A Multilevel Examination of Stereotype Accuracy among Campus Groups: The Role of Group Entitativity

Abstract

As individuals move through the world, they form impressions of other groups of people. These impressions can be referred to as stereotypes (Jussim, 2012). Lay people often view the use of stereotypes negatively, relating them to prejudice and discrimination. Certainly, individuals’ stereotypes of groups may influence the extent to which they engage in discriminatory behaviors (Jussim, Cain, Crawford, Harber, & Cohen, 2009). However, as Jussim et al. and others (Judd, Ryan, & Park, 1991; Lee, McCauly, & Jussim, 2013; Ryan, 2002) note, stereotype accuracy can be empirically evaluated. Additionally, it may be useful to understand how accurate individuals’ stereotypes of group are for a few reasons. First, understanding how stereotypes begin and develop informs other research on intergroup perceptions and relations. Second, even stereotypic judgments that have negative consequences may be useful in certain circumstances just as other accurate information is helpful. Finally, as new group members are socialized, it may be useful to understand how group-level characteristics influence new members’ perceptions of those groups.

The purpose of this study is to assess the new group members’ judgments of their groups across two waves of data approximately six months apart as a function of actual group characteristics: stereotypicality (i.e., the extent to which group members have more stereotypic versus counterstereotypic traits), positivity (i.e., the extent to which group members have more positive than negative traits), and entitativity (i.e., how “group-y” a group is). We first intend to establish the new group members’ perceptions of their ingroups are accurate and that other group-level characteristics, such as entitativity influence new group members’ perceptions of their ingroups at Wave 2. Given that individuals’ perceptions of their groups change as they are initiated into the group (Moreland & Levine, 1982; 1989), we also examine how changes in perceived positivity and stereotypicality over time depend on group-level characteristics.

Stereotype Accuracy

Stereotypes can be evaluated along three different dimensions, often referred to as the EPA model of stereotypes (Lee et al., 2013). The E refers to the evaluation of stereotypes, that is, whether they are positive or negative in valence. The P refers to the potency of stereotypes, that is how automatically they are activated, and A referred to the accuracy of stereotypes, that is how much individuals’ perceptions of groups correspond to actual group members’ characteristics. All of the above dimensions may be evaluated along a continuous scale, that is, for example, researchers may examine the extent to which individuals’ perceptions of groups are accurate or inaccurate. The present study examines two of the above components: evaluation (i.e., valence), and stereotype accuracy.

Stereotype accuracy, in particular, can be evaluated through two means: discrepancy scores and correlations (Lee et al., 2013). Neither technique is better or worse than the other; they simply provide different types of information. Discrepancy scores are calculated by assessing a stereotypic belief and comparing it to some sort of accuracy criteria. For example, an individual might hold the belief that most basketball players are tall. The accuracy of this stereotypic belief could be assessed by asking the individual to estimate the average height of an NBA player and assessing how close that is to the average height of players in the NBA. The difference between the two scores would indicate how (in)accurate that stereotype was.

Correlations can be assessed by asking individuals to rate their impressions of groups along stereotypic dimensions and correlating those impressions with self-ratings from the actual group members. For example, a researcher might ask individuals to estimate the how studious the average graduate student is. They might then correlate these ratings with the average rating of studiousness among a group of graduate student. The strength of the correlation then indicates the accuracy of those beliefs. The latter technique is used to assess accuracy in this study.

It is important to note that in discussing stereotype accuracy, researchers assess the extent to which descriptive stereotypes of groups, that is, for example, the base rate of individuals who possess a certain trait, are accurate (Lee et al., 2013). Stereotype accuracy is not possible to assess for prescriptive stereotypes (i.e., individuals’ perceptions of how groups ought to behave). Additionally, assessing the accuracy of stereotypes does imply that characteristics conferred upon a group are right, or fair, or socially just. As Jussim, et al. (2009) note, “the accuracy of stereotypes is an empirical question, not an ideological one (p. 199). That is, it is possible to empirically assess the extent to which stereotypes correspond to actual group traits without making a moral judgment about whether that stereotype is “right.”

Indeed, studies suggest that, by and large, stereotypes of group tend to be accurate. Measures of stereotype accuracy are among the strongest effects in social psychological literature. For example, consensual stereotypes of gender or racial groups generally correlate to group characteristics at or above .50 (Richard, Bond, & Stokes-Zoota, 2003). Furthermore studies of other types of groups, such as sororities, (Ryan & Bogart, 1997, 2001) suggest that individuals perceptions of groups are generally accurate. I therefore expect that new group members’ perceptions of the groups that they join will be accurate.

H1: Established group members’ self- ratings along stereotypic versus counterstereotypic traits at Wave 1 will predict newcomers impressions of the group at Wave 2.

H2: Established group members’ self-ratings along positive versus negative traits at Wave 1 will predict newcomers impressions of the group at Wave 2.

Stereotypes and Socialization

Individuals perceptions of groups in which they have membership (i.e., ingroups) are different from those in which they do not have membership (i.e., outgroups). For example, individuals judge their ingroups as being more variable than outgroups (Jones, Wood, & Quattrone, 1981; Linville, Fischer, & Salovey, 1989; Park & Judd, 1990; Park & Rothbart, 1982; Park, Ryan, & Judd, 1992) and more positive than outgroups (Brewer, 1979; Tajtel, 1969). It is also possible that new group members may view their ingroups differently from more established group members. For example, Moreland and Levine (1982; Levine & Moreland, 1994) suggested that individuals’ impressions of their own groups change as a function of group socialization.

Socialization is the initiation process through which new group members become full group members and includes five distinct phases: investigation, socialization, maintenance, resocialization, and remembrance (Moreland & Levine 1982; 1989). During socialization, new group members are generally highly motivated to view the group positively and learn about the group (Ryan, Robinson, & Hausmann, 2004). Socialization may also be a time during which new group members have negative experiences, as groups (e.g., organizations, sororities, fraternities) may not live up to the high expectations that new group members formed during their initiation (Wanous, 1997; Wanous, Poland, Premak, & Davis, 1992). Indeed in a study of new sorority pledges, Ryan & Bogart 1997) found that new members’ positive impressions of their ingroups decreased over time.

H3: Perceptions of positivity will decrease from Wave 1 to Wave 2.

There is also evidence that perceptions of group stereotypicality may also change during the socialization process, although to my knowledge, no study has examined the extent to which group characteristics affect individual groups members’ perceptions of their ingroups. Ryan and Bogart (1997) found that stereotypicality decreased during group socialization, and that individuals perceived their ingroups to be increasingly more disbursed over time. This is consistent with evidence suggesting that individuals tend to use individuating information to make judgments of targets when it is available (Crawford, Jussim, Madon, Cain, & Stevens, 2011).

H4: Perceptions of group stereotypicality will decrease from Wave 1 to Wave 2.

Few studies have examined the extent to which full group members’ traits influence new group members’ stereotypes of their ingroups. However, there is reason to expect that group-level characteristics will influence perceptions of positivity and stereotypicality over time. For example, Ryan and Bogart (2001) in a re-analysis of Ryan and Bogart (1997) found that despite decreases in general stereotypicality, new members’ stereotypes of group members were increasingly accurate over time, suggesting that the tendency for groups to be viewed less stereotypically over time may be weaker among more stereotypic groups. It is also possible that as newcomers often have some negative experiences during socialization that groups who are themselves high in positivity (that is, groups who are self-rated as having more positive than negative traits) may buffer the extent to which positive perceptions decrease.

H5: There will be a Wave X Group Stereotypicality cross-level interaction indicating that the tendency for groups to be seen less stereotypically over time will be less pronounced among more stereotypic groups.

H6: There will be a Wave X Group Positivity cross-level interaction indicating that the tendency for groups to be seen less positively over time will be less pronounced among groups who rate themselves as having more positive than negative traits.

Entitativity

Another way in which groups may vary is the extent to which they are seen as a social unit (Licket et al., 2000), or in other words, how entitative they are. Campbell (1958) defined entitativity as groups that had a common fate, similarity, proximity, and completed boundaries, also referred to as pregnance. Individuals readily distinguish between entitative groups (e.g., a sports team) and non-entitative groups (e.g., individuals waiting at a bus stop; Licket et al., 2000). Further evidence also suggests that consistent with Campbell’s (1958) conceptualization, entitative groups exhibit similarity (i.e., homogeneity; Crawford et al., 2002), tend to have common goals and outcomes (Clark & Wegener, 2009), are highly interdependent (Crump, Hamilton, Sherman, Lickel, & Thakkar, 2010) and are organized (Newheiser, Sawoka, & Dovidio, 2012).

Entitativity also impacts individuals perceptions of groups. For example, individuals may see highly entitative outgroups more stereotypically (Brewer & Harasty, 1996; Welbourne, Harasty, & Brewer, 1997). Entitativity may also impact perceptions of group positivity. For example, Dang, Lui, Ren, and Su (2018) found that individuals perceptions of groups were more polarized for entittaive groups that for non-entitative groups. That is, positive impressions of entitative groups become more positive and negative perceiptions of entitative groups become more negative. This is consistent with evidence suggesting that individuals may perceive more entitative groups as more competent (Yzerbyt,  Kervyn, & Judd, 1998) and may view entitative groups with cooperative intentions more positively (Castano, Yzerbyt, & Bourguignon, 2003). However, if entitative outgroups who are seen as having negative intentions were regarded more suspiciously (Newheiser et al., 2012).

Most studies that examine group entitativity examine individuals perceptions of entitative outgroups, that is, individuals are asked to judge a target group with whom they do not share group membership. However, few studies have examined individuals perceptions of entitative ingroups. Indeed, Yzerbyt, Castano, Leyens, & Paladino (2000) called for more studies investigating ingroup members’ perceptions of entitative ingroups. Castano, Yzerbyt, & Bourguignon (1998) found that individuals tend to identify more strongly with highly entitative groups, suggesting that individuals may regard their ingroups more positively if they are highly entitative.

H7: There will be a cross-level main effect of group entitativity on perceived positivity at such that more entitative groups will be seen more positively at Wave 2.

Interestingly, although individuals may regard entitative outgroups more stereotypically (Brewer & Harasty, 1996), the opposite might be true of entitative ingroups. Yzerbyt and Estrada (1999) found that in a group decision-making task, members of more essentialist groups were less likely to change their initial position. This is consistent with information suggesting that individuals are motivated to maintain a sense of uniqueness among ingroup members (Brewer, 1991). Additionally, it is also consistent with evidence suggesting that individuals are likely to rely on individuating information when it is available (Crawford et al., 2011). Thus, as individuals in highly entitative groups may be likely to try to maintain a sense of uniqueness, and others in the ingroup may be highly attuned to this individuating information, highly entitative ingroups may be seen less stereotypically by their ingroup members.

H8: There will be a cross-level main effect of group entitativity on perceived stereotypicality such that more entitative groups will be seen less stereotypically at Wave 2.

Furthermore, the socialization processes of entitative groups, who are highly organized and cohesive, are likely different from the socialization process in less entitative groups. Due to the highly organized nature of more entitative groups (Newheiser et al., 2012) and the fact that individuals in highly entitative groups highly identify with those groups (Castano, Yzerbyt, & Bourguignon, 1998) highly entitative groups may see themselves more positively at Wave 2, the tendency for groups to see themselves less positively over time may be weaker among more entitative groups. Conversely, the tendency for groups to see themselves less setereotypically over time may similarly be stronger among more entitative groups.

H9: There will be a Wave X Group entitativity cross-level interaction, such as the tendency for groups to see themselves less positively over time may be weaker among more entitative groups.

H10: There will be a Wave X Group entitativity cross-level interaction, such as the tendency for groups to see themselves less stereotypically over time may be stronger among more entitative groups.

Method

Participants. New members (41.7% female, 58.3% male) of 11 campus groups (See Table 1 at a Midwestern university completed questionnaires at two time points, approximately six months apart (Wave 1, N = 149; Wave 2, N = 127). Most (85.23%) participants who participated in Wave 1 (W1) were retained at Wave 2 (W2), consistent to the retention rate of Ryan and Bogart (1999, 2001). Participants who dropped out of the study did so primarily because they dropped out of their campus groups. Most participants were White (90.01%), followed by Asians (3.97%), Blacks (2.65%), and Latinx (2.65%) participants.

Full members of campus groups participated by providing self-ratings of attribute dimensions at W1 (n = 241). Full member’s self-ratings on attribute dimensions served as the accuracy criteria for perceptions of groups. Most (95%) group members provided accuracy criteria. Percentages of group members who provided ratings ranged from 82% to 100%. New member participants were paid $10 each for each wave in which they participated. Each campus group was paid $100 for its participation.

Procedure. Newcomers completed two questionnaires during the Fall (W1) and Spring (W2) of the academic year approximately six months apart. Both questionnaires included a percentage estimation task (Judd et al., 1991; Park & Judd, 1990; Ryan & Bogart, 2001). In this task, participants were asked to indicate the percentage of group members who possessed 16 attribute dimensions on a 0 to 100% scale. The 16 attribute dimensions for each group varied with respect to stereotypicality (stereotypic vs. counterstereotypic) and valence (positive vs. negative). Attributes were selected from interviews that were conducted towards the end of the previous academic year. More specifically, full members of each group were interviewed separately to identify positively and negatively valenced attributes that they believed most students (i.e., group members and non-members) would consider either characteristic or uncharacteristic of their group. This process yielded for each participant group a set of 16 attributes with four attributes in each of the four stereotypicality (stereotypic vs. counterstereotypic) X valence (positive vs. negative) cells. Examples of attributes for each group are reported in Table 1.

Perceived stereotypicality measures were created by subtracting the mean perceived percentage of counterstereotypic traits from the mean percentage of stereotypic traits for each group. Measures of perceived positivity were similarly created by subtracting the mean perceived percentage of negative traits from the mean percentage of positive traits for each groups.

It seemed reasonable that individuals involvement in their groups might influence the extent to which they viewed the group positively and stereotypically and that entitativity might be confounded with involvement. I therefore sought to control for the effects of individual involvement in groups over time. At both waves new group members completed two items assessing the average amount of time they spent with their group members and the average amount of time they spent on group activities per week: 1-2 hours, 3-5 hours, 5-10 hours, 10-15 hours, or more than 15 hours. This measure was subsequently treated as a 5-point scale, with higher valued indicating greater group involvement.

At Wave 1, new group members also completed a five-item measure of perceived group entitativity (5 items, α = .86) on a 1 (Not very much at all) to 7 (A great deal) scale. Example items were, “How cohesive is your group?” and “How much unity do you think the members of your group feel?” ICCs suggested that aggregating to the group level was appropriate, ICC(2) = .375, χ²₍₁₀₎ = 110.83, p < .001.

Finally, full group members completed self-ratings of the same 16 attribute dimensions that new group members rated at W1, using 20 point scales ranging from (1 = Not at all) to (20 = Very much) scales. Accuracy criteria were created by first calculating the percentage of group members in each group who possessed each attribute. Those items were then averaged to create composite measures of the average percentage of individuals in each group who possessed stereotypic, counterstereotypic, positive, and negative traits. The measure of actual group stereotypicality was created by subtracting the average percentage of group members who possessed counterstereotypic traits from the average percentage of group members who possessed stereotypic traits. Actual group positivity was similarly calculated by subtracted the average percentage of group members who possessed negative traits from those who possessed positive traits.

Results

Means and Correlations among Measures

Means and standard deviations of perceived stereotypicality, positivity, and involvement at W1 and W2 are included in Tables 2 and 3, respectively. Groups were generally perceived as more stereotypic than counterstereotypic and more positively than negatively at both waves, suggesting that the manipulation of the attribute dimensions was successful across groups. Correlations among perceived stereotypicalty, positivity, and group involvement over time (See Table 4) indicated that groups who were perceived more stereotypically were also perceived more positively at both waves. Additionally, across-wave correlations for positivity, sterotypicality, and involvement indicated that the constructs were relatively stable over time. Interestingly, involvement was unrelated to either outcome of interest.

Means of group-level data (See Table 5) indicated that full-group members rated themselves as having more stereotypic than counterstereotypic traits and more positive than negative traits, again suggesting that the manipulation of traits was effective. Groups also appeared to be highly entitative overall. Correlations among Level 2 variables were not significant, likely due to the small number of groups (n = 11).  However, strong non-significant trends indicated that groups who were themselves highly stereotypic also had more positive than negative traits, r(9) = .42, p = .194. High entitative groups were also more stereotypic, r(9) = .48, p = .130 and had more positive than negative traits, r(9) = .43, p = .183.

Multilevel Analyses

To determine whether multilevel analyses were appropriate, I estimated “unconditional” or intercepts-only models for perceived and positivity. In both cases, significant ICCs indicated that multilevel modeling was appropriate for perceived group stereotypicality (ICC(2) = .278, χ²₍₁₀₎ = 85.99, p < .001) and perceived group positivity (ICC(2) = .256, χ²₍₁₀₎ = 70.62, p < .001).

Separate series of multilevel models were used to test perceptions stereotypicality (Table 6) positivity (Table 7), respectively as a function of Wave, involvement over time, and actual group characteristics. The process for testing perceived group stereotypicality is described below. In Model 1, the main effects Wave (coded as W1 = -1, W2 = 0) and involvement over time (grand mean centered) were added at Level 1 as random predictors of perceived group stereotypicality. In Model 2, the main effects of actual group stereotypicality and entitativity (both grand mean centered) were added on perceived stereotypicality at W2. In Model 3, cross-level interactions were added testing the Gruup Stereotypicality X Wave and Entitativity X Wave interactions. The procedures for testing perceived group positivity were the same, except measures of actual group positivity (grand-mean centered) and perceived group positivity were used in place of actual group stereotypicality and perceived group stereotypicality, respectively. Figures 1 and 2 summarize the associations tested in the stereotypicality and positivity analyses, respectively.

Stereotypicality analyses

The results of the first model indicated that adjusting for involvement over time and perceived stereotypicality at W1, groups were perceived as more stereotypic than counterstereotypic at W2. Furthermore, perceived stereotypicality at Wave 2 varied significantly across groups, χ²₍₁₀₎ = 56.85, p < .001.Although neither Level 1 effect was significant, adding them improved model fit Δ χ²₍₅₎ = 14.98, p > .05 and decreased prediction error PRPE = 4.06%. Indeed, the effect of Wave suggested that consistent with expectations, controlling for involvement, new members viewed their groups somewhat less stereotypically at W2, partially supporting H4, although the effect was not significant.

Both cross-level main effects of actual group stereotypicality and entitativity were significant in Model 2. Controlling for group entitativity, groups who were themselves highly stereotypic were perceived by new group members as highly stereotypic at Wave 2, consistent H1, and the stereotype accuracy literature more generally. Additionally, consistent with expectations (H8), more entative groups were perceived less stereotypically by their group members at Wave 2, holding actual group stereotypicality constant. Both variables accounted for 99.71% of the variability in perceived stereotypicality at W2 and significantly improved model fit Δ χ²₍₂₎ = 46.00, p < .001.

Only one of the cross-level interactions in Model 3 was significant. The Entitativity X Wave interaction was not significant, inconsistent with expectations that groups that the tendency for newcomers to see their ingroup as less stereotypic over time would be stronger among more entitative groups (H10). However, a Group Stereotypicality X Wave interaction indicated that there was a polarizing effect of actual group stereotypicality such that highly stereotypic groups were viewed more stereotypically over time, whereas groups low in actual stereotypicality were perceived somewhat less stereotypically over time (See Figure 3). This effect was somewhat consistent with H5, suggesting that the tendency to perceive groups less stereotypically over time might be weaker among groups who were themselves more stereotypic. However, it is unclear why highly stereotypic groups might be perceived more stereotypically over time. In any case, the addition of both interaction terms significantly improved model fit Δ χ²₍₂₎ = 5.664, p < .05, PRPE = 96.6%.

Positivity

Controlling for changes in positivity over time and group involvement, groups viewed themselves more positively than negatively at Time 2, and this effect varied significantly across groups, χ²₍₁₀₎ = 29.14, p < .001. However, neither the effect of involvement nor the effect of Wave was significant, nor did they improve model fit, Δ χ²₍₅₎ =7.60, p > .05, PRPE = 0.52%, inconsistent with the expectation that groups would be perceived less positively over time (H3) and with findings from Ryan and Bogart (1997, 2001).

Main effects of group positivity and entitativity were added on perceived group positivity in Model 2. There was a significant effect of actual group positivity. Groups who themselves were high in actual positivity were perceived more positively than negatively by new members at W2, consistent with H2. However, there was no effect of entitativiy on perceived positivity, inconsistent with expectations that more entitative groups would be viewed more positively by new group members (H7). Additionally, the addition of both effects degraded model fit (Δ χ²₍₂₎ = -3.23) and increase prediction error (PRPE = -13.23%).

I assessed a separate model only assessing the main effect of group positivity on perceived group positivity to examine whether the increase in prediction error was due to the effects of group entitativity. This model similarly indicated that higher group positivity was marginally associated with higher perceived positivity at W2, b = 0.67, SE = 0.31, t(9) = 2.17, p = .058. However, this effect also increased prediction error over Model 1 (PRPE = -3.24%) and degraded model fit (Δ χ²₍₁₎ = -3.08). It is difficult to know what to make of these effects given that there was no problematic multicollinearity and no indication of heteroscedsasticity. It is possible these effects may be due to nonlinear associations among predictors.

In any case, cross-level interactions testing the Group Positivity X Wave and Group Entiativity x Wave itnteractions in Model 3 revealed two significant interactions. A Group Positivity X Wave interaction indicated that groups who were high (vs. low) in positivity were perceived less positively over time (See Figure 4), inconsistent with the expectation that decreases in perceived positivity over time would be attenuated among groups higher in positivity (H6). In retrospect, the effect is likely due to regression to the mean. In any case, a marginal cross-level Entitativity X Wave interaction indicated that perceptions of group positivity were relatively stable over time for highly entitative groups, whereas groups low on entitativity were perceived less positively over time by their group members, consistent with H9 (See Figure 5).

Discussion

The purpose of the present study was to examine changes in new group group members perceptions of their ingroups over time as well as to investigate the effect of group characteristics on new members’ perceptions of their groups both in terms of how accurate new members’ perceptions of their groups were and in terms of how established group member’s characteristics influenced changes in new members’ perceptions of their ingroups during the socialization process.

Stereotypicality

Consistent with stereotype accuracy literature (Lee et al., 2013), we found that group members’ self ratings on stereotypicality and positivity predicted new members perceptions of their groups. These results are also consistent with Ryan and Bogart (1997, 2001), suggesting that stereotypes of groups such as campus groups and organizational groups are generally accurate. Additionally, the large size of the effect is consistent with other stereotype accuracy studies (e.g., Jussim 2012).

Researchers and the general public generally view stereotypes as inaccurate (Judd et al., 1991; Lee, et al., 2013, Ryan, 2002) and was ways to foment prejudice and discrimination (Jussim et al., 2009). However, as Jussim et al. (2009) note stereotype accuracy can be assessed empirically and should not be viewed as an ideological issue. It should be noted that stereotype accuracy does not imply that group-level characteristics apply to all individuals in the group. Of course, individuals in any group vary in the extent to which they exhibit different traits. However, individuals’ ability to accurately perceive groups in the absence of individuating information is important, and accurate stereotypes might be best conceived as reasonable perceptions of groups that can be further revised as individuating information is available (Ryan et al., 2000).

Analyses of new groups members’ changes in perceived stereotypicality and positivity over time indicated that consistent with expectations and with Ryan and Bogart (1997, 2001), groups were viewed somewhat less stereotypically over time. However, inconsistent with Ryan and Bogart, groups did not view themselves less positively over time, although the coefficient associated with Wave in the positivity analyses indicated that there was a nonsignificant trend in that direction. In retrospect, given that group members’ experiences with a group during socialization may influence their perceptions of group positivity (Moreland and Levine 1982; 1984), it is possible that differences in the socialization experiences across the groups may account for this nonsignificant result.

Interestingly, changes in perceived group characteristics over time depended on group-level stereotypicality but did not seem to depend on group-level positivity. The effects of positivity simply indicated a regression to the mean. However, the tendency for groups to be viewed somewhat less stereotypically over time was less pronounced among groups who were themselves high in stereotypicality. Indeed, groups who were highly stereotypic were perceived somewhat more stereotypically over time. It is difficult to speculate about the nature of these results. The divergent pattern for more versus less stereotypic groups may suggest that group members perceptions became more accurate over time. In other words, group members may have adjusted their judgments of the groups as a function of group characteristics – upwards for more stereotypic groups and downwards for less stereotypic groups. However, I was unable to test those effects in these data.

Entitativity

Consistent with expectations, groups who were more entitative were perceived more positively and less stereotypically at W2. This is consistent with literature indicating that individuals in more entatative groups may more strongly identify with the group (Castano et al., 1998) and may seek maintain their individuality (Brewer, 1991), suggesting that more individuating information may be readily available in more entitative groups (Crawford et al., 2011).

Interestingly, the Entitivity X Wave cross-level interaction was only significant for perceptions of group positivity but not for group stereotypicality. The tendency for groups to be viewed less positively over time was weaker among more entitative groups. It is possible that due to their similarity (Crawford et al., 2002) and the fact that more entitative groups have common goals (Clark & Wegener, 2009; Rydell & McConnell, 2005, & Welbourne, 1999), high group entitativity may decrease the probability that new group members have negative experiences during socialization. Furthermore, given that individuals highly identify with entitative ingroups (Castano et al., 1998) individuals in highly entiative groups may more readily downplay negative experiences during socialization among more entitative groups.

The lack of effects on perceptions of stereotypicality over time is unclear. It is possible that group entitativity affected group members’ perceived stereotypicality at Waves 1 and 2 in the same way, that is that groups were perceived less stereotypically at both Waves of time, but that it did not impact the changes in perceived stereotypicality over time. It is also possible that it may take longer than six months of changes in stereotypicality due to group entitativity to emerge. It remains for future research to address these questions.

Limitations and Future Directions

To my knowledge, this is one of the only studies to examine the effects of group level characteristics on new group members’ perceptions of their groups over time among naturally occurring groups. Such a design is high in ecological validity, but does not allow for the manipulation of group characteritics. As such, although I was able to establish temporal order, I am unable to make any causal conclusions about the effects of group characteristics on individuals’ perceptions of their groups.

Additionally, data were only gathered at two points in time over a relatively short time period. It is possible that the effects of socialization might be stronger if newcomers had more of an opportunity to learn about their groups that the effects of group characteristics on change over time might be stronger. Additionally, two time points does not allow for me to test potential curvilinear effects in the change in positivity or stereotpyicality over time, which has been suggested by other work (Ryan & Bogart, 1997, 2001).

Finally, I did not directly examine group members’ efforts to socialize their groups. Therefore, although it is certainly reasonable to expect that more entitative groups may socialize their ingroup members differently from less entitative groups, I was unable to assess or test those effects. Future studies might consider more directly measuring variables implicating in the socialization process to directly test the effects of group socialization on individuals perceptions of their groups.

Conclusion

New group members learn about their groups from existing group members. Those newcomer perceptions’ may be influenced by group characteristics, resulting in more accurate group stereotypes and changes in how group members are perceived over time. These effects allow researchers to better understand how impressions of groups are formed and how newcomers integrate themselves into already existing groups.

Appendices

Table 1

Campus Groups and Examples of Attributes as a Function of Attribute Stereotypicality and Valence

		Stereotypic			Counterstereotypic
Groups		Positive	Negative		Positive	Negative
Military organization		patriotic	cocky		compassionate	weak
Aviation club		attentive	too meticulous		easygoing	impractical
Frisbee club		outgoing	slackers		goal-oriented	uptight
Environmental club		eco-friendly	doom-sayer		into networking	greedy
Martial arts club		agile	withdrawn		enterprising	clumsy
Business club		career-minded	materialistic		liberal	impractical
Geography club		outdoorsy	dorky		competitive	materialistic
Chemistry club		analytic	introverted		personable	bossy
Academic organization		studious	nerdy		laidback	lazy
Business honors society		enterprising	pushy		empathic	introverted
Business fraternity		career-minded	self-serving		relaxed	touchy feely

Table 2

Means and Standard Deviations of Level 1 Variables at Wave 1

Group	n	Involvement	Perceived Stereotypicality	Perceived Positivity
Military organization	29	2.10 (0.60)	14.12 (16.86)	50.54 (17.13)
Aviation club	6	2.42 (0.73)	12.58 (13.35)	49.63 (17.46)
Frisbee club	11	1.36 (0.45)	18.09 (13.40)	51.14 (16.70)
Environmental club	13	1.08 (0.19)	16.55 (9.18)	56.90 (16.36)
Martial Arts club	5	1.90 (0.42)	17.63 (13.92)	70.50 (14.59)
Business club	11	1.00 (0)	19.20 (14.57)	50.16 (11.00)
Geography club	9	1.61 (1.22)	32.90 (14.11)	65.22 (18.41)
Chemistry club	8	1.19 (0.37)	3.50 (15.79)	21.66 (13.03)
Academic organization	27	2.44 (0.84)	20.47 (15.05)	49.46 (24.11)
Business honors society	16	1.34 (0.54)	23.84 (10.70)	49.91 (20.56)
Business fraternity	14	1.57 (0.55)	5.73 (11.58)	35.43 (14.09)

Note. Standard Deviations are provided in parentheses.

Table 3

Descriptive Statistics among Level 1 variables at Wave 2

Group	n	Involvement	Perceived Stereotypicality	Perceived Positivity
Military organization	22	2.32 (0.66)	17.39 (10.79)	54.40 (17.12)
Aviation club	4	1.63 (1.25)	12.47 (23.17)	46.00 (30.86)
Frisbee club	9	1.17 (0.25)	18.98 (14.73)	47.24 (25.41)
Environmental club	13	1.04 (0.14)	15.18 (14.38)	51.35 (14.41)
Martial Arts club	5	2.60 (0.96)	13.02 (9.33)	62.50 (21.51)
Business club	6	1.08 (0.20)	22.06 (16.24)	52.88 (17.88)
Geography club	8	1.93 (0.98)	35.55 (13.37)	67.41 (18.45)
Chemistry club	7	1.14 (0.24)	-5.07 (19.03)	25.35 (18.22)
Academic organization	26	2.38 (0.88)	12.73 (16.29)	42.18 (23.90)
Business honors society	14-15	1.37 (0.52)	29.29 (13.44)	48.77 (20.14)
Business fraternity	12	2.00 (0.85)	1.60 (11.99)	32.29 (24.30)

Note. Standard Deviations are provided in parentheses.

Table 4

Correlations among Level 1 Measures Across All Groups

Variable	1	2	3	4	5	6
Involvement
1. Wave 1
2. Wave 2	.55***
Perceived Stereotypicality
3. Wave 1	.04	.08
4. Wave 2	-.01	-.05	.70***
Perceived Positivity
5. Wave 1	-.01	.05	.49***	.33***
6. Wave 2	-.03	.10	.33***	.56***	.63***

Note. Wave 1 N = 149-150. Wave 2 N = 126-127.

***p < .001, **p < .01, *p < .05, †p < .10

Table 5

Means Group Level Stereotypicality, Positivity, and Entitativity

Group	N	Stereotpyicality	Positivity	Entitativity
Military organization	51 (29)	17.16	45.10	6.17
Aviation club	6 (6)	14.58	47.92	5.63
Frisbee club	6 (11)	6.25	27.08	3.80
Environmental club	14 (13)	14.29	50.00	5.52
Martial Arts club	8 (5)	15.85	47.54	5.52
Business club	13 (11)	11.30	28.61	4.21
Geography club	8 (9)	20.99	47.77	4.20
Chemistry club	8 (8)	0.89	28.57	3.99
Academic organization	57 (27)	7.55	54.29	3.99
Business honors society	38 (16)	20.14	32.01	4.73
Business fraternity	32 (14)	4.29	35.55	5.20

Note. Mean scores on stereotypicality, positivity, and entitativity for all campus groups. The number of new group members is indicted in parentheses. The number of full group members is indicated outside of the parentheses. Stereotypicality and Positivity ratings were provided by full group members, whereas judgments of entitativity were supplied by new group members.

Table 6

Relationships of Actual Ingroup Stereotypicality, Entitativity, and Newcomer Involvement to Newcomers’ Perceptions of Ingroup Stereotypicality

Predictor	Model 1			Model 2			Model 3
	b	SE	t	b	SE	t	b	SE	t
Intercept	15.80	3.64	4.336***	16.80	0.98	17.27***	16.70	0.82	20.34***
Wave	-0.98	1.77	-0.52	-0.75	1.66	-0.449	-0.95	0.98	-0.98
Involvement	-0.14	1.81	-0.77	1.50	1.41	1.064	1.29	1.46	0.89
Stereotypicality				1.64	0.11	14.744***	1.86	0.13	14.58***
Entitativity				-7.55	0.53	-14.27***	-6.67	0.70	-9.575***
Stereotypicality X Wave							0.49	0.11	4.25**
Entitativity X Wave							1.27	0.91	1.40

Note. Number of Level 2 Groups = 11.

***p < .001, **p < .01, *p < .05, †p < .10

Table 7

Relationships of Actual Ingroup Positivity, Entitativity, and Newcomer Involvement to Newcomers’ Perceptions of Ingroup Positivity

Predictor	Model 1			Model 2			Model 3
	b	SE	t	b	SE	t	b	SE	t
Intercept	48.46	3.54	13.68***	48.25	3.38	14.27***	48.37	3.10	15.63***
Wave	-1.61	1.59	-1.01	-1.77	1.54	-1.14	-1.62	0.95	-1.72
Involvement	1.40	1.86	0.75	0.62	1.71	0.36	0.53	1.83	0.29
Positvity				0.67	0.28	2.41*	0.49	0.30	1.65
Entitativity				-0.06	3.88	-0.02	2.19	4.26	0.52
Positivity X Wave							-0.23	0.07	-3.18*
Entitativity X Wave							2.55	1.30	1.96†

Note. Number of Level 2 Groups = 11.

***p < .001, **p < .01, *p < .05, †p < .10

Figure 1. Model of the perceived stereotypicality.

Figure 2. Model of perceived positivity.

Figure 3. Perceived stereotypicality as a function of Wave and actual group stereotypicality.

Figure 4. Perceived Positivity as a function of actual group positivity and Wave.

Figure 5. Perceived group positivity as a function of Wave and Group Entitativity