Experiment 1 method and results

In Experiment 1, participants were assigned to one of the two arbitrary groups, saw repeated and non-repeated positive and negative stereotypical statements about their in-group and out-group, and evaluated stereotypical statements about both groups as “true” or “false”. We predicted that participants would rate repeated stereotypical statements about their in-group more frequently as “true” than any other stereotypical statements. Materials Dot Estimation Task. For minimal grouping purposes, we used Ratner and Amodio’s (2013) Numerical Estimation Style Test (NEST). NEST includes 3 practice and 10 test trials. Clusters of black dots (consisting of 98-200 identical dots) were presented in the white background followed by a number estimation question in each trial. Stereotypes. We selected traits from Anderson (1968) and translated these traits to Turkish. An English language expert checked Turkish versions of the traits. Three Turkish language experts rated the top 200 personality traits that have the highest meaningfulness score in the original study. These three experts agreed on 92 positive traits and 84 negative traits. Among those, for positive stereotypes, we selected the top 40 traits having the highest likeability rates, and for negative stereotypes, we selected 40 traits having the lowest likeability traits. For stereotypical sentences, we used a very simple sentence structure that is the combination of group names (over-estimators or under-estimators) and the traits. We chose to keep the sentences as simple as the real stereotypes about real groups for face validity. Identification Measures. We followed Cameron (2004) for identification measures by including three dimensions of social identity; centrality, in-group affect, and in-group ties. Therefore, we used a four-item measure (“I identify myself with other over-estimators/under-estimators”, “I see myself as an over-estimator/under-estimator”, “I am glad that I am an over-estimator/under-estimator”, “I have strong ties with other over-estimators/under-estimators”). We used a 7-point Likert scale for these items indicating 1= “totally agree” and 7= “totally disagree”. We also asked participants to choose the best option showing their relationship with their in-group on the Pictorial Item of Identity Fusion Scale (Swann et al. , 2009). Participants. 96 undergraduate students (73 female, 22 male, 1 unspecified; Mage= 21, SDage= 1. 47) from various departments at Duzce University and Abant Izzet Baysal University participated in the study. They participated either voluntarily or for course credit. Only one student was omitted from the data because s/he gave uniform answers showing that s/he did not pay attention to the task. Design. Participants were assigned to one of the two groups randomly (i. e. , “over-estimators” vs “under-estimators”). All participants evaluated stereotypes from eight categories in a 2x2x2 within-participants design: repetition (repeated vs. new), group (in-group vs. out-group), valence (positive vs. negative). Procedure. The participants took part individually in the experiment. The experimenter welcomed, read the consent form, made the participant sit in front of the computer and after taking the consent she began the experiment on SuperLab software. Participants took the experiment as follows; minimal group assignment, statement evaluation, and identification measurements with demographics. First, participants had three practice and ten test trials for the group assignment. In each trial, they saw dots for 5 s and we asked them to estimate how many dots they have seen without time limitation. Participants waited for 20 s by thinking that the program was calculating their results. A false feedback appeared on the computer screen saying “According to the numerical estimation test you are overestimator/underestimator”. Second, the statement exposure and evaluation phase started. At the beginning of the exposure phase, they were instructed to read 40 statements and try to remember them. They randomly saw statements from four categories (in-group positive, in-group negative, out-group positive, out-group negative). Each statement appeared for 2. 5 s and a fixation dot appeared for 1 s. After that, participants judged 20 new and 20 old statements from the above-mentioned four categories in a binary fashion (Unkelbach, 2007). After truth judgments, they evaluated the statements as “seen” or “not seen” in the first phase for manipulation check. Third, identification with the group was measured through 4 items and Pictorial Identity Diffusion Scale. After demographics and two political orientation questions, the experimenter provided the information that the grouping information was fake, debriefed the purpose of the experiment, and thanked participants. Results Truth Evaluations. We analysed means of truth ratings with a 2 (repetition: repeated vs non-repeated) X 2 (group: in-group ve out-group) X 2 (valence: positive vs negative) repeated measures ANOVA. Participants are more likely to evaluate repeated stereotypes (M = 0. 50, SD= 0. 18) as true compared to non-repeated stereotypes (M = 0. 36, SD= 0. 17), F (1, 94) = 48. 448, p<. 001, ηp² =. 340) as expected. In addition, ANOVA showed a main effect of valence. Participants are more likely to evaluate positive stereotypes (M =. 50, SD= 0. 20) as true compared to out-group stereotypes (M = 0. 36, SD= 0. 18), F (1, 94) = 32. 157, p<. 001, ηp² =. 255. Furthermore, participants are more likely to evaluate in-group stereotypes (M = 0. 46, SD= 0. 18) as true than out-group stereotypes (M = 0. 40, SD= 0. 17), F (1, 94) = 11. 030, p=. 001, ηp² =. 105. As ingroup favouritism suggests, The group effect was larger for positive statements compared to negative statements, F (1, 94) =28. 581, p<. 001, ηp² =. 233. Participants judged positive stereotypes (M = 0. 59, SD= 0. 28) as more true than negative stereotypes (M = 0. 33, SD= 0. 19) when the stereotype is about their in-group whereas they did not judge positive stereotypes (M = 0. 41, SD= 0. 25) as more true than negative stereotypes (M = 0. 40, SD= 0. 22) when the stereotype is about their out-group. There was no significant interaction of repetition, group, and valence, F (1, 94) =2. 087, p=. 152. Reaction Times. The measured reaction times were checked if they are at three standard deviations above or below the average of the participant’s reaction times. A 2 (repetition: repeated vs non-repeated) X 2 (group: in-group ve out-group) X 2 (valence: positive vs negative) ANOVA was used for analysis of the data. Results indicated that participants responded faster to in-group statements (M = 2770) than out-group statements (M = 2948), F(1, 94) =12, 189, p<. 05, ηp2 =. 114. In addition, participants responded slightly faster to positive statements (M = 2813) than negative statements (M = 2904), (F(1, 95) =3, 217, p=. 076). However, the effect of repetition was not significant (F(1, 94) =1, 093, p>. 05). Manipulation check. We checked whether participants could recognize repeated statements. A paired sample t-test showed participants responded to repeated statements ``Yes, I saw it” (M = 56%) more frequently than “No, I didn't see it” (M = 34%).