The stereotype rub-off effect – Organizational stereotypes modulate behavioural expectations, expectancy violation and punishment after transgressions

Transgressions committed by employees of non-profit (vs. of for-profit) organizations seem to be judged more harshly by the public. This research studies the underlying process of this relationship. We show that organizational stereotypes of morality and warmth “ rub-off ” from organizations to individuals affiliated with them (Study 1, N = 297). We show that stereotypes of morality and warmth predict expected communal sharing and market pricing behaviour. (Study 2, N = 300). Next, we identify downstream effects of this stereotype rub-off effect in case of transgressions. We show that communal sharing expectations elicit greater perceived expectancy violation and consequently higher punishment when employees commit transgressions (Study 3, N = 402). In sum, as a result of high perceived morality and warmth and subsequent expectations of communal sharing, transgressions of employees affiliated with non-profit organizations prompt increased expectancy violation in observers, leading to harsher punishment. Our findings have important implications for public relations management of non-profit organizations.


Introduction
In February 2018, managers of the non-profit organization Oxfam were accused of organizing sex parties in Haiti in 2010.Allegedly, eight Oxfam employees including the country director had frequently used the services of local Haitian sex workers.While other organizations had experienced similar sex scandals without major public outcries, the public's response towards Oxfam was particularly harsh.The scandal received extensive media attention, former supporters boycotted Oxfam's products, and plummeting donations brought the organization to the verge of collapse.The organization took action and the employees in question were investigated and eventually fired; however major reputational damage had already been done.This example shows that when employees commit transgressions, even outside the work environment, the public's reaction appears to vary as a function of the type of organization employees are affiliated with.
This paper aims to answer the question why transgressions committed by employees affiliated with non-profit organizations seemingly elicit sharper reactions from the public than comparable transgressions committed by employees affiliated with for-profit organizations, even if they transgress outside the organizational context.Drawing on previous literature on organizational stereotypes (Aaker, Vohs, & Mogilner, 2010, Cuddy, Glick, & Beninger, 2011;Johnson, 2019), relational models (Fiske, Haslam, & Fiske, 1991;Haslam, 1995;Haslam, 2004;Haslam & Fiske, 1999) and expectancy violation (Burgoon & Hale, 1988;Burgoon, 2015;Purkiss, Perrewé, Gillespie, Mayes, & Ferris, 2006), we identify the underlying process explaining the public's differential reactions to transgressions of employees affiliated with non-profit and for-profit organizations.Specifically, previous research shows that the public stereotypes non-profit organizations as predominantly moral, while for-profit organizations are perceived as predominantly competent (Aaker et al., 2010).We propose that people extend the stereotypes they hold towards organizations to individual employees affiliated with them, meaning that employees of non-profit organizations are also perceived as moral and warm in daily life outside the organization's context, while employees of for-profit organizations are seen as competent.We label this phenomenon the "stereotype rub-off effect".
Building upon literature regarding relational models (Fiske et al., 1991;Haslam, 1995;Haslam, 2004;Haslam & Fiske, 1999), we propose that fundamental attributions of who a person is, in terms of morality, warmth and competence, prompt different expectations regarding how this person should behave in their relationships with others.Specifically, we propose that perceptions of morality and warmth drive expected communal sharing (CS) behaviour and perceptions of competence drive expected market pricing (MP) behaviour.
In case of employee transgressions, such as in the sex scandal reviewed in the introduction, expectations of market pricing and communal sharing are violated.Interestingly, previous research shows that, in most cultures, violations of communal sharing principles elicits stronger moral outrage than violations of market pricing principles (Fiske & Tetlock, 1997).Consequently, we propose that, since individuals affiliated with non-profit organizations are seen as more moral and warmer due to the stereotype rub-off effect, the public expects communal sharing behaviour from them and therefore responds more harshly to transgressions, showing greater intent to punish transgressors.
In a series of studies, we show that organizational stereotypes of morality, warmth and competence "rub-off" on individual employees of organizations.The public attributes morality, warmth and competence based on organizational affiliation, such that employees of non-profit organizations are perceived as strongly more moral, warm, and as slightly more competent than their for-profit counterparts.Indeed, we find that the public expects predominantly moral and warm individuals to act in the spirit of communal sharing, while predominantly competent individuals are expected to act in line with market pricing principles.Finally, we examine the downstream effects of the stereotype rub-off effect when individual employees of an organization commit transgressions.We show that transgressions violate the public's expectations regarding communal sharing behaviour, but not market pricing behaviour.Importantly, perceptions of morality and warmth, expected communal sharing behaviour and expectancy violation serially mediate the relationship between organizational affiliation and punishment.As a consequence, observers of the same transgression experience significantly greater expectancy violation when the transgression is committed by employees affiliated with non-profit as opposed to for-profit organizations, which results in harsher punishment.
The present research contributes to the literature of organizational stereotypes (Aaker et al., 2010;Cuddy et al., 2011;Johnson, 2019) and expectancy violation (Burgoon, 2015;Purkiss et al., 2006;Rim, Park, & Song, 2018) in three ways.First, we examine the cognitive processes underlying the public's reactions to transgressions and find that organizational affiliation modulates organizational stereotypes of warmth, morality and competence, which then affect expected relational models.Second, we contribute to expectancy violation literature by identifying the origins of observers' expectations in the organizational context, namely organizational stereotypes and expected communal sharing or market pricing behaviour.Third, we demonstrate the consequences of expectancy violation for the relationship between the public and employees of non-profit organizations.We show that, due to different expectations, the public judges transgressions more harshly when they are committed by employees of non-profit compared to for-profit organizations.
Taken together, our research provides strong support for the proposition that organizational affiliation (non-profit vs. for-profit) is an important factor modulating observers' expectations and their intent to punish transgressions committed by employees.We discuss additional factors that might fulfil a signalling function affecting observers' expectations from others and subsequent punishment of transgressions.We discuss theoretical and practical implications, as well as limitations and promising routes for future research.

Organizational stereotypes
In general, stereotypes are "widely held assumptions about certain types of people that are represented cognitively as extensive, wellorganized categories or schemata" (Andersen, Klatzky, & Murray, 1990;Gunia & Levine, 2019).While stereotypes allow for a quick and intuitive assessment of other individuals and groups, they can also cause distorted judgement and biased behaviour (Bordalo, Coffman, Gennaioli, & Shleifer, 2016).Importantly, stereotypes are also formed about the main dimensions people use to evaluate their social surroundings.
In order to evaluate their social surroundings, people resort to two fundamental dimensions, which have been differently labelled by prior stereotype research as agency and communality (Conway, Pizzamiglio, & Mount, 1996), competence and warmth (Cuddy, Fiske, & Glick, 2007), or competence and morality (Abele & Wojciszke, 2014;Wojciszke, 1994Wojciszke, , 2005)).A particular stream in the stereotyping literature primarily distinguishes between the dimensions of competence and warmth (Cuddy et al., 2007;Fiske, Cuddy, Glick, & Xu, 2018), however, this categorization has been criticized for omitting important details about stereotypes against certain groups and thus not capturing stereotyping adequately (Leach, Bilali, & Pagliaro, 2015;Leach, Minescu, Poppe, & Hagendoorn, 2008).A recently emerged perspective suggests morality over warmth as the dominant driver of impression formation, but also highlights the important role of warmth as complementary information besides moral character (Goodwin, Piazza, & Rozin, 2014;Goodwin, 2015).Specifically, research suggests that morality is relevant in order to assess the intentions of others towards oneself, while warmth is important as it predicts the success of a person to recruit friends and allies to support their intentions, thereby providing complementary information rather than replacing morality as a dimension (Brambilla, Rusconi, Sacchi, & Cherubini, 2011;Goodwin et al., 2014, Goodwin, 2015).It is not yet clear whether the observed dominant role of morality over warmth in interpersonal impression formation also holds in the organizational context of our research.Hence, we follow the line of reasoning of Goodwin (2015) and consequently measure and analyse all three dimensions to systematically examine potential differences between the morality and warmth perceptions on subsequent expectations of behaviour.
Prior research suggests that consumers judge organizations along the same basic dimensions of morality, warmth, and competence as they judge other people (Aaker, Garbinsky, & Vohs, 2012;Aaker et al., 2010).Whereas non-profit organizations are stereotyped predominantly as warm (Leach, Ellemers, & Barreto, 2007), for-profit organizations are commonly perceived as predominantly competent (Aaker et al., 2010).Generally, attributions of morality, warmth and competence are not mutually exclusive; instead of representing two ends of the same continuum, people attribute both morality (or warmth) and competence to others (Wojciszke, 2005).However, since stereotypes describe widely held, but oversimplified images of people or things, the dimensions are often considered orthogonal when evaluating out-groups (Yzerbyt, Kervyn, & Judd, 2008;Kervyn, Yzerbyt, & Judd, 2010).
In this paper, we propose that the extent to which organizations are stereotypically perceived as moral and warm versus competent also affects the public's stereotypes about individual employees affiliated with the organization.Specifically, we expect that employees of non-profit organizations will be perceived as more moral and warm individuals, while employees of for-profit organizations will be perceived as more competent individuals due to their respective organizations' stereotypes.In other words, we propose that organizational stereotypes of morality, warmth and competence not only apply to organizations as a whole, but "rub off" on individual employees affiliated with the organization.We develop the following hypotheses: H1a: Affiliation with non-profit organizations is positively related to perceptions ofmorality.H1b: Affiliation with non-profit organizations is positively related to perceptions ofwarmth.H1c: Affiliation with for-profit organizations is positively related to perceptions ofcompetence.
In addition, we propose that rubbed-off stereotypes of warmth, morality and competence affect expectations regarding the expected behaviour of the stereotyped individual.In other words, social stereotypes in the fundamental dimensions of person perception shape expectations on how the stereotyped individual should act.Specifically, we draw on relational models theory to suggest that strong attributions of morality, warmth, and competence are linked to expectations that others will act in accordance with two distinct relational models, communal sharing and market pricing, which we discuss below.

Relational models theory
Relational Models Theory (Fiske et al., 1991;Fiske, 1992) posits that people organize their social relationships with others using four underlying models, namely, communal sharing, market pricing, equality matching, and authority ranking.Relational models provide rules and norms which serve as guidelines for socially acceptable behavior in relationships with others.The theory has received extensive theoretical and empirical validation (e.g., Fiske and Haslam, 1997;Haslam, 2004;Haslam & Fiske, 1999) and has seen a wide range of applications across disciplines such as sociology, social psychology and organizational studies (O'Regan & Choe, 2019;Schoenherr & Dopko, 2019;Wellman, 2017).In communal sharing (CS) relationships, individuals split their social surroundings in in-and out-group without further distinctions and people give and take as they need within the in-group.In contrast, in market pricing (MP) relationships, individuals compare entities by using a single utility metric, mostly money.Previous research conceptualizes relational models as "the implicit social grammar of interpersonal relationships" rather than "explicit folk understanding" (Haslam, 2004), meaning that people categorize their relationships unknowingly.However, when asked, they have no trouble matching their own personal relationships to the proposed dimensions of the model (Haslam & Fiske, 1992).
Prior research successfully applied relational models theory to the organizational context by focusing on the CS and MP dimensions of the model, describing relations between leaders and followers (Game, 2008), and organizations and the public (McGraw & Tetlock, 2005;McGraw, Schwartz, & Tetlock, 2012).Results regarding the organization-public relationship show that the public expects most organizations to maximize profit and thus act in the spirit of market pricing, exchanging goods and services for money.However, organizations, which operate in communality-based domains, such as religion or healthcare, were found to be exempt to this rule: the public expects them to prioritize communal obligations over market pricing goals (McGraw et al., 2012).
Building on these insights, we propose that, since non-profit organizations are seen as warm and pursue moral goals, such as eradication of world hunger or saving the environment, and as a consequence their employees are seen as moral and warm, the public expects them to be individuals, which should generally act in the spirit of CS in both their work life as well as in private life.Now the question becomes relevant whether stereotypes of morality and warmth differ in their effects on expected communal sharing behaviour.We consider expected communal sharing behaviour as expectations of generalized pro-social behaviour, which we expect to be driven by perceptions of both morality and warmth due to the theoretical and conceptual closeness.For example, one would expect an employee of the non-profit organization "Save the Children" to be perceived as a person who would act consistent with CS principles by metaphorically "giving the shirt off their back" for others, because their organizational affiliation prompts stereotyping this person as moral and warm.By contrast, one would expect an employee of the for-profit organization "The Coca-Cola Company" to be perceived as a person that would act consistent with MP principles, for example by maximizing personal gain in relationships with others, because their organizational affiliation prompts stereotyping this person as competent.
H2a: Perceived morality is positively related to expected communal sharing behavior.H2b: Perceived warmth is positively related to expected communal sharing behavior.H2c: Perceived competence is positively related to expected market pricing behavior.
Relational models set expectations of how organizations and people working for those organizations should act.Accordingly, these expectations can inadvertently be violated by peoples' actual behaviour, for example in the form of transgressions.Interestingly, in most cultures, violations of communal sharing models elicit a sharper reaction than the violation of market pricing models and lead to increased moral outrage following the violation (Fiske & Tetlock, 1997).In line with this, we suggest that the public will judge transgressions committed in daily life, and thus outside organizational responsibilities, more harshly when the transgressor is affiliated with a non-profit compared to a for-profit organization, as the former face higher expectations to act according to communal sharing principles.We propose that the perceived difference between expected and observed behaviour will be greater for employees of non-profit organizations, or in other words, the same transgression will prompt different levels of expectancy violation, depending on who committed it and the public's corresponding expectations towards them.These propositions align with Expectation Violation Theory, which we introduce below.

Expectancy violation theory
Expectancy Violation Theory (EVT) is a communication theory concerned with expectations that people have regarding others (Burgoon & Hale, 1988;Burgoon, 2015;Purkiss et al., 2006).Burgoon and Hale (1988) define expectations as enduring cognitions from one party (A) in a relationship to the other party (B) regarding the behavior that is expected from B in the relationship.Similarly, B will also have expectations regarding the behavior of A. In the present research, we apply EVT to support our proposition regarding observers' different responses to transgressions committed by employees affiliated with non-profit versus for-profit organizations.
Expectations can be considered products of both social norms and additional information that one party has about the other (Burgoon & Walther, 1990), such as organizational affiliation.For strangers, social norms and personal characteristics such as culture or gender, and relationship factors such as status or trust are taken into account to form expectations about others (Burgoon, 2015).Importantly, EVT proposes that expectation violation causes arousal and subsequent cognitive assessment of the violation.In the current research, this means that when the observed behavior of the employee does not match observers' expectations, this results in experiences of expectancy violation, which trigger cognitive and behavioral reactions.
The difference between the expectations of the observer and the actual behavior of the observee may be positive when expectations are exceeded and negative when others' behavior falls short of what was expected (Burgoon & Hale, 1988).When shown behaviour is different from what was expected, observers assign violation valence depending on (1) their evaluation of the behaviour as positive or negative, (2) the behaviour exceeding or falling short of expectations, and (3) the severity of the violation.Previous research shows that EVT may be applied beyond the borders of the close interpersonal contexts, as over time, people also form expectations regarding media figures (Cohen, 2010), customer service staff (Houston, Grandey, & Sawyer, 2018), and organizations (Helm & Tolsdorf, 2013).
The current research focuses on negative expectancy violation, thus on behaviour falling short of expectations.We suggest that observers experience stronger expectancy violation when a transgression is committed by an employee affiliated with a non-profit compared to a for-profit organization, because the difference between expected behavior and the observed transgression will be greater when communal sharing was expected compared to when market pricing was expected (e.g., Fiske & Tetlock, 1997).
H3a: In case of transgressions, affiliation with non-profit organizations is positively related with expectancy violation.H3b: In case of transgressions, affiliation with for-profit organizations is negatively related with expectancy violation.
Negative violations of expectations yield various negative consequences, such as negative affect (Burgoon & Hale, 1988), perceptions of deception (Bond et al., 1992) and a reduction of perceived closeness with the individual who violated expectations (Cohen, 2010).In this research, we propose that negative expectancy violation will manifest in observers' intent to punish individuals for their transgressions.In order to explain the public's different response to the same transgression when it was committed by employees affiliated with Oxfam.Specifically, we propose that the relationship between organizational affiliation and punishment is sequentially mediated by attributions of morality, warmth and competence (M1), expected CS or MP behaviour (M2) and expectancy violation (M3).We derive the following hypothesis:

Overview of studies
We tested our hypotheses in a series of studies using different samples.First, we conducted an extensive pilot (MTurk; N = 200) test to examine morality, warmth and competence perceptions across 20 large international non-profit and for-profit organizations.We used this pilot test to select organizations for our main subsequent studies (details are provided in the web-appendix).Study 1 (N = 297; MTurk) establishes the existence of the stereotype rub-off effect and tests whether attributions of morality, warmth and competence indeed rub-off on individual employees of non-profit and for-profit organizations.Study 2 (N = 300; Prolific) examines the influence of warmth, morality and competence attributions on expected behavior of others.We show that high perceived morality and warmth lead to expectations of CS behavior, while high perceived competence leads to expectations of MP behavior.Finally, Study 3 (N = 402; representative sample Dutch consumers from an ISO certified market research panel) tests the proposed three-step sequential mediation model and examines downstream behavioral consequences of the stereotype rub-off effect using an experimental game setting.

Study 1
Study 1 examines whether observers transfer stereotypes from the organization as a whole to individual employees of the organization.To be able to select organizations, we first ran a pilot test (see web-appendix) in which 20 non-profits and for-profits were evaluated on their perceived morality, warmth and competence.Results of the pilot showed that non-profit organizations are evaluated as significantly more moral, warmer (cf.Aaker et al., 2010), but surprisingly also as slightly more competent than for-profit organizations.This finding deviates slightly from previous research on organizational stereotypes (Aaker et al., 2010), but may be due to the specific organizations in our sample, which were all large international organizations operating for several years.Respondents might have concluded that non-profits operating in such contexts are more competent than for-profits.
However, this slightly unexpected finding did not affect the design of our main study, because used the pilot test to choose the organizations that were most distinct regarding their profile of morality and warmth versus competence.We selected the World-Wildlife Fun (WWF) and Save the Children as examples of non-profit, and Amazon and The Coca-Cola Company as examples of for-profit to be used in our subsequent studies.The selected organizations differed on morality and warmth, but were not perceived as immoral or not warm.In Study 1, we predict that employees of non-profit organizations will be perceived as more moral and warmer than their for-profit counterparts.In addition, we hypothesized stronger attributions of competence for employees of for-profit organizations; however, the results of the pilot test decreased our confidence in this hypothesis.

Method
Design and Participants.An a priori power analysis using G*Power (Faul, Erdfelder, Lang, & Buchner, 2007) was conducted to estimate sample size.With an effect size = 0.0625, α = 0.05 and power = 0.95, the projected sample size is approximately N = 153 for the simplest between/within group comparison.Thus, our proposed sample size of N = 300 will be more than adequate for the main objective of this study and adds statistical power.In total, we recruited 300 participants via MTurk for Study 1 (M age = 39.38,SD age = 10.62;141 male, 157 female, 2 other).Data was collected with the help of the organization "MTurk Data", which provided access to a representative sample, guaranteeing high quality participants for a Qualtrics study.Importantly, MTurk Data was not involved in creation of the experiment or the interpretation of results, but merely provided participants for our research.We used a between-subjects design and randomly assigned participants to a forprofit and a non-profit condition.In the non-profit condition, we introduced participants to an employee of either Save the Children or the WWF.In the for-profit condition, we introduced participants to an employee of either Amazon or the Coca-Cola Company.Three participants were removed from the dataset for failing the attention check described below.We continued the analysis with 297 participants (M age = 39.32,SD age = 10.61;138 male, 157 female, 2 other).Participants who completed the pretest on Mturk regarding their perception of organizations' morality, warmth and competence were banned from participating in Study 1 to avoid potentially biased responses.
Procedure and Dependent Measures.Participants read the following instruction: "In this study, we are interested in impressions people form about other people when they only have very limited information on that other person.Therefore, your task is to imagine you are being introduced to a person called Bob Miller.Bob Miller is an American, male and 43 years old.He lives in a mid-sized city in the United States.Bob Miller currently works as a manager for the organization < organization name > .Please take some time to picture the described person and form an impression of them." Participants indicated on seven-point Likert scales (ranging from 1 = Strongly disagree to 7 = Strongly agree) whether they perceived Bob as competent (competent, intelligent, skilled; α = 0.92), moral (moral, sincere, trustworthy α = 0.95) and warm (warm, friendly; r = 0.84, p < .001).
Participants then filled out an attention check examining how thoroughly they read the information given to them during the research.Participants read: "Astrology has always had a big influence on people's lives.Every day, people check their horoscope based on the date of birth.For some people, this is more important than others, but the fact is that it still provides highly profitable market.Some people claim that the sign of your horoscope also influences how thoroughly you read texts.This question is designed to test this.This astrology test is simple, when asked for your astrological sign you must indicate your favourite sport in the text box below.This question will help us to better understand the influence of horoscope signs on behavioural decision-making.Based on the text above, what astrological sign have you been asked to enter?" Participants indicated their age and gender and were given the opportunity to leave comments or ask questions regarding the study.Participants were then thanked, compensated and debriefed.

Results and discussion
We created a dummy variable indicating whether Bob Miller was described to participants as an employee of a for-profit or a non-profit organization (0 = for-profit and 1 = non-profit).We then conducted a t-test to compare means for morality, warmth and competence ratings of Bob Miller.As expected, the analysis shows significant differences on all three dimensions.Participants perceived Bob Miller as significantly more moral when he was affiliated with a non-profit (M = 5.90, SD = 0.80) as opposed to a for-profit organization (M = 5.09, SD = 1.06), t (295) = -25.36,p < .001. Bob Miller was further perceived as significantly warmer when affiliated with a non-profit (M = 6.05,SD = 0.83) compared to a for-profit organization (M = 4.93, SD = 1.10), t(295) = 9.83, p < .001.In line with the results of the pilot test, Bob Miller was also seen as slightly more competent affiliated with a non-profit (M = 5.92, SD = 0.70) compared to a for-profit organization (M = 5.64, SD = 0.87), t(295) = 3.12, p = .002.We conducted three simple regression analysis and found that the effect of organizational affiliation on perceived morality (β = 0.40, t(295) = 7.39, p < .0001)and perceived warmth (β = 0.50, t(295) = 9.83, p < .0001)were much stronger than for perceived competence (β = 0.20, t(295) = 3.12, p = .002).
The results suggest that employees of non-profit organization are indeed perceived as more moral and warm than their for-profit counterparts, confirming hypothesis H1a and H1b.However, based on our findings regarding perceptions of competence we reject hypothesis H1c, as employees of the WWF and Save the Children were also seen as significantly more competent than employees of Coca-Cola and Amazon.It is important to note that while the difference in competence was significant, it was considerably smaller than differences pertaining to morality and warmth.The result can potentially be explained by the choice of organizations for Study 1, as the WWF indeed scored slightly above average on competence but was chosen for a lack of better alternatives of organizations scoring high on morality and warmth and average on competence.In sum, results of Study 1 show that organizational stereotypes of warmth, morality and competence "rub-off" from the organization to individual employees affiliating with the organization.

Study 2
Study 1 established that organizational stereotypes rub-off on individuals within the organization.As a next step, Study 2 examines the link between perceptions of morality, warmth and competence and expected behavior of others.We hypothesize that observers will expect highly moral and warm individuals to act according to communal sharing principles, while predominantly competent people will be expected to act according to market pricing principles.

Design and
Participants.An a priori power analysis using G*Power (Faul et al., 2007) was conducted to estimate sample size.With an effect size = 0.0625, α = 0.05 and power = 0.95, the projected sample size is approximately N = 153 for the simplest between/within group comparison.Thus, our proposed sample size of N = 300 will be more than adequate for the main objective of this study and adds additional statistical power.
We initially gathered responses from 300 participants (M age = 35.80,SD age = 12.85; 149 male, 149 female, 2 other) via Prolific.However, eight participants did not pass the attention check, resulting in the collection of eight additional participants, who all passed the attention check.We continued the analysis with the 300 participants who all successfully passed the attention check (M age = 35.93,SD age = 12.93; 146 male, 152 female, 2 other).We used a between-subjects design randomly assigning participants to one of three conditions describing Bob Miller as a particularly moral (N = 102), warm (N = 99) or competent (N = 99) person.
Procedure and Dependent Measures.We used the same description of Bob Miller as in Study 1, but omitted information on organizational affiliation and instead manipulated his perceived morality, warmth and competence, using items from Study 1. Participants read the following instructions: "In this study, we are interested in impressions people form about other people when they only have limited information on that other person.Therefore, your task is to imagine you are being introduced to a person called Bob Miller.Bob Miller is an American, male and 43 years old.He lives in a mid-sized city in the United States."Depending on the condition participants were assigned to, they further read "In addition, Bob Miller is an especially honest person.Family, friends and co-workers agree that he is one of the most trustworthy and sincere people they have ever met" (morality condition); "In addition, Bob Miller is an especially friendly person.Family, friends and co-workers agree that he is one of the warmest people they have ever met" (warmth condition); or "In addition, Bob Miller is an especially intelligent person.Family, friends and co-workers agree that he is one of the most competent and skilled people they have ever met" (competence condition).
In order to measure expectations regarding communal sharing and market pricing behavior, we adapted items from Haslam (1995) relational models scale.Specifically, we presented participants with 6 statements reflecting communal sharing and market pricing behavior.Communal sharing: He would prioritize cooperation with other people; He would give the shirt off his back for others; He would cancel plans to provide help for others; α = 0.81.Market pricing: He would act in a business-like way towards others; He would act purely rationally towards others; He would keep track of how much reward he is getting for the amount of time, effort or money he spends in relationships with others; α = 0.45.Due to the insufficient Cronbach's Alpha of the market pricing scale, we decided to remove the third item and build the market pricing index from items 1 and item 2 (r = 0.44, p < .001).
Participants then indicated on seven-point Likert scales (ranging from 1 = Strongly disagree to 7 = Strongly agree) whether they expect Bob Miller to act according to the statements.Participants filled out the same attention check as in Study 1, indicated their age and gender, and were given the opportunity to leave comments or ask questions regarding the study.Participants were then thanked, compensated, and debriefed.

Results and discussion
We conducted three t-tests to compare means regarding the expected communal sharing or market pricing behavior across conditions.
The findings suggest that, compared to competence, both perceptions of morality and warmth lead to greater expectations of communal sharing behavior.By contrast, compared to morality and warmth, perceptions of competence lead to greater expectations of market pricing.Extending the findings of Study 1, Study 2 shows that perceptions of morality, competence and warmth affect expectations of behavior corresponding to communal sharing and market pricing relational models.Since individuals affiliated with non-profit organizations are stereotyped as more moral and warmer and only as slightly more competent, the public expects them to act in the spirit of communal sharing more than in accordance with market pricing.These findings are important, as they show that morality, warmth and competence stereotypes impact expected communal sharing and market pricing behavior.In our final study, we examine how these behavioral expectations influence expectancy violation and punishment when employees affiliated with non-profit or for-profit organizations transgress societal norms.

Study 3
Our third study examines consequences of the stereotype rub-off effect and the resulting behavioral expectations on expectancy violation and punishment.Based on the findings of Studies 1 and 2, Study 3 tests the proposed three-step sequential mediation model, in which organizational stereotypes of morality, warmth, and competence (M1), communal sharing and market pricing expectations (M2) and expectancy violation (M3) mediate the relationship between organizational affiliation and punishment.We propose that the relationship between communal sharing and expectancy violation is moderated by whether or not the actor commits a transgression.When employees of non-profit organizations, who are seen as moral and warm and are expected to behave in line with communal sharing, commit transgressions, observers will experience increased expectancy violation, which in turn prompts greater punishment compared to employees of for-profit organizations.
To test this with a behavioural measure, we used an experimental game -often applied as a way to study social decision making (cf.van Dijk & De Dreu, 2021) -to capture willingness to punish.Specifically, we used the third-party punishment game (or altruistic punishment game), a situation in which an independent observer C observes Person A and B who play a dictator game.Person A makes a decision that affects the financial situation of both A and B, but B cannot have an influence on this decision.After observing the decision by A, C can decide to give up money to punish Person A. Prior research showed that many people offer a (nearly) equal split in dictator games, and that strong deviations from this equal split, e.g., 80-20 offers, are being punished (Fehr & Fischbacher, 2004;Leliveld, van Dijk, & van Beest, 2012;Van Doorn, Zeelenberg, & Breugelmans, 2018).Our experimental game thus mimics a situation in which people observe transgressions (unequal splits) and are given the opportunity to punish the transgressor, allowing us to test the effect of organizational affiliation on willingness to punish transgressors.A supplementary study, measuring participant's intent to punish transgressors rather than their behaviour can be found in the web-appendix.

Design and
Participants.An a priori power analysis using G*Power (Faul et al., 2007) was conducted to estimate sample size.With an effect size = 0.0625, α = 0.05 and power = 0.95, the projected sample size is approximately N = 186 for the simplest between/within group comparison.Thus, our proposed sample size of N = 400 will be more than adequate for the main objective of this study and should also allow for expected attrition and our additional objectives of controlling for possible mediating/moderating factors.
We recruited participants via the Dutch market research agency "PanelWizard".PanelWizard provides access to a high quality, ISO certified panel of 33.000 Dutch members for market-and academic researchers.We obtained a balanced sample of the Dutch population according to the key variables age, gender, family situation, employment, education and region.Since participants in our panel were exclusively Dutch, the experiment was translated from English to Dutch by a native speaker of Dutch in our research group and was peer reviewed by other native speakers in order to solve the few remaining issues.
Initially, we collected data from 400 participants (Age groups: 18-25, N = 58; 23-34, N = 57, 35-44, N = 69; 45-54, N = 82; 55-64, N = 67; 65 and older, N = 67; 199 male, 201 female).We randomly assigned participants to the conditions of a 2 (Organizational Affiliation: non-profit vs. for-profit) × 2 (Transgression: yes vs. no) betweensubjects design.We excluded 35 participants from the analysis as they failed one or more of the three attention checks described below.Preemptively accounting for the proportion of expected attention-check failures, we collected responses from an additional 44 participants, which all passed the check.Next, we excluded 7 participants answers after detecting their answers as multivariate outliers on either the morality, warmth, competence, communal sharing, market pricing, expectancy violation or punishment dimensions.
Procedure and Dependent Measures.Participants gave informed consent and were told that they would take part in a decision-making experiment in which their choices could influence other participants' compensation for taking part in the experiment.Participants read the following instruction: "The decision-making situation is as follows.In total, there are 3 people involved in the game: Player A, Player B, and Player C. Participants are assigned to their roles based on the sequence they complete the study, meaning that the first participant who starts this study is player A, the second is player B, the third is Player C. The fourth participants to start is Player A again, etc. etc.Before we continue, we first need to assign roles.As explained, this is based on the timing of the start of your participation.Please advance to the next page for role assignment."Importantly, no role-assignment took place and all participants received the role of Player C to enable us to manipulate information about Person A, but we needed this cover story to ensure the credibility of the following experiment.
We explained to participants that their role as Player C involves evaluating a decision made by another participant in the research, Person A, which affected another participant, Person B. We explained that A possesses 100 chips each worth 10 eurocents and is tasked with making the decision how to divide these 100 chips between him/herself and B, who possesses no chips.B cannot reject A's decision.We laid out to participants that before evaluating A's decision, we would ask about their general impression of Player A, based on some general information.This is where we manipulated organizational affiliation by presenting participants with a short table containing Player A's nickname in the experiment (BLmiller), his age group (35-44), his gender (male) and his current employer.We stated that Player A worked for either The Coca-Cola Company (for-profit condition) or The World-Wildlife Fund (nonprofit condition).Participants were then asked to form in impression about Player A. Subsequently, we measured perceived morality, competence and warmth of Person A using the same items as in Study 1 (morality α = 0.91; warmth r = 0.61; competence α = 0.85).We then measured expectations of CS and MP behavior using the same items as in Study 2 (CS α = 0.71; MP α = 0.66).
Next, we presented participants with A's decision on the allocation of the 100 chips between himself and player B. We manipulated the decision to reflect either an equal split of 50-50 chips between players (no transgression condition) or a clearly unequal distribution of 80-20 chips in favor of A (transgression condition).We then measured participants' expectancy violation on a seven-point Likert-scale (ranging from 1 = Strongly disagree to 7 = Strongly agree) using three items (Given my previous impression of Player A: He/She acts like I expected him/her to act; I am negatively surprised by his/her behavior; I am positively surprised by his/her behavior; α = 0.62).The first and last item were reverse coded, such that higher scores indicate increased negative expectancy violation.Next, participants were told that they received 50 chips (each with a value of 10 cents) themselves in their role as player C. Participants could now choose whether to keep the chips, or whether to spend their chips to lower the amount of player A's chips.Specifically, for each chip they would give up, Person A would lose 3 chips.We also explained that of all participants, 10 participants would be randomly selected and would be paid accordingly.We measured punishment as the number of chips given up by participants in order to lower the number of chips of Person A.
In the final stage, participants filled in three attention checks asking about their role in the experiment (A, B, C), the gender of the player A (male, female) and the working place of player A based on the information they received (open question).Participants were then debriefed, and received an explanation on the research setup and the non-existence of Players A and B. Participants had the option to give comments and to withdraw consent from the research.Participants were thanked and compensated.

Results
Table 1 shows the correlation matrix of Study 3. Table 2 shows means and standard deviations across the for-profit vs. non-profit and the transgression vs. no transgression condition.Tables 3-5 provide an overview of the model pathways for models 1-3.
First, we examine model 1, in which morality (M1), expected communal sharing behaviour (M2) and expectancy violation (M3) sequentially mediate the relationship between organizational affiliation and punishment.See Fig. 1 for an overview of the findings.In line with findings from Study 1, the relationship between organizational affiliation and morality was positive and significant (B = 0.55, p < .001)showing that participants perceived Player A as more moral when he was described as working for a non-profit as opposed to a for-profit organization.In addition, confirming the findings of Study 2, the relationship between morality and expected communal sharing behaviour was also positive and significant (B = 0.48, p < .001).Participants expected A to act according to communal sharing principles based on his perceived morality.Next, we examined the consequences of the stereotype rub-off effect when A committed a transgression.The effect of expected communal sharing behaviour on expectancy violation was negative and significant (B = -0.29,p < .001)when no transgression was committed.However, it was positive and significant (B = 0.36, p < .001)when a transgression was committed.This shows that when participants expected communal sharing behaviour, an occurring transgression prompted expectancy violation, whereas no transgression prompted expectancy confirmation.
Next, we analysed the indirect effect of organizational affiliation on punishment through morality, expected communal sharing behaviour and expectancy violation.In the no transgression conditions, the indirect effect was negative and significant (B = -0.21,CI 95% = [-0.355,-0.094]); however, in the transgression condition it was positive and significant (B = 0.25, CI 95%, = [0.094,0.471]).This shows that, in line with our hypothesis, organizational affiliation with non-profit organizations indirectly increased punishment intentions through morality, expected communal sharing behaviour and expectancy violation when A committed a transgression, but not in the absence of a transgression.
Second, we examined model 2, in which warmth (M1), expected communal sharing behaviour (M2) and expectancy violation (M3) sequentially mediate the relationship between organizational affiliation and punishment.See Fig. 2 for an overview of the findings.The effect of organizational affiliation on warmth was positive and significant.Confirming the results of Study 1, organizational affiliation significantly predicted judgements of warmth (B = 0.57, p < .001).In addition, the effect of warmth on expected communal sharing behaviour was significant and positive (B = 0.47, p < .001).Aligning with the findings of Study 2, participants expected Player A to act in line with communal sharing principles when he was perceived as warm.Similar to model 1 with morality as M1, the effect of expected communal sharing behaviour on expectancy violation was negative and significant when no transgression was committed (B = -0.27,p = .001).However, it was positive and significant when a transgression was committed (B = 0.39, p < .001).Next, we analysed the indirect effect of organizational affiliation on punishment through warmth, expected communal sharing behaviour and expectancy violation.Similar to model 1, and in line with our hypothesis, when no transgression occurred, the indirect effect was negative and significant (B = -0.19,CI 95% = [-0.327,-0.088]); however, in the transgression condition it was positive and significant (B = 0.27, CI 95% = [0.111, 0.488]).This shows that organizational affiliation with non-profit organizations indirectly increases punishment through warmth, expected communal sharing behaviour and expectancy violation in case of transgression, but not in the absence of a transgression.
Last, we examined model 3, in which competence (M1), expected market pricing behaviour (M2) and expectancy violation (M3) sequentially mediate the relationship between organizational affiliation and punishment.See Fig. 3 for an overview of the findings.The effect of organizational affiliation on competence was not significant (B = 0.10, p = .200).Contrary to the results of Study 1, participants did not expect Player A to be a more competent individual when he was affiliated with a non-profit organization.The effect of competence on expected market pricing behaviour was significant and positive (B = 0.37, p < .001).Aligning with the findings of Study 2, perceived competence drove market pricing expectations.The effect of expected market pricing behaviour on expectancy violation was negative and non-significant when no transgression was committed (B = -0.10,p = .200),but was negative and significant when a transgression was committed (B = -0.27,p = .001).Next, we analysed the indirect effect of organizational affiliation on punishment through competence, expected market pricing behaviour and expectancy violation.Importantly, the findings differ from the results of models 1 and 2. The effect was negative and nonsignificant when no transgression was committed (B = -0.01,CI 95% = [-0.042,-0.010]); it remained negative and non-significant in the transgression condition (B = − 0.03, CI 95% = [-0.082,0.015]).Compared to models 1 and 2, this shows the relative unimportance of competence judgements and subsequent market pricing expectations on punishment in case of transgressions.Put differently, organizational affiliation with non-profit organizations did not indirectly increase punishment through competence, expected market pricing behaviour and expectancy violation in case of a transgression.
5.3 Discussion.Study 3 contributes four major findings.First, we validate the results of Studies 1 and 2, by showing that organizational affiliation with non-profit as opposed to for-profit organizations leads to strong attributions of morality and warmth towards the individual affiliated with the organization.However, as in Study 1, we do not find the hypothesized relationship between affiliation with a for-profit organization and attributions of competence.Instead, affiliation with non-profit organizations also increased perceived competence in Study 2, albeit to a much lesser extent compared to perceptions of morality and warmth.Study 3 does not show evidence to confirm this finding regarding competence.Second, Study 3 confirms both morality and warmth as predictors for expectations regarding communal sharing, but not for market pricing behaviour.By contrast, competence is identified as a predictor of market pricing, but not of communal sharing behaviour.Third, we provide evidence that organizational affiliation increases punishment for transgressions through its indirect effect via morality and warmth stereotypes, expectations of communal sharing resulting from these and the associated increased expectancy violation in case of transgressions.Importantly, this indirect effect was not observed for attributions of competence, resulting expectations of market pricing and expectancy violation.

Findings
We started with the observation that employees face harsher repercussions from the public for the same transgressions when they are affiliated with non-profit compared to for-profit organizations.Aiming to explain this phenomenon, we proposed that stereotypes of morality, warmth and competence rub-off from organizations to individual employees even outside the organizational context, and thus effectively lead observers to hold people to higher expectations of communal sharing when these people are affiliated with non-profit as opposed to for-profit organizations.The present research extends current knowledge in three important ways.
First, we show that employees affiliated with non-profit organizations are indeed stereotyped by others according to the common organizational stereotypes of morality and warmth, a phenomenon we labelled the stereotype rub-off effect (Study 1).Specifically, we found that employees affiliated with non-profit organizations are stereotyped as more moral and warm than their for-profit counterparts.Previous research (Aaker et al., 2010) suggests that for-profit organizations are generally stereotyped as more competent than non-profit organizations and thus we expected the stereotype to rub-off on individuals within the organization accordingly.However, both the pilot test we conducted and the results of Study 1 challenge this assumption, as some non-profit organizations in our pilot test, such as the WWF, Save the Children, and most notably, Doctors without Borders, were rated as more moral, warm and as more competent than for-profit organizations in our sample.One potential explanation for why certain non-profit organizations are rated as more moral, warm, and competent is that people may attribute competence, but not morality and warmth, to organizations indiscriminately (Kirmani, Hamilton, Thompson, & Lantzy, 2016).This means that the mere fact that an organization survives on the market may be enough evidence for the public to attribute a certain level of competence to organizations.However, Study 3 showed no evidence of increased expectations of competence due to affiliation with a for-profit organization.These finding are surprising and show that attributions of competence may not be as clearly associated with for-profit organizations as previous research suggests.These considerations invite followup research to systematically test the associations between type of organization and stereotype content, as well as potential boundary conditions for these associations.Even though our findings regarding competence were unexpected, they do not decrease the validity of the stereotype rub-off effect in the domains of morality and warmth.Results of the pilot tests showed that non-profit organizations were rated as more moral and warmer than forprofit organizations and only as slightly more competent.This difference is found again in Study 1 when examining how stereotypes rub-off on individuals affiliating with different organizations.Results show that morality, warmth and competence attributions followed the same pattern, as individuals affiliated with non-profit organizations were rated as more moral and warmer and only as slightly more competent, thus matching the stereotypes of the organization.The findings complement and extend previous research by Aaker et al. (2010) by showing that not only non-profit organizations as a whole are perceived as more moral, but that organizational stereotypes also extend to individual members, especially when the morality and warmth dimensions are concerned.
Second, our findings show that organizational affiliation indirectly impacts the behavior people expect from others, such that employees of non-profit organizations are expected to act according to communal sharing principles and employees of for-profit organizations are expected to act in the spirit of market pricing (Study 2).Specifically, stereotypes of morality and warmth, flowing from the type of organization employees affiliate with, lead to expectations of communal sharing behavior, while attributions of competence lead to expected market pricing behavior.Not surprising in light of their theoretical and conceptual closeness, both constructs increase expectations of communal sharing behavior to almost the same extent.In the contexts where impression formation based on organizational affiliation, we did not find notable differences between morality and warmth and consider them to be equal predictors of expected communal sharing behavior.These findings add to the discussion on whether morality and warmth should be treated as distinct constructs (Cuddy et al., 2007;Fiske et al., 2018) or whether they are theoretically too closely linked to justify a distinction between them (Leach et al., 2007;Leach et al., 2007;Brambilla & Leach, 2014).
Third, we demonstrate that the stereotype rub-off effect and its consequences on expected behavior affect experienced expectancy violation in case of transgressions (Study 3).Specifically, the public experiences increased expectancy violation when transgressions are committed by employees of non-profit organizations, who are stereotypically seen as warm and moral and who are expected to act according to communal sharing principles.Study 3 adds to expectancy violation literature by identifying organizational affiliation as an origin of expectations in the relationship between organizational members and the public.Note that organizational affiliation is likely one of several indicators people use to infer morality, warmth and competence and build subsequent expectations of communal sharing or market pricing behavior.For example, prior research suggests social class as another determinant of warmth and competence perceptions (Durante, Tablante, & Fiske, 2017).We showed that organizational affiliation contributes to stereotypes of morality, warmth and competence "rubbing off" to affect expectation s of subsequent behaviour; our findings  potentially apply to a greater variety of signals that observers use to infer stereotypes.Therefore, the relevance of the presented findings extends beyond the organizational context.
Taken together, our analysis provides strong support for the rationale that employees affiliated with non-profit organizations are stereotypically seen as more moral and warm than their for-profit counterparts.Observers' strong attributions of morality and warmth in turn manifest in expectations that these employees will act in line with communal sharing principles, an expectation that is not evident for employees of for-profit organizations.The different expectations observers form about employees affiliated with non-profit organizations ultimately result in increased expectancy violation when these employees commit transgressions, leading to harsher punishment.
While their greater perceived morality and warmth might give nonprofit organizations an advantage over other companies, we have shown the risks associated with morality and warmth stereotypes when employees commit transgressions.This is illustrated by the introductory example of Oxfam, a non-profit company, which suffered substantially harsher repercussions for their scandal than comparable for-profit organizations whose managers engaged in similar transgressions.This paper explained the underlying process for this observed phenomenon.It appears that employees of non-profit organizations are particularly vulnerable to the consequences of transgressions because they are held to higher behavioural standards than their for-profit counterparts.

Limitations & future research
While providing support for our theoretical rationale across three studies, our research is not without limitations.First, aligning with our goal to provide first support for the proposed effect, we confined ourselves to conducting highly controlled experimental studies.We encourage follow-up research in more realistic and rich contexts, for example by analysing news coverage of managers' transgressions as a function of organizational affiliation.Such approaches will also valuably inform the search for additional factors observers might rely on to infer morality, warmth, and competence from others.
Second, the three-item scale used to assess market pricing behaviour was adapted from Haslam (1995).We selected three from the six original items to measure expectations regarding market pricing behaviour and repeated the process for the communal sharing scale.In both cases, we chose three items that fitted the purpose of our research best.i.e., we selected only the items not assuming a pre-existing close interpersonal relationship between the participant and the actor to fit the context of our research.The resulting scales produced a satisfactory reliability index for the communal sharing scale, but the reliability index for the market pricing scale was insufficient, forcing us to proceed with a twoitem index to measure market pricing adequately.While this may limit the interpretation of the market pricing concept, it does not affect our major findings with respect to the stereotype rub-off effect.In addition, the main findings regarding the effects of morality and warmth stereotypes on expected communal sharing behaviour and subsequent expectancy violation and increased punishment in case of transgressions are not altered by this limitation.Nevertheless, future research should develop refined measurements of market pricing expectations outside of close interpersonal relationships.
Third, in our studies participants received only limited information, such as age, gender, location and employment about the person they were asked to evaluate.We purposely chose to use these key factors in order to isolate the downstream effects of organizational affiliation as much as possible without adding additional noise to the experimental design.The information available to participants can be considered similar to information obtained from (online) newspaper articles, which typically include a transgressors name, age, country and organizational affiliation.Consequently, this limited information is then used in readers' impression formation processes.While our findings show support for our main hypotheses in a research context with limited information about the transgressor, future research is necessary to examine the stereotype rub-off effect in more personal, information rich contexts in order to study the generalizability of the effect.

Conclusion
The present research shows that, presumably unbeknownst to themselves or their organization, employees are seen as more moral, warm, and as slightly more competent when they affiliate with nonprofit organizations.Due to strong attributions of morality and warmth, the public expects others to behave in line with communal sharing principles.Ultimately, the demonstrated stereotype rub-off effect results in stronger expectancy violation when employees of nonprofit organizations committed transgressions.Our findings underline the relevance of anticipating potential hazards for organizational reputation that might result from employees' transgressions.Importantly, our findings suggest that adequate and fast responses to scandal and crisis may be particularly important for non-profit organizations.Negligence of non-profit organizations' potential vulnerability to the public's condemnation might ultimately impair the positive contributions of non-profit organizations to society and calls for further identification of stereotypes in stakeholder management.

Declaration of Competing Interest
We declare that there is no competing interest.

Table 2
Variable Means across Conditions, Study 3.

Table 5
Summary of Pathways Model 3.