.14, 98.4% U.S. Americans; see Note 2). In determining sample size a priori (power ≥ 80%, α = .05, two-tailed), we used the effect size found in Study 2b (gHedges = 0.41). The necessary sample size was 190.

Procedure

The study ostensibly examined interactions in the workplace. Participants read about an interaction that someone supposedly experienced at work and were asked to imagine being a part of the scenario. In this scenario, they and a male colleague had completed a very important project into which they had put equal amounts of hard work. Because her contract expires soon, it is very important for the female protagonist (i.e., the participant) to impress her boss when she and her colleague present the results. The next morning, the participant learns that her colleague already talked to the boss because he thinks that “these things are often more effectively communicated between guys, fewer misunderstandings and such.” This sexist remark represented the transgression and was followed by help: “. . . but I made sure to put in a good word for you to help with your contract renewal. I’ve got to run. I’ll be around later.” For the experimental manipulation, participants were randomly assigned to either the experimental condition, in which they imagined expressing thanks: “You say: ‘Thank you so much for putting in a good word for me! Bye,’” or not: “You say: ‘Bye.’” To approximate behavior and strengthen the manipulation, participants were asked to copy the response word for word into a text box. Participants then indicated to what extent they experienced certain thoughts and emotions. The statements, which were supposedly provided by previous participants, contained the forgiveness and gender-work-specific system justification measures embedded in distractor items.

Moving on in the scenario, participants read that they were considering what to do next by thinking about what their two female best friends, who have similar jobs, would do. One friend was portrayed as approving and the other as disapproving of the colleague’s behavior. At this point, we administered the protest intentions measure, which again supposedly consisted of randomly selected statements expressed by previous participants. To increase credibility, participants could add statements.

Participants responded to demographic questions, a manipulation check, and a measure aimed to examine whether participants perceived the colleague’s transgression as an intergroup transgression. Finally, participants were fully debriefed and compensated.

For the complete scales, see the SOM.

Forgiveness was assessed with three items, for example, “I forgive my colleague” (α = .89; 7-point scales, 1 = not at all to 7 = very much).

Gender-work-specific system justification (GWSJ)

We adjusted the gender-specific system justification measure from Jost and Kay (2005) to a work context. Five items measured GWSJ, for example, “In general, work relations between men and women are fair” (α = .88).

Protest intentions were assessed with 11 items adapted from the previous studies (7-point scales, 1 = I would definitely not engage in, 7 = I would definitely engage in). Actions were, for example, “calling out my colleague” or “complaining about my colleague to the women’s representative” (α = .83).

Manipulation check

As a manipulation check, we administered the item “In the described interaction, did you thank your colleague for his help?” to which participants could respond “yes” or “no.”

Intergroup check

Participants indicated how strongly (1 = not at all to 7 = very much) they perceived their colleague’s transgression within the interaction as “an interpersonal transgression” and “a gender-related transgression.”

Manipulation check

A χ2 test showed a significant association between the condition and the expression of thanks-indicator variable, confirming the different nature of the two conditions, χ2(1) = 158.43, p < .001.

Intergroup check

A paired-samples t test showed that the colleague’s transgression was significantly more perceived as a gender-related transgression (M = 5.51, SD = 1.84) than an interpersonal transgression, M = 4.58, SD = 1.85, t(247) = −5.51, p < .001.

Table 1 displays means, standard deviations, and correlations among study variables.

Table 1.

Study 3: Descriptive Statistics and Intercorrelations (Pearson’s r).

Variables	No thanks, M (SD)	Thanks, M (SD)	1	2	3
1. Expression of thanks[a]			—
2. Forgiveness	2.34 (1.38)	2.77 (1.53)	.15*	—
3. Gender-work-specific system justification	3.79 (1.87)	3.69 (1.61)	–.03	.13*	—
4. Protest intentions	4.24 (1.11)	4.20 (1.24)	–.02	−.42**	−.37**

Coded 0 = no thanks and 1 = thanks.

p < .05. **p < .01.

Expression of thanks and protest intentions

The two experimental conditions were coded 0 = no thanks (n = 127) and 1 = thanks (n = 121). Against expectations, an independent-samples t test showed no significant effect of the expression of thanks on protest intentions, t(246) = 0.30, p = .763, 95% CI [−0.25, 0.34].

Mediation

To examine the indirect effect of the expression of thanks on protest intentions through forgiveness, we conducted a mediation analysis. As expected, expressing thanks positively predicted forgiveness (b = 0.43, SE = 0.18, p = .021, 95% CI [0.06, 0.79]), which in turn negatively predicted protest intentions (b = −0.34, SE = 0.05, p < .001, 95% CI [−0.43, −0.25]). The mediation was confirmed by a significant indirect effect (b = −0.15, SE = 0.07, 95% CI [−0.31, −0.02]). The direct effect was nonsignificant (b = 0.10, SE = 0.14, p = .467, 95% CI [−0.17, 0.37]).

Next, we included GWSJ as a subsequent mediator to forgiveness (see Figure 1). The indirect effect was significant (b = −0.01, SE = 0.01, 95% CI [−0.05, −0.001]; for detailed results, see the SOM). Expressing thanks positively predicted forgiveness, which in turn positively predicted GWSJ, which negatively predicted protest intentions. The direct effect was not significant (b = 0.07, SE = 0.13, p = .615, 95% CI [−0.19, 0.32]).[5]

Figure 1.

Serial mediation model tested in Study 3.

*p < .05. ***p < .001.

Post hoc analysis

An explorative analysis revealed that a large number of participants perceived the sexist transgression as very unfair (69.8% had a mean value of 1 on a 1-7 scale), and the sample mean was M = 1.57 (SD = 0.07). Compared with the value distributions of fairness in the other studies, which detected significant main effects, the scenario in Study 3 probably was too “unfair” to detect the main effect. We explored this assumption post hoc.

The interaction between the expression of thanks and fairness perception approached significance (b = −0.23, SD = 0.12, p = .057, 95% CI [−0.463, 0.007], f2 = 0.01). Using the Johnson–Neyman technique (see Hayes, 2013), we found that the undermining effect of thanks expression on protest intentions was in fact significant for fairness values above 4.46 (at 95.5th percentile: b = −0.76, SD = .39, p = .049, 95% CI [–1.663, −0.004]), and not significant for values of 4.46 and below (at 4.5th percentile: b = 0.06, SD = .16, p = .714, 95% CI [−0.249, 0.363]).

In sum, we found support for the sequential mediational process: Expressing thanks motivated forgiveness, which enhanced system justification, which undermined protest intentions. Although we did not find the main effect of expressions of thanks on protest intentions, we illustrate that this was probably due to a floor effect in perceptions of fairness: The pacifying effect occurred only for those with values above the midpoint of the fairness scale. We can speculate that a more normally distributed perception of fairness might have revealed the main effect. Therefore, in our next study, we chose a transgression, which would not be perceived as completely unfair by the majority of the sample.

Study 4

So far, we have shown that expressions of thanks undermined protest intentions when the idea of expressing thanks was raised by us. Hence, it is unclear whether disadvantaged group members would express thanks spontaneously in the context of unequal treatment and whether these natural expressions of thanks would inhibit protest intentions. Second, we wanted to test whether the pacifying effect would translate to real-life situations and behaviors and, finally, whether it is more than just a short-lasting effect. To address these limitations of ecological validity, we conducted Study 4.

To connect our findings to a richer context, we examined the pacifying effect for naturalistic expressions of thanks in an experiential context using real protest behavior. To show that the pacifying effect of thanks is not a fleeting phenomenon, we included a follow-up.[6]

Participants were recruited on the campus of a German university in exchange for course credit. The final sample consisted of 93 female undergraduate psychology students (Mage = 22.12, SDage = 3.44, 96.8% German; see Note 2).

Because this was a lengthy lab study for which we explicitly needed undergraduate psychology students, we aimed to recruit at least 30 participants per cell. A sensitivity analysis showed that to reach 80% power, a sample of N = 93 would require an effect of gHedges = 0.61.

The study was advertised to undergraduate psychology students as supposedly assessing evaluations of the previous application procedure for research assistants at our department. While participating in the study, students could also apply for a research assistant position at our lab (and we have contacted those students who were interested in the position). Two independent evaluators ostensibly preselected candidates based on their task performance and mutual evaluations among the participants.

As in Study 3, we targeted women as the lower power group because we could recruit female students feasibly without having to reveal that gender is the targeted dimension. Participants came into the lab and were supposedly interacting with another female and male student, who were apparently in the adjoining rooms, and would communicate with the participant through the computer. During a task, in which participants suggested improvements to the procedure, the male participant made a sexist remark. He suggested a quota for male research assistants because psychology was female dominated and men were beneficial to psychology, given that they won all the Nobel prizes. This indicated the high-power group members’ transgression.

In a later task, he wrote in a chat that he will be a gentleman and transfer his course credits to the “ladies” because he did not need them and was only participating to apply for the research assistant position. This formed the high-power group members’ help. Participants saw that the female student expressed thanks for the male student’s help. This was meant to incorporate naturalistic demands to increase the salience of gratitude norms.

Then, we experimentally manipulated the opportunity to express thanks: Two thirds of participants were given a line of communication in response to that chat where they could write to the group. The other third could not respond to the chat. We used a 2:1 ratio because we expected that some participants would not express thanks although they could. Thus, we had three conditions: (a) expressing thanks when there is an opportunity to express thanks, (b) not expressing thanks when there is an opportunity to express thanks, and (c) no opportunity to express thanks.

Afterward, participants could evaluate the other two participants regarding their suitability as research assistants. These statements included protest behavior measures. Next, participants could write a message to the other students, which assessed another protest behavior. Supposedly, participants’ statements were then directly sent to one of the independent evaluators. Because the male student was interested in applying for the research assistant position, whether participants protested or not mattered in real life and supposedly jeopardized the success of his application.

Participants were contacted again about 1 week later and responded to the same protest measures online (Mdays = 6.81, SDdays = 1.45). To justify the follow-up, we told participants that their statements will be sent to the other evaluator to guarantee an independent preselection. After the follow-up, participants were fully debriefed and compensated (details of the procedure can be found in the SOM).

Protest behaviors

Participants indicated protest behavior by responding “yes” (1 = protest) or “no” (0 = no protest) to whether they (a) argue against or (b) veto the male student getting the position, (c) protest or (d) recommend (reversed) that he is nominated, or (e) want to file a complaint against him. Participants could (f) write out the complaint and (g) directly confront the male student in a message. Two raters who were blind to the hypotheses coded the open format answers. Interrater reliability was κcomplaintT1 = 1.00, κconfrontationT1 = 0.97, κcomplaintT2 = 1.00, κconfrontationT2 = 0.86 (all ps < .001). Discrepancies were resolved by the third independent rater. All seven indicators were averaged to a scale with higher scores indicating stronger engagement in protest behavior (αT1 = .79, αT2 = .75).

Out of those participants who had the opportunity to express thanks, 35 (58%) expressed thanks spontaneously, while 26 did not. Thirty-two participants had no opportunity to express thanks. There was no significant difference in protest levels among those who could not thank and those who did not express thanks although they could at both measurement times. Therefore, we collapsed the two control conditions into one to increase test power (n = 58), tT1(56) = −0.16, p = .871, 95% CI [−0.16, 0.14]; tT2(56) = 0.98, p = .333, 95% CI [−0.07, 0.20].[7]

The expression of thanks was negatively associated with protest behaviors shown immediately at T1 (r = –.34, p = .001) and in the follow-up at T2 (r = –.28, p = .007). Protest at T1 was positively associated with protest at T2 (r = .75, p < .001).

Expression of thanks and protest behavior

Welch’s t test showed a significant effect of expression of thanks on protest behavior for both measurement points, tT1(90.50) = 3.98, p < .001, 95% CI [0.09, 0.27], gHedgesT1 = 0.74, 95% CI [0.31, 1.18]; tT2(88.89) = 3.18, p = .002, 95% CI [0.05, 0.20], gHedgesT2 = 0.61, 95% CI [0.18, 1.04]. In line with expectations, participants who expressed thanks showed less protest behavior (MT1 = 0.09, SDT1 = 0.16; MT2 = 0.10, SDT2 = 0.13) than participants who did not express thanks (MT1 = 0.27, SDT1 = 0.28; MT2 = 0.23, SDT2 = 0.25).[8]

In sum, Study 4 extended the generalizability of our findings. We replicated the pacifying effect of thanks in a real-life context beyond hypothetical scenarios and behavioral intentions, targeting participants’ real social identities as women and students. Importantly, we found the effect for naturally occurring expressions of thanks, while limiting self-selection effects, experimental demands, and reactance. Moreover, our results show that the pacifying effect persists 1 week later, suggesting that it affects real and on-going relationships as well as.

General Discussion

The present work pioneers research on the harmful side of expressing thanks. Our investigation indicates that the positive act of thanking can be problematic within an intergroup context marked by social injustice. Across five studies, we provide direct evidence for a pacifying effect of “thanks” on members of low-power groups. Specifically, we show that expressions of thanks for benefits provided by high-power groups can directly, or indirectly, undermine their efforts to challenge the intergroup hierarchy. Results of Studies 2 and 3 further highlight that the underlying processes are forgiveness of the high-power group member’s transgression and system justification.

Taken together, our research program provides a cumulative understanding of the pacifying effect of gratitude expressions. We tested and overall supported our hypotheses in multiple and heterogeneous ways: in the lab and online, with correlational data and different experimental manipulations, a within-study replication, for behavioral and longitudinal data, and across different contexts. This speaks to the generality of our findings. Finally, we showed that the pacifying effect of thanks affects real-life protest. With this, our findings question the universal appropriateness and benefits of expressing thanks and identify it as an everyday mechanism through which members of low-power groups might be unintentionally endangering improvement of their status position.

The Power of “Thanks”

Our findings emphasize the power of expressions of thanks and show how their restrictiveness manifests in the expressers’ behavior. In support of prior theorizing (Lambert, Clark, Durtschi, Fincham, & Graham, 2010), the sole act of thanking not only communicated acceptance of help to the helper but also the self. Studies 1 and 3 especially back this interpretation, because they yielded the pacifying effect even when the expression of thanks was not voluntary. This implies that the act of thanking seems to override levels of gratitude or willingness to thank.

Second, our findings place the expression of thanks as a determinant of reciprocal, cooperative behavior within intergroup interactions: Thanking participants “paid it back” through mitigating a potential threat to the high-power helper’s power position. We provide experimental evidence showing that this effect is not due to self-selection based on individual characteristics. Expressions of thanks are consequential, even in the absence of gratitude. This speaks to their normative nature as outlined in the introduction.

Pacification Through Forgiveness and System Justification

We found that forgiveness of the high-power benefactor explained the decrease in low-power group members’ protest intentions observed after thanks were expressed. For example, women who thanked a sexist colleague for subsequent help forgave him and were less willing to stand up for themselves (and other women), even though his help was patronizing.

Expressing thanks and thereby granting forgiveness could be a way for low-power group members to cope with inequality or situations where protest is dangerous or costly, for example in abusive relationships (Wood et al., 2016). However, the experience of situational power via forgiveness might reinforce the translation of the thankful stance into a hierarchy-supporting belief system. For instance, some conservative Christian women who were abused by their husbands reported that they viewed forgiveness as the duty of a good Christian wife (Nash, Faulkner, & Abell, 2013). Such narratives should encourage attention to behaviors within power contexts which might promote the representation of the low-power group through complementary stereotypes: A representation which crosses a low status in a power dimension with a high status in a moral dimension (here: “powerless but grateful/forgiving”) has been found to increase system justification (Kay & Jost, 2003). Indeed, we found support for this compensatory effect: Expressing thanks signaled forgiveness, which triggered the perception among low-power group members that unequal status relations are fair. With this, we also provide the first empirical evidence that expressions of thanks can have a system-justifying function (Eibach et al., 2015). Thus, although the sense of justice and fairness should positively influence well-being (e.g., Reis, Sheldon, Gable, Roscoe, & Ryan, 2000), our research suggests that it could also mask the structural injustice between the helper and the one helped. This diverts from protest, which bears the potential to improve sociopolitical conditions for the low-power group in the long run.

Instead of providing benefits to secure forgiveness, high-power transgressors could choose to apologize. Although apologies can also result in forgiveness, when they contain certain elements (Kirchhoff, Wagner, & Strack, 2012), they might be more beneficial for members of low-power groups than help. By explicitly acknowledging their wrongdoing, transgressors commit to a consensus of values with the victims and this could increase chances that the transgressor will not transgress again. However, if transgressors do not want to give up transgressing at the expense of their power, they might choose to instead provide benefits to pacify victims, while appearing generous (Okimoto, Wenzel, & Hedrick, 2013).

Theoretical and Practical Implications

Within psychological literature, the problematic role of gratitude or its expression has not yet been studied. Thus, the most novel contribution of our work is that it shows that gratitude and expressions of thanks, which are concepts almost everyone perceives in positive terms, can have negative effects. With this, we provide direct evidence for the claim that the “positivity” of positive psychological phenomena cannot be established independent of context (McNulty & Fincham, 2012). Moreover, our research not only connects but also advances the literature on intergroup contact, intergroup helping, and gratitude.

First, our work has implications for intergroup contact research, because it suggests a mechanism that could be central in explaining the demobilizing effect of positive intergroup contact on members of disadvantaged groups. Positive contact with the advantaged group could pacify disadvantaged group members because they might be feeling grateful for the advantaged groups’ kindness. This idea has neither been considered nor studied within intergroup contact research.

Our work further advances research on intergroup helping by empirically laying the bridge to protest research. We go beyond the effects of receiving help for low-power groups and show that expressing thanks for it can inhibit protest on behalf of the ingroup and (unintentionally) signify agreement to dependency.

Third, we contribute to answering the call voiced by gratitude researchers for empirical evidence for harmful effects of gratitude (expression) which might help to (a) explain null effects or negative effects on well-being–related outcomes of gratitude interventions and (b) identify situations in which gratitude (expression) can be harmful (see D. E. Davis et al., 2015; Wood et al., 2016). Based on our findings, the major implication for gratitude research and exercises in clinical contexts and mindfulness practice is to acknowledge the relative power difference between the target of thankfulness and the expresser of thanks. Encouraging members of socially disadvantaged groups to be grateful for what they have has many positive consequences for the individual (cf. Wood et al., 2010) but may derail from perceiving structural inequality and depress entitlement to just treatment. Researchers and practitioners could also attend to the restrictiveness of gratitude norms which members of low-power groups face and the status-reinforcing character of certain benefits.

Finally, the present research has implications for social change. Within societies, which privilege certain social groups above others, it seems crucial that members of disadvantaged groups advocate for their group if they aim to achieve status improvement. Our findings suggest that expressing gratitude toward those who are already privileged instead silences those holding lesser power and encourages cooperation with the high-power group. This is problematic given that members of disadvantaged groups are already discouraged from communicating anger about discrimination (e.g., Kaiser & Miller, 2001). In addition, through the receipt of thanks, members of high-power groups might feel affirmed and remain unchallenged in providing help, which boosts their dominance and maintains the social hierarchy.

How can members of disadvantaged groups escape the pacifying effect of expressions of thanks? Our research shows that advocacy for their ingroup was higher when thanks were not expressed. This does not imply that members of disadvantaged groups should stop thanking. Displaying gratitude is a kind and considerate act, which can improve well-being and enrich social interactions. Members of disadvantaged groups should not be denied these benefits. Withholding thanks might lessen non-harmful help from advantaged groups. Members of disadvantaged groups seem to face a dilemma: While thanking the advantaged group might inhibit advocacy for their ingroup, not thanking might deprive them of individual benefits and attach to them the stigma of ingratitude.

However, withstanding demands to express gratitude at least poses an opportunity for resistance for low-power group members. Thus, when, for example, users of soup kitchens do not express thanks for food, this should not be judged as a sign of “attitude” (Stein, 1989) but could be seen as an attempt to preserve and communicate a critical stance on the differences in privilege between volunteers and users. The present research suggests that it might be protective for members of low-power groups at times to display “ingratitude” to avoid self-censorship, boycott unwanted assistance, or resist dependency (Eibach et al., 2015). This argumentation parallels with the reasoning that, although it might yet seem antithetic to the tradition in health psychology, “psychological . . . discomfort can be psychologically and politically healthy” (Allen & Leach, 2018, p. 332).

Indeed, former immigrants from relatively poorer countries have started to publicly resist continuous demands to be grateful for their citizenship, criticizing that these demands prescribe submissiveness and otherness (e.g., Gorelik, 2012; Nayeri, 2017). Thanking advantaged groups for being granted rights, which they naturally enjoy, could encourage perceptions that equal rights for the disadvantaged are gifts and not a natural course of action (Eibach et al., 2015).

Limitations and Future Research

In our research, we implemented direct transgressions to stimulate protest in a study setting. Therefore, one limitation could be that the forgiveness process is specific to the context of advantaged-group transgressions because without a transgression there is nothing to forgive. However, this should neither imply that this specific process is rare or that the pacifying effect of thanks is contingent on proximal transgressions. First, intergroup relations between groups of unequal power are often marked by transgressions, because high-power groups chronically possess more resources, social rights, and status than low-power groups (Sidanius & Pratto, 1999). Transgressions are not always direct and tied to a specific transgressor, like in our studies. They can be subtle or structural, but they are part of hierarchical intergroup contexts (e.g., Autin & Butera, 2016; Fiske, Dupree, Nicolas, & Swencionic, 2016). Of course, not all advantaged group members transgress and many advocate for social equality and support disadvantaged group members’ protests. Nevertheless, we can infer from the intergroup helping literature that the mere existence of social inequality or unequal power relations should be sufficient to find pacifying effects of gratitude. Help or benefits can appear as compensation for disadvantage as long as there is relative power between the groups involved in the helping act. For example, one way how advantaged group members can sustain their high-power position is by providing benefits or help to disadvantaged group members (for an overview, see van Leeuwen & Täuber, 2010). We would therefore also expect the pacifying effect of thanks in cases where help is not preceded by a direct transgression, and this could be tested in the future.

Another interesting future research question is whether receiving thanks from high-power groups would have a pacifying effect on members of low-power groups. People in high-power positions evaluate generous acts from people with less power more cynically and ascribe instrumental intentions to them, reducing power holders’ desire to reciprocate (Inesi, Gruenfeld, & Galinsky, 2012). Thus, advantaged group members might prefer expressing gratitude when the generous acts are not threatening to the power hierarchy, such as service or unpaid labor. Thanking the disadvantaged for their services could constitute a paternalistic appeasement gesture to ensure the latter’s loyalty (Jackman, 1994). Receipt of thanks increases feelings of being socially valuable, which in turn prompts prosocial behavior (Grant & Gino, 2010). Therefore, disadvantaged groups might feel pleased that they are needed and loyally avoid protest.

Finally, the type of gratitude expressions that we study should not be confused with a strategic type of gratitude expression, by which, through calculating deliberation, members of disadvantaged groups might attempt to advance their individual status. This can be classified as an individual mobility strategy, which does not address changing the unfair conditions of existing power relations (Ellemers, 2001). Moreover, our findings show that the pacifying effect persists over time and that those who did not protest immediately also did not protest later.

Conclusion

When social groups are deprived of rights and resources and put into a state of disadvantage and need, they might be led to believe that anything advantaged groups offer beyond hostility is a gift for which one should be thankful. Our research shows that members of low-power groups voluntarily thank for benefits that reinforce their weakness. Moreover, the communication of thankfulness can unknowingly prevent them from voicing dissent with unjust treatment and undermine their involvement in status-advancing measures. This implies that, in practice, it is essential to stimulate a critical reflection of gratitude norms and a redefinition of appropriate situations when to express thanks. As previous research suggests, giving thanks entails many benefits and is the grease in the wheels of socioeconomic interactions. Therefore, we do not prescribe to stop thanking altogether but encourage low-power groups to be more cautious of the power context and the type of help before expressing thanks. By focusing on everyday behavior, we hope to propose an accessible opportunity for members of disadvantaged groups to regain control. Our findings support what the opening quote by civil rights activist Malcolm X implies: refusing to say “thank you” can in itself be a form of protest.

Click here for additional data file.

Supplemental material, Ksenofontov_OnlineAppendix for The Harmful Side of Thanks: Thankful Responses to High-Power Group Help Undermine Low-Power Groups’ Protest by Inna Ksenofontov and Julia C. Becker in Personality and Social Psychology Bulletin

Click here for additional data file.

Supplemental material, The_Harmful_Side_of_Thanks_Supplemental_Material_CORRECTED for The Harmful Side of Thanks: Thankful Responses to High-Power Group Help Undermine Low-Power Groups’ Protest by Inna Ksenofontov and Julia C. Becker in Personality and Social Psychology Bulletin