In-group Favoritism in Talent Attribution
Julia Plaisier
Bachelor Thesis Julia Plaisier (10340459) Universiteit van Amsterdam Noah Millman
31-05-2016 Word count: 4538
Abstract
There is a tendency to prefer people belonging to the in-group over the out-group. Social Identity Theory explains this favoritism with group-based self-esteem and the striving for positive distinctiveness. From this theory it can be expected that in-group favoritism is not only shown in relevant aspects but also in irrelevant aspects. Therefore, I conducted an online study with Ajax and Feyenoord supporters. Participants had to attribute talent to painters belonging to the in-group and the out-group. There was no in-group favoritism found in this study. Thus, Social Identity is not supported but further research is needed to make strong conclusions about this theory.
Introduction
Groups are a pervasive feature of our social lives. We interact with people who share common group identities and also find ourselves interacting with others who belong to different groups (Balliet, Wu & de Dreu, 2014). Being a fan of a football team is an example of belonging to a social group. When being a fan of a football team you experience passion for the game and in particular for your own team. You might feel excitement when your team won a game and even feel sad and disappointed when your team is doing badly. When a significant player of your team transfers to another relevant club this player can be seen as a traitor. This is because people evaluate in-group members more positively than out-group members (Perdue, Dovidio, Gurtman & Tyler, 1990). Supporters are convinced of the
extraordinary qualities that their team possesses, they tend to reward in-group members more than out-group members (Tajfel, Billig, Bundy & Flament, 1971). Supporters see their team as the best team in the whole league. For example: supporters of FC Barcelona will most likely agree that Messi was named best player in Europe (UEFA, 2015). But supporters of Real Madrid might find that Ronaldo, who was also in the running for this title, should have been named best player. Would Real Madrid supporters have this opinion because they truly find Ronaldo better or just because he plays for their team and Messi does not?
Preferring the player playing for your own team is called in-group favoritism. Because the player from your team is part of your in-group you will tend to favor this player over a player from the out-group (Turner, Brown & Tajfel, 1979). Messi, who plays for the opponent, is therefore seen as out-group, which makes him less likeable for the Madrid supporters. This preference not only occurs in big events such as naming the best player in Europe but also more generally by preferring everything about your team (Mullen, Brown & Smith, 1992). This phenomenon can be seen in situations where there is an in-group and
out-group. People favor their own in-group compared to the out-group (Turner et al., 1979). They also evaluate in-group members more positively than out-group members (Brewer, 1979).
People evaluate in-group members more positively than out-group members to maintain or create a positive impression about the in-group. They do this because they base their self-image on the group that they belong to (Voci, 2006). Assuming that people tend to seek a positive identity it is necessary to also keep seeing your in-group positively. The most common way to do this would be to perceive that your group is more positive than another relevant social group (Tajfel, 1981). To think that the team that you are supporting is better in playing football than the opponent logically follows from the above. Your team being good at football is an important aspect when you are a football supporter but one can expect that being better than another relevant social group on an unimportant domain would also promote a positive group-image. If in-group favoritism is shown in irrelevant domains such as talent in general instead of just football talent, this would mean that for example advertisement
agencies could use this phenomenon to enhance convincingness of their advertising. Also for hiring procedures it would be relevant to know how easily people can be manipulated into like someone and judging their competence.
For this purpose, I conducted a study with Dutch football supporters of Ajax and Feyenoord. These are two of the most rivaling teams in the Netherlands. My aim was to explore whether supporters show in-group favoritism in talent attribution when rating paintings made by in-group members (e.g. Ajax or Feyenoord supporters). In this case the talent for painting is a non-related talent for football supporters but should by my expectations still cause in-group favoritism.
In the following piece I will start with a review of previous research done on this topic. I will start my line of reasoning by first explaining and define a few key concepts and then move on to explaining how these relate to each other. Then I will discuss the Social
Identity Theory (SIT) and the predictions for in-group favoritism that follow from this theory. Next I will describe the method and results of my study and I will conclude with a discussion of my findings.
Literature Review
People organize themselves and others into groups and we are all members of different types of groups (Brewer, 1996, p. 1). An example of one of those groups is called the group, the people who do not belong to this group are, naturally, named the out-group. The in-group can simply be defined as the in-group you feel that you belong to. This in-in-group consists of other people with whom you share something; this can be a preference or even a part of your identity. When you belong to such a group the other people that do not belong to this group are seen as the out-group. This does not necessarily have to be a group that is the exact opposite of yours, a group can be seen as the out-group simply by not belonging to the in-group. Unfortunately, the mere perception of belonging to one of two distinct groups is sufficient for in-group bias in the distribution of monetary rewards (Turner et al., 1979). A form of in-group bias is known as in-group favoritism. According to Turner et al., in-group favoritism is a descriptive concept of referring to any tendency to favor the in-group over the out-group, in behavior, attitudes, preferences or perception (1979). In a study conducted by Tajfel et al. (1971) they showed that in a minimal group, participants tend to maximize outcomes for the in-group, showing in-group favoritism.
Social Identity Theory
Social Identity Theory (Tajfel & Turner, 1979) tries to explain in-group favoritism. This theory states that everyone has a social identity. A person’s social identity is those
aspects of his self-concept conduced by the social groups to which the person perceives themselves to belong. In general, individuals are motivated to establish positively valued differences between the in-group and a relevant out-group to achieve a positive social identity (Turner et al., 1979). This group-based self-esteem can provide motivation to distinguish the in-group from the out-group. Because the view of the in-group is part of their self-concept it is assumed that people strive to value this group as positive. This search for positively valued distinctiveness can lead to in-group favoritism (Turner et al., 1979).
For in-group favoritism to occur there are a few conditions according to SIT. To start, the person showing group favoritism should have the perception of belonging to the in-group, identifying with it and therefore base their self-esteem on the appreciation of this group (Ellemers, Kortekaas & Ouwerkerk, 1997). Also, the quality on which this person is judging an in or out-group member should be either relevant or salient. Then, it is important that the out-group is being seen as a comparison group that is relevant. And last, it should be ambiguous what the appreciation is for the attribute so that it can be seen as favoritism and not just as a general objective opinion (Turner et al., 1979). For example, no one can convincingly keep arguing that a person in his or her amateur football team is better than Messi (or Ronaldo). This is because the talent of the amateur is not comparable to the professional player and therefore not ambiguous.
It is seen to be relatively less important to strive for positive distinctiveness relative to out-groups when the out-group is experienced as already being distinctly different (Brewer, 1996). To explain in-group bias with the striving for positive distinctiveness there are
assumptions one must make. The first one is that the appreciation of the group largely consists of the comparison between the social groups. Hereby the evaluation of the in-group cannot be made without comparing it to the out-group (Brewer, 1991). The evaluation is not absolute but rather relative to other groups. The second assumption of SIT is that positive
distinctiveness signifies the fundamental need for a positive self-esteem, this self-esteem is based on the social identity (Brewer, 1996).
Even though the assumptions made by SIT and the positive distinctiveness view of this theory that predict that in-group favoritism can only occur when the aspect on which the groups are being compared is relevant or salient. I think that it is very interesting to study if in-group favoritism can still occur when judging irrelevant aspects of the groups that are being compared. This is because you could expect that if people base their self-esteem on the group’s esteem and evaluation, even irrelevant aspects on which the group is seen as better than other groups should enhance their self-esteem. And every change in a positive direction will satisfy the basic fundamental need for a positive self-esteem (Oakes & Turner, 1980).
Talent Attribution
Every person views talent and giftedness differently, according to Gagné (1985) talent refers to a performance that is distinctly above average in one or more fields of human
performance. Giftedness corresponds to competence, which is distinctly above average in one or more domains of ability.
As described in the paragraphs above, the evaluating of features of an in or out-group member are assumed to be lead by a striving for positive distinctiveness. This striving is driven by the need for a positive group-based self-esteem because the social identity of a person consists of this. In this study I am going to look at how people attribute the talent of in and out-group members. Note that talent attribution is different from just liking something. In this case it is important to understand that someone can for example feel neutral about a song that they are listening to but still recognize that the singer is talented. To have a talent for something is here defined as a performance that is distinctly above average.
In this study I looked at how people would attribute the talent of a painter belonging to their in-group and to a relevant out-group. Since a big part of being a football supporter is comparing your team to the rivaling teams, this is expected to be a relevant and salient out-group. Therefore I chose to use participants living in Amsterdam and Rotterdam, assuming that people generally support the team based in their own city. Even though Rotterdam based football club Feyenoord has not won the Dutch championship in over ten years it is still seen as one of the most important opponents. For example, the match between Ajax and Feyenoord is always called classic (‘de klassieker’).
Looking at this situation from a SIT perspective I expected that the groups used are relevant. The talent attribution from the painter belonging to either the in-group or the out-group is expected to evoke in-out-group bias. Because supporting a team is part of a person’s social identity and someone will derive their self-esteem based on that of the group, it is expected that people will show a striving for a positive distinction relative to the out-group. This will lead to in-group favoritism when evaluating the talent of the painter.
Based on existing literature described in the paragraphs above, my prediction of my study is as follows. I expect to see that participants will attribute more talent to an in-group member than to an out-group member (Hypothesis 1a). I expect that this effect arises from a group based self-esteem, because of striving for a positive social identity (Hypothesis 1b). Thus, in this study it is expected that Feyenoord supporters will attribute more talent to a painter who is also a Feyenoord supporter than to a painter who is an Ajax supporter. The same goes for Ajax supporters.
Methods
Sample
For this study I collected a sample using Ajax and Feyenoord supporters. The assumption was made that people that were born and raised in Amsterdam would naturally support Ajax if given the choice between Ajax and Feyenoord. The same assumption was made for the participants living in Rotterdam. Although Rotterdam has two large football clubs, Feyenoord and Sparta, I assumed that even Sparta supporters would still favor Feyenoord over Ajax because of the rivalry between the two cities.
To determine my minimal sample size, I ran a power analysis using G*power for a mixed design based on a previously conducted study that found a main effect of ηp2 = .0235. Assuming this effect size and parameters ∝ = .05 and 𝛽𝛽 = .8, the calculated sample size should be at least N = 140. My analyses were done using a significance threshold of ∝ = .05
The minimal sample size was set at N = 140, which means each group of participants (Ajax fans and Feyenoord fans) had to have a minimal sample size of N = 70. However, because of the limited amount of time I also set our goal at collecting as much data as I could in the coming weeks, knowing that I would have to stop collecting data when time ran out even if I had not yet met the minimal sample size of N = 140.
After approximately three weeks of active and passive data collection I closed the online survey. Unfortunately, I had not reached the goal of collecting data from 140 participants. I collected a sample size of N = 119 but when looking at the data it turned out that there were a lot of people who dropped out after filling out only the first few questions. In the end I only got a useful sample of N = 58 with 51 Ajax supporters and 7 Feyenoord
supporters. There was one possible outlier that I identified using a box-plot. Because of the already very small sample size this participant was not excluded from the analyses.
Eventually, all analyses were performed with 51 Ajax fans and 7 Feyenoord fans, leaving a sample size of N = 58.
Materials
The questionnaire consisted of three parts. First the participant was presented with an informed consent form and a small explanation of my study. Then the first part started. The first part consisted of a few basic questions such as their gender, city of residence and asking them which football club they supported. Of these questions only the answers of the question which football club they were a supporter of was used for analyses. The second part consisted of the main questions for my research. Talent attribution was measured using a series of paintings. Participants had to answer questions attributing talent to the painter using a 7-point Likert scale (1 = Completely Disagree, 7 = Completely Agree). An example statement: ‘This painter is talented”. This statement was followed by a few other questions concerning the painting that was shown. These were questions such as “The colors of this painting give me a good feeling”. There were six different paintings, three that had a vignette stating that an Ajax supporter made them and three by a Feyenoord supporter. The third part of my questionnaire consisted of questions from the Collective Self-Esteem Scale (Luhtanen & Crocker, 1992). This scale has a reliability ranging from having a Cronbach’s ∝ = .83 to ∝ = .88. Item-total correlations ranged from .51 to .80. In order to check whether the subscale from the
Collective Self-Esteem Scale had been reliable for my study I used a Cronbach’s alpha analysis with a cut-off point of Cronbach’s ∝ = .70. I only used the questions from the subscale ‘Importance to Identity’ because only this subscale was particularly relevant for my study. These were only four statements of which participants had to say whether they agreed with it on a 7-point Likert-scale (1 = Completely Disagree, 7 = Completely Agree). The last two questions were needed for our manipulation check. These were again two statements for
which a 7-point Likert-scale (1 = Completely Disagree, 7 = Completely Agree) was used. The statements were “I and other Ajax supporters belong to the same group” and “I and other Feyenoord supporters belong to the same group”. Every participant, regardless to which fan group they belonged had to answer both questions. At the end of the questionnaire
participants were asked for their email address, they could fill this in if they wanted to have a chance to win the gift card that would be given to a random participant.
Procedures
The study was conducted via only surveys using Qualtrics. The survey was distributed via social media such as Facebook. I also sent the link to the survey to several people in my family who live in Rotterdam so that they would distribute this to their friends who support Feyenoord. I also approached several private Facebook groups for Feyenoord supporters. The survey was online available for approximately three weeks. When posting and distributing the survey via social media I also added a short text urging people to help me by filling in the short questionnaire. I also mentioned that I was looking for football supporters and that there was the possibility of winning a gift card from Bol.com with the value of 15 euros.
The order of the paintings in the questionnaire and also if they were made by an Ajax supporter or Feyenoord supporter, were randomized. The questions belonging to the
Collective Self-Esteem subscale were not randomized and had the same order for every participant.
My hypothesis was that participants attribute more talent to an in-group member than to an out-group member. My other hypothesis was that this favoritism would arise from a group-based self-esteem because of striving for a positive social identity. Following from these hypotheses I expected Ajax fans to attribute more talent to the Ajax painters than to the
Feyenoord painters and expected Feyenoord fans to attribute more talent to the Feyenoord painters than to the Ajax painters.
Data Analysis
Group membership (in-group/out-group) of the painter was the independent within-subjects variable and the group membership of the participant the independent between-subjects variable. The dependent variable was talent attribution. To see whether supporters of Ajax identify more with Ajax supporters than with Feyenoord supporters and the other way around I used a factorial mixed ANOVA (repeated measures) to conduct this manipulation check. The main analysis was also performed using a factorial mixed ANOVA. For the moderation of the main effect by group-based self-esteem an ANCOVA was performed.
Results
Checks and Data Screening
The participant’s group membership variable was not normally distributed for the Ajax group (p < .05) according to the Kolmogorov-Smirnov test. This means the assumption of normality has been violated. This is a limitation for my study but I decided to still do the analysis because in SPSS there is no non-parametric test available for a mixed ANOVA. Levene’s test for homogeneity was not significant for both groups (p >.05). The interaction of identification and fandom/group membership was significant and positive (F (1, 56) = 21.694;
p =. 00; partial η2 = .279). Ajax supporters identified more with other Ajax supporters than
with Feyenoord supporters and Feyenoord supporters identified more with Feyenoord
supporters than with Ajax supporters (see Figure 1 & Table 1). There was no main effect for fandom (F (1, 56) = .996; p = .323; partial η2 = .017), meaning that there was no significant difference in degree of identification between the two groups (Ajax/Feyenoord).
Figure 1.
Interaction Between Identification and Fandom.
Table 1.
Means and Standard Deviations for Identification.
Identification Supporter Mean Std. Deviation
Ajax Ajax 3.88 1.82
Feyenoord 2.71 1.60
Feyenoord Ajax 2.57 1.41
Factorial Mixed ANOVA
Again, unfortunately the assumption of normality had been violated for the Feyenoord group (p <. 05). The other assumptions had been met. There was no interaction between group membership of the participant and talent rating of the painter (F (1, 61)= .004; p =. 952; partial η2 =. 00). There was no significant main effect either (F (1, 61)= .011; p =. 916; partial η2 =. 00), meaning Ajax supporters did not rate the Ajax painter significantly higher than the Feyenoord painter, the same goes for the Feyenoord supporters. For the means and standard deviations of the talent rating of the painter by the supporters see Table 2.
Table 2.
Means and Standard Deviations for Talent Rating
Painter Supporter Mean Std. Deviation N
Ajax Ajax 4.24 1.19 51
Feyenoord 4.33 1.33 7
Feyenoord Ajax 4.31 1.10 51
Feyenoord 4.24 1.51 7
Moderation by Group-Based Self-Esteem
Finally, I performed an ANCOVA to look at the effects of the covariate group-based self-esteem. In this analysis talent rating was the outcome variable, participant fandom/group membership was the between independent variable and painter fandom/group membership was the within independent variable. Levene’s test of equality of error variances is not significant (p> . 05) which means the assumption has not been violated. There was no
significant interaction (F (1, 55)= .563; p = .456; partial η2 = .010). There was no significant main effect found for participant fandom/group membership (F (1, 55)= .001; p = .001; partial η2 = .00); meaning participants did not rate painters belonging to their own group higher when their self-esteem was more group based.
Discussion
Conclusion
I found that there was an interaction between fandom/group membership and identification. The Ajax supporters identified more with other Ajax supporters than with Feyenoord supporter and the other way around There was no main effect, meaning that fandom did not affect identification. For the main analysis there was no significant effect found between fandom/group membership and talent rating. Meaning that participants did not rate painters from their own group significantly higher than painters from the other group. This shows that there was no in-group favoritism shown by participants in this study.
Caveats and Limitations
There were a few caveats and limitations in my study. One of the first things that might explain the results of my study, or the lack of significant results is the fact that I did not succeed in reaching the sample size. Because of using almost half of my previously calculated sample size my study might not be reliable enough to interpret results from. Even though the lack of a large sample size there were no outliers or other problematic participants so the data of the participants that I did manage to collect were reliable. The logical solution for this problem is to make sure there is enough time to collect data in further research concerning this topic.
Another limitation of my study concerns the distribution of participants between the groups. My data collection closed due to lack of more time but by that time only 7 people claiming to be Feyenoord supporters had filled in my survey. I had to use their data but the Feyenoord supporters were a large minority in my study. The data of only 7 Feyenoord supporters was compared to data of almost 8 times as much Ajax supporters. This is a limitation for obvious reasons; it is very hard to get useable results out of data that almost cannot be compared. In a next study it is recommended to try and reach the same amount of participants in each group. Since my recruitment plan involved letting people know about my study via my own personal Facebook account it was to be expected that people from my own city would fill in the survey first, because I naturally know more people supporting Ajax than that I know people supporting Feyenoord. In a next study I would handle this in another way, making sure that I was reaching as much people from one group as I was doing from the other.
People that had filled in my survey reported that they had found it obvious that the study was searching for links between attributing talent or liking the painter and their fandom. Before starting my actual study I had done a small pilot study and had let some people fill in the first version of the survey, after receiving the comments that they found the survey very transparent I added a few questions to a survey, asking participants not only for their talent rating of the painter but also some other questions about the painting that were supposed to keep participants in the dark. I think this helped but it did not completely solve the problem. Because I do not know how many participants understood how much of my survey I cannot conclude that the survey was bad but it is not good that participants came back with this comment. A solution would be to change the survey a little but it can also be expected that the problem will solve itself when having a larger sample size.
Another caveat concerns the randomization of questions. It might be useful to
randomize the questions from the subscale of the Collective Self-Esteem Scale since the order in which the questions are shown might influence the reliability of the scale. Also the order in which the talent rating questions are presented should be randomized in future research. Because the questions ask people for their appreciation of a painting the order might influence how they think about the talent of the painter. For example rating the painting as giving you a happy feeling might change the way you think about the painter’s talent, randomizing the order of the questions can control for this effect.
Lastly, it is needed to mention the violation of the assumption of normality in both the manipulation check analysis as in the main analysis. It is conventional to instead conduct a non-parametric test in the case of violation of assumptions. In this case doing non-parametric tests was not an option because SPSS does not provide these. Therefore I continued with the parametric test. It is imaginable that this might have had consequences for my results.
Implications
The results of this research are implying that current theories regarding the subject of group favoritism are insufficient in explaining my results. In this study I researched in-group favoritism in talent attribution expecting to see in-in-group favoritism because of in- group-based self-esteem. Social Identity Theory supported this expectation. In this study I have not been able to support this theory but because I did not even find any type of favoritism I also cannot alter the theory. Because of the limitations in my study I do not find it reasonable to say Social Identity Theory is an overrated theory but it does need to be researched more in a study with less limitations. It also cannot be concluded that group –based self-esteem plays any role in in-group favoritism. I found that group based self-esteem had no effects.
Future Research
Future research concerning the topic of in-group favoritism will have to clear up whether this study found no results because of problems with the sample or because of other reasons. Furthermore, it should be studied whether people do show favoritism when being provided with a more reliable survey. Also because of lack of effect of group based self-esteem in this research this should be studied in the future in order to answer more questions about the existence of in-group favoritism in talent attribution and the underlying mechanisms that play a role in this phenomenon.
Bibliography
Balliet, D., Wu, J., & De Dreu, C. K. (2014). Ingroup favoritism in cooperation: A meta-analysis. Psychological Bulletin, 140(6), 1556.
Brewer, M. B. (1979). In-group bias in the minimal intergroup situation: A cognitive-motivational analysis. Psychological Bulletin, 86(2), 307.
Brewer, M. B. (1991). The social self: On being the same and different at the same time. Personality and social psychology bulletin, 17(5), 475-482.
Brewer, M. B. (1996). Intergroup relations (2nd ed.). New York: McGraw-Hill.
Ellemers, N., Kortekaas, P., & Ouwerkerk, J. W. (1999). Self‐ categorisation, commitment to the group and group self‐ esteem as related but distinct aspects of social
identity. European journal of social psychology, 29(23), 371-389.
Gagné, F. (1985). Giftedness and talent: Reexamining a reexamination of the definitions.
Gifted Child Quarterly, 29(3), 103-112
Luhtanen, R., & Crocker, J. (1992). A collective self-esteem scale: Self-evaluation of one's social identity. Personality and Social Psychology Bulletin, 18, 302-318.
Messi named best player in Europe for second time. (2015, 27 augustus). Opgehaald van http://www.uefa.com/uefachampionsleague/news/newsid=2274915.html
Mullen, B., Brown, R., & Smith, C. (1992). Ingroup bias as a function of salience, relevance, and status: An integration. European Journal of Social Psychology, 22(2), 103-122. Oakes, P. J., & Turner, J. C. (1980). Social categorization and intergroup behaviour: Does
minimal intergroup discrimination make social identity more positive?. European
Perdue, C. W., Dovidio, J. F., Gurtman, M. B., & Tyler, R. B. (1990). Us and them: social categorization and the process of intergroup bias. Journal of Personality and Social
Psychology, 59(3), 475.
Tajfel, H. (1981). Human groups and social categories: Studies in social psychology. Cambridge: Cambridge University Press.
Tajfel, H., Billig, M. G., Bundy, R. P. & Flament, C. (1971). Social categorization and intergroup behavior. European Journal of Social Psychology, 1, 149-178.
Tajfel, H., & Turner, J. C. (1979). An integrative theory of intergroup conflict. The social
psychology of intergroup relations, 33(47), 74.
Turner, J. C., Brown, R. J. & Tajfel, H. (1979). Social comparison and group interest in ingroup favouritism. European Journal of Social Psychology, 9(2), 187-204.
Voci, A. (2006). The link between identification and in-group favouritism: effects of threat to social identity and trust-related emotions. British Journal of Social Psychology, 45, 265-284.
