Abstract
Entitativity is essential for individuals to experience a grouping of people as a “group.” However, entitativity is primarily studied at the individual level. If it is truly a fundamental component of group outcomes and processes, it should be considered a group-level construct. We establish that group members can share entitativity perceptions. We propose that entitativity develops in work and social groups through different self-categorization processes. Social groups can take advantage of top-down processes to establish similarity of goals and characteristics. Workgroups use both top-down and bottom-up processes with differing effects on these two forms of similarity. We propose that shared entitativity affects individual level attitudes and behavior. Results support our theoretical model. Shared entitativity explains between 2% and 11% of outcome variance in workgroups and 3% to 14% of the outcome variance in social groups. Shared similarity relates to shared entitativity differently for social and workgroups. Shared entitativity is theoretically and practically important for successful groups.
Keywords
Entitativity is the assessment of a social unit as a group (Blanchard et al., 2020; Campbell, 1958; Lickel et al., 2001). It is an essential social cognition facilitating individuals’ perception of a collection of people as a “group.” For example, people waiting at a bus stop are ostensibly a group. But when these same people participate in a book club or a staff meeting, they are “groupier.” Entitativity is this variance in groupness that makes people perceive that they are indeed part of a group.
To date, entitativity is studied primarily at the individual level. An individual’s entitativity has a powerful link to their identification, commitment, and satisfaction with their group (Blanchard et al., 2020; Hogg et al., 2007; Jans et al., 2011). Entitativity clearly matters when it comes to individual group perceptions and outcomes. However, if entitativity is truly important for group processes and group outcomes, then it should be shared by group members.
Consider these examples. First, one person at a bus stop may believe they are waiting for the bus with a group of people. This belief may affect the individual’s perceptions and behaviors at the bus stop, based on previous entitativity research (e.g., Hogg et al., 2007). There may be negative consequences for that individual if they begin to overtly act like the bus stop is a group and the other individuals do not agree. Second, members of a new book club may perceive the potential of a group, but only after a period of time will they perceive the book club as a stable group of engaged book readers. Finally, employees may not know why they were invited to attend a meeting until the leader clarifies the attendees’ shared work goals (cf., Blanchard & Allen, 2022). In these examples, it is only when other people (at the bus stop, in the book club, or at the meeting) share a perception of groupness (i.e., entitativity) that the entity becomes an intersubjective group. Further, shared entitativity likely has a stronger effect than individual members’ group perceptions and outcomes.
In this study, we extend existing entitativity literature to focus on shared, group-level entitativity. Before doing so, we first distinguish between social groups and workgroups. Members of social groups and workgroups may focus on different member characteristics as they develop shared entitativity. For example, similarity is an important antecedent to entitativity (Blanchard et al., 2020; Campbell, 1958; Lickel et al., 2001). Social groups may assume member similarity from some overarching social category such as when people join a book club, a sports group, or a political organization. Further, book club members likely expect more initial member similarities than, for example, bank tellers. Nonetheless, bank tellers may be socialized to believe that they provide the best customer service, creating a shared similarity that separates them as an entity from tellers at other banks. Distinguishing between social groups and workgroups allows researchers to better theorize about small groups, particularly workgroups for whom entitativity has been understudied.
Our contribution is threefold. First, we establish that entitativity is meaningful at the group level. Second, we test two different theoretical paths to entitativity for social groups and workgroups, allowing for a more precise understanding of their shared entitativity development. Finally, we examine how shared entitativity affects individual outcomes for group members. Entitativity as a shared perception should affect individual members above and beyond their individual entitativity. If entitativity is truly important for groups, we expect to see the groups’ shared perceptions of it emerge as a useful explanation for important individual outcomes.
The Development of Group-Level Entitativity
Campbell (1958) originally conceptualized entitativity as an objective characteristic of groups. He suggested that although it is easier to identify a solid object (such as a stone or a teacup) as an “entity,” groups of people can also be identified as entities. Characteristics of some groups (i.e., proximity, similarity, common fate, and clear perceptual boundaries) make this process easier, and subsequently, they have higher “entitativity.” At the time, Campbell’s goal was to challenge the argument that individuals are the largest unit of analysis available for social scientists (Allport, 1961; Thompson & Fine, 1999). Therefore, entitativity is grounded in the idea that groups can and should be their own unit of analysis.
In the 1990s, when research on entitativity began to re-emerge, the individual became the focus of analysis once again (Brewer & Harasty, 1996; Hamilton et al., 1998). This is reflected in the current definition of entitativity as an individual’s cognitive assessment of a social unit as a group (Blanchard et al., 2020; Hamilton et al., 1998; Lickel et al., 2001). Researchers have identified entitativity’s antecedents, including interactivity, similarity of characteristics, similarity of goals, and history of interactions (Blanchard et al., 2020; Crump et al., 2010; Ip et al., 2006; Lickel et al., 2001) as well as its outcomes such as stereotypes, ingroup and outgroup biases, group identity, and group identification (Gaertner & Schopler, 1998; Lickel et al., 2001). Research has also provided evidence that entitativity is conceptually and empirically distinct from its antecedents and outcomes (Blanchard et al., 2020; Hogg et al., 2007; Jans et al., 2011).
We argue that examining shared entitativity will move our theorizing and research forward. Group researchers, particularly in the organization sciences, have recognized the importance of group-level phenomena on individuals (N. J. Allen & O’Neill, 2015; Hitt et al., 2007; Kozlowski et al., 2013). For work and social groups, a high shared entitativity should indicate a successful group with positive effects on individual members, whereas low shared entitativity should indicate a group in which members feel more like individuals than group members. If only one member believes the group has high entitativity, the other members are unlikely to engage in behaviors to support the group and its members.
What contributes to group-level entitativity? Because group members share a context, the antecedents to entitativity at the group level are likely similar to those determined at the individual level. Currently, well-established entitativity predictors at the individual level include interactivity, similarity, and history of interactions (Blanchard et al., 2020; Ip et al., 2006; Lickel et al., 2001). In our example of people at the bus stop, shared entitativity would develop as they begin to interact with each other and establish some sort of commonality, even if it is just that they are commuting to work at the same time every day.
In this study, we focus primarily on similarity. Similarity is the primary antecedent for which an individual’s perceptions can differ from other group members, unlike other antecedents like interactivity or the history of the group. Interactivity, particularly the quantity of it, is relatively objective (Lickel et al., 2001), and the history of the group (i.e., the amount of time the group has existed) is also a more objective characteristic of the group. In contrast, group members may vary substantially in their similarity perceptions which makes it important to understand how they are shared or not. Similarity is more subjective and, thus, where we begin our investigation into shared entitativity.
Further, there may be differences for work and social groups in how similarity affects shared entitativity. We apply self-categorization theory to explain these differences, arguing that similarity of characteristics may be difficult to develop as a shared antecedent in workgroups yet easier to develop in social groups. Similarity of goals, however, may be more easily shared in both workgroup and social groups.
Theoretical Framework: Self-Categorization
The processes through which entitativity develops may be different for workgroups and social groups in ways that have not been previously acknowledged. These differences may be theoretically important, particularly for organizational scientists who are interested in understanding why some workgroups are strong and productive while others are not. One difference is that organizational scientists conceive of groups as fully interactive workgroups and teams whereas social psychologists also include much larger groups like ethnicity or nationality (cf., Blanchard & Allen, 2022). Organizational scientists should not simply apply social psychology theories to workgroups without understanding these differences in group conceptualizations.
We suggest that the differences in how entitativity develops in work and social groups are based on self-categorization. Self-categorization is the process by which people cognitively sort themselves into groups (Blanchard & Allen, 2022; Hogg & Reid, 2006; Hogg & Terry, 2000; Turner & Reynolds, 2012). Generally, self-categorization occurs when individuals perceive similarities with others.
There are two cognitive processes at play in self-categorization. The first is a deductive, top-down process (Jans et al., 2011; Tajfel, 1982). Here, similarity perceptions are inferred from an overarching social category. People perceive shared characteristics and goals because of that shared social category. For example, upon joining a book club, individuals may assume a shared love of a particular genre of books and the goal of having lively discussions after reading them.
Leaders can influence deductive perceptions of shared similarity within a group by creating similarity “from the top.” Leaders can focus group members on their shared characteristics and their shared goals (Berson et al., 2015; Vanderstukken et al., 2019). In a workgroup, particularly, effective leaders focus group members on their shared values and goals (i.e., their shared identity) for being in that particular workgroup (Ashforth et al., 2011).
A key point here is that if similarity perceptions are developed deductively, from the top, they are likely shared by all the group members, and shared similarity that develops deductively is more likely to lead to shared entitativity. Further, the shared perceptions do not depend on the group members interacting with each other because they come from a social category or an effective leader.
The second cognitive process of self-categorization is an inductive, bottom-up process. This process occurs when an overarching social category is not available. Instead, people develop similarity perceptions (or not) during their interactions with other members (Postmes et al., 2005). Sometimes, they project their own identity onto others, assuming that other groups members are “like me” in what is called a self-anchoring process (van Veelen et al., 2013). During self-anchoring, individuals note a shared similarity with other members based on their own personal characteristics and goals. For example, one bank teller may self-anchor similarity to others because they are all college graduates. Another may self-anchor similarity to others because they are all extroverted. If every member of the group perceives similarity through self-anchoring, then their shared similarity would lead to shared entitativity, even if the characteristics are objectively different.
A second inductive process can occur if members, instead of focusing on shared similarity, focus on their “shared differences” with each other (Jans et al., 2011; Postmes et al., 2005). The logic of this inductive self-categorization process is that members focus on their unique ability to make contributions to and influence the group. Group members perceive that it is their differences that create a stronger feeling of groupness (Jans et al., 2011, 2012). In this self-categorization process, everyone’s ability to be a distinct individual is the uniting feature of the group. In this type of process, since everyone focuses on their uniqueness, there would be no relationship between shared similarity and shared entitativity.
The two types of inductive self-categorization processes present competing paths to shared similarity perceptions. Self-anchoring leads to shared similarity whereas focusing on shared differences leads to shared dissimilarity. Further, at the group level, every person in the group must use the same inductive self-categorization process or there will not be a relationship between shared similarity and shared entitativity. If some members of the same group self-anchor similarity with others and other members focus on their shared differences with others, then there will unlikely be a shared group-level relationship to shared entitativity.
How do we reconcile these different self-categorization processes to predict shared entitativity? If there is an overarching social category, then it is more likely that members deductively perceive a shared similarity that positively predicts shared entitativity. If groups have an effective leader, then it is more likely that members deductively perceive a shared similarity that positively predicts shared entitativity. If there is not an overarching social category or effective leader, then group members may engage in inductive self-categorization processes during their interactions with each other. Inductive self-categorization processes will only lead to shared, group-level similarity—or group-level differences—if everyone in the group uses the same inductive self-categorization process.
Social groups most likely have an overarching social category. Book club members prefer a particular book genre and want to discuss books with the group. A soccer team likes soccer and wants to win with their team. The established social categories for social groups facilitate their shared perceptions of both similarity of characteristics and similarity of goals. Shared entitativity should develop from both of these forms of shared similarity.
Workgroups are different. Shared goals can develop from a deductive similarity process. The organizational culture, serving as a social category, may create perceptions of shared goals (Ashforth et al., 2011). Effective leaders can create perceptions of shared goals especially during effective meetings (J. A. Allen & Rogelberg, 2013; Zhang & Chiu, 2012). Further, some workgroups may have clear shared goals by the nature of their work. For example, members of a non-profit to ease hunger likely share the overarching goal of reducing hunger. Therefore, for workgroups, we expect shared goals to be related to shared entitativity.
Perceptions of shared characteristics are likely less clear for workgroups. Some workgroups, like firefighters or nurses, might share a professional identity that serves as a social category from which to deduce shared characteristics. A strong organizational identity may also create a social category to deduce shared characteristics, but other types of workgroups may not have a strong professional or organizational identity. Therefore, it is likely most workgroup members perceive shared characteristics (or shared differences) through inductive self-categorization processes. Further, all group members across all the workgroups must use the same bottom-up, inductive self-categorization process for there to be a relationship between shared entitativity and shared (dis)similarity of characteristics, which is unlikely. Therefore, we do not expect a relationship between shared entitativity and shared similarity of characteristics in workgroups.
To summarize, we hypothesize the following (see the top half of Figure 1):
Group-level antecedents to entitativity and individual-level outcomes.
Group-Level Entitativity’s Effects on Individual Group Members
While we propose group-level antecedents to entitativity, we are also interested in how shared entitativity affects individual members (the bottom half of our model in Figure 1). If shared entitativity is important, it should have a clear effect on group members. Once group members’ perceptions converge on “we think. . .” instead of “I think. . .,” these group perceptions likely have a powerful influence on the individuals (Ashforth et al., 2011).
We examine four key individual-level outcomes: identification, satisfaction, commitment, and prosocial behavior. These outcomes are believed to come from successful groups (cf., Ashforth et al., 2011). Further, they are a combination of cognitive and behavioral outcomes, potentially demonstrating the power of group-level entitativity on members’ thoughts and actions. Our theoretical logic follows.
Individual-Level Outcomes
Identification, an individual’s perception of oneness with an aggregate (Ashforth et al., 2008; Mael & Ashforth, 1992), is one of the most frequently studied outcomes of entitativity (Hamilton et al., 1998; Hogg et al., 2004, 2007; Jans et al., 2011). Individual group members identify more strongly with their group when they perceive the group to have high entitativity (Hogg et al., 2007). Individuals should also identify more strongly with a group when the other members agree that it is “more of a group” (i.e., high group-level entitativity). Previous research also links individual entitativity to satisfaction and commitment (Blanchard et al., 2020). Theoretically, there are strong arguments that shared experiences at the group level have powerful effects on individuals (Ashforth et al., 2011; Kozlowski & Klein, 2000). Therefore, when shared entitativity is high, individual identification, satisfaction, and commitment should be higher, too.
We also include prosocial behavior as a behavioral outcome in our model. Proposing that entitativity affects behaviors shows that its influence goes beyond perceptions and is not an artifact of a self-reported attitude (Conway & Lance, 2010). Prosocial behavior, broadly defined, is behavior oriented toward benefiting other people (Penner et al., 2005). Prosocial behavior has some similarities to organizational citizenship behavior, an extra-role behavior enacted by employees that is positive for the organization (Lepine et al., 2002). In this study, we examine a specific prosocial behavior in which group members represent the group in a positive way to others (i.e., boosterism; Podsakoff et al., 2000). A classic real-world example is when employees speak highly of or promote their company outside of work. This type of prosocial behavior is relevant for both work and social groups, as it does not require the specification of formal in-role behaviors.
Previous entitativity research has not examined group member behaviors. We nonetheless expect that shared entitativity relates to individual behaviors (in addition to perceptions) because prosocial behaviors are strongly linked to identification, satisfaction, and commitment, particularly within workgroups (Bergeron, 2007; Tepper et al., 2004). When individuals have positive perceptions about the group, they are also likely to engage in behaviors that benefit the group.
In summary, we hypothesize the following cross-level relationships (see Figure 1):
Methods
Overview
Before collecting data, we registered our project and hypotheses on Open Science Framework. We test our model with two samples: the first is composed of only workgroups and the second is composed of only social groups.
Participants
As part of a class exercise, undergraduate students were asked to identify both a workgroup and a social group in which they participated and to send them the research survey. We provided definitions of workgroups and social groups to assist them in identifying the different groups. We matched the members of the groups to each other in three ways: the assigned name of the group as decided by the undergraduate student, the student’s name who recruited them, and descriptions of the group. We were able to match all collected surveys using this strategy. We also reviewed the group descriptions to ensure that they were, indeed, work or social groups.
We collected data from 71 workgroups (n = 347 individual participants) and from 86 social groups (n = 396 individual participants). We eliminated all groups with less than three responses. Group sizes ranged from 3 to 22 members. The average size for workgroups was about seven members (M = 6.90, SD = 5.21) and for social groups was about six members (M = 5.83, SD = 3.70). Sample workgroups included, “Guest Services at [Car Dealership],” “[Grocery Store] Workgroup,” and “Ticket Office Reps.” There was one group of firefighters, who likely had a stronger professional identity than other groups. None of the other groups had a clear professional identity. Social groups consisted of many hobby and social interest groups as well as religious and sport groups. Sample group names include, “Foodies,” “Our Church Group,” and “Virginia’s Book Club.”
In the workgroups, 230 (67%) participants identified as female. The mean age was 30.70 years (SD = 12.94). The majority of participants (60%) identified as White. The remaining participants identified as Black (18%), Hispanic or Latino (10%), Asian American (6%), Native Hawaiian, Pacific Islander, or American Indian (<1%), or chose not to respond (5%). In the social groups, 271 (69%) participants identified as female. Participants had a mean age of 23.48 years (SD = 9.22). The majority of participants (51%) identified as White. The remaining participants identified as Black (26%), Hispanic or Latino (14%), Asian American (4%), Native Hawaiian or Pacific Islander or American Indian (<1%), or chose not to respond (5%).
Measures
We assessed group-level measures (entitativity, similarity of characteristics, and similarity of goals) using a referent shift consensus model (e.g., emphasizing “we” instead of “I’”, as in: “we feel like a group”; Chan, 1998). Conceptually, moving from “I think. . .” to “we think. . .” is an important part of an individual’s group-level perceptions (Ashforth et al., 2011). Thus, this enhanced our ability to capture a group phenomenon and distinguish it from individual experiences in the group. For the individual-level items, we focused wording on the individual (e.g., “I am satisfied with this group”). Unless otherwise noted, all scales are measured using a 7-point Likert-type scale where 1 = Strongly Disagree and 7 = Strongly Agree.
Entitativity was measured using three items validated by Blanchard et al. (2020): “We are a unit,” “We are a group,” and “We feel like a group to me.” Similarity of characteristics (four items) and similarity of goals (three items) were assessed using measures validated by Blanchard et al. (2020). Sample items include “We have similar values” and “We share a common goal,” respectively.
Group identification was measured using five items (Mael & Ashforth, 1992). A sample item is: “When someone criticizes this group, it feels like a personal insult.” Group satisfaction was measured using three items (Cammann et al., 1983). A sample item is “All in all, I am satisfied with this group.” Group commitment was assessed with four items (Klein et al., 2014). A sample item is: “How dedicated are you to this group?” For commitment, responses to these items were on a 5-point scale (1 = Not at All; 5 = Extremely).
For prosocial behavior, we used an optional, voluntary activity after the end of the survey. Our goal was to make this task an “extra-role” behavior as well as to make it challenging. First, the participants were told that the survey was complete, followed with a question in which members had the opportunity to explain the group to outsiders:
The following question is optional.
For this question, we would like to know more about your group. Please provide a detailed description of your group. For example, you can discuss the history, purpose, and goals of the group, how often you meet, and/or activities you do together.
Since participants were asked to explain in detail by writing about the group, it was more challenging than to simply click a response on the survey. This prosocial behavior focuses on boosterism, the component of citizenship behavior that involves promoting the organization to outsiders (Podsakoff et al., 2000). Therefore, we assessed a voluntary, extra-role, prosocial behavior by asking the participants to describe their group by writing about it in significant detail.
Originally, we expected that participants would either write in great detail about their group (e.g., demonstrating prosocial, extra-role behavior) or not. We planned to use the number of characters written in this optional response as an indication of their prosocial efforts. However, when we reviewed the comments, we determined that the content of what they wrote varied substantially. Some participants copied their organization’s description from the web, providing a long description that did not take much effort to produce (as an extra-role behavior). Some participants wrote a simple description of the group: “We provide 24-hour medical care and supervision, activities, and meals to the elderly in a residential assisted living setting.” Some participants wrote very positive comments about the importance and value of their group: “We’re a unit that has stuck together through the good times and the bad times. We always help one another in our time of need. We all love each other very much.” Further, we were surprised to discover that several participants wrote negative comments about the group: “[This group] is that situation where you have a few really good apples (aka good workers), and the rest are rotten, making being a ‘collective unit’ difficult.”
Because the purpose of this measure was to assess how participants promote the group to outsiders, we created an ordinal scale for the prosocial behavior. If they did not write anything, they did not engage in a voluntary, pro-social behavior. As representative as the absence of a behavior, these responses were assigned 0. Participants were assigned 1 for descriptive comments, which follows the prompt as we wrote it. Participants were assigned 2 for positive and emotional comments, which went well beyond the written survey instructions. Participants who wrote negative comments did not engage in a prosocial behavior. These comments represent an anti-social behavior by “bad-mouthing” the group to an outsider (Note. We perform sensitivity analyses in the results to test our ordinal scale.). Therefore, they were assigned −1. Three research assistants (who were blind to the group’s entitativity score) independently coded the content of the responses reaching an initial agreement of 94% for both the work and social groups. They resolved any differences by discussing discrepancies in the codes.
Control variables
Because identification, satisfaction, and commitment are well-studied individual-level phenomena, we included control variables to account for established relationships already known in the literature. We asked participants to report their tenure in the group (1 = Less than 1 month, 2 = 1–6 months, 3 = 6 months to 1 year, 4 = 1–3 years, 5 = More than 3 years) and if they were the leader of their group (0 = No, 1 = Yes).
Further, we wanted to account for each individual’s unique level of entitativity (Evans et al., 2019). While our focus is shared entitativity, we know that individuals will (likely) differ from the group mean. Although we focus on shared entitativity, the entitativity of particularly enthusiastic or alienated individuals—as different from the rest of the group—will also affect individual outcomes. In order to control for this effect, we include individual differences in entitativity. We calculated this as each individual’s entitativity minus their group’s entitativity. Thus, if an individual’s entitativity is higher than the group mean, they will have a higher individual difference. This allows us to distinguish individual effects from group effects. If shared entitativity remains significant in a model that provides a unique estimate of individual entitativity, then we can be more confident in the importance of shared entitativity (Woznyj et al., 2017).
Results
Aggregation of Group-Level Data
The first step when testing a group-level construct is to determine whether the construct can and should be examined at the group level. We calculated two agreement statistics, rwg(j) and ICC(1). The rwg(j) statistic assess within-group agreement. A minimum rwg(j) of .70 is recommended to claim that group members agree on a particular value (Bliese, 2000). Our calculations show that for the workgroups, similarity of characteristics (rwg(j) = .78), similarity of goals (rwg(j) = .78), and entitativity (rwg(j) = .81) all surpass that cutoff. The same is true for the social groups with similarity of characteristics (rwg(j) = .85), similarity of goals (rwg(j) = .87), and entitativity (rwg(j) = .89)
ICC(1) assesses interrater reliability (Bliese, 2000). An average ICC(1) value for group-level constructs in organizational research is approximately .22 (Woehr et al., 2015), meaning that 22% of the variance in individual ratings can be attributed to group membership . Our calculations show that, for the workgroups, similarity of characteristics (ICC(1) = .22), similarity of goals (ICC(1) = .27), and entitativity (ICC(1) = .22) all meet or surpass that average. For the social groups, the interrater reliability measures for similarity of characteristics (ICC(1) = .17), similarity of goals (ICC(1) = .15), and entitativity (ICC(1) = .15) are lower. They are, however, considered medium to strong indicators of reliability for group level measures (LeBreton & Senter, 2008). The higher levels for workgroups correspond with existing evidence that workgroups generally have stronger interrater reliability than social groups (cf., Woehr et al., 2015).
The descriptive analyses at the group and individual levels for the workgroups are in Table 1. The mean workgroup entitativity is 5.72 (out of 7) suggesting that most participants believe their workgroup is relatively “groupy.” Group-level similarity of characteristics is lower than group-level similarity of goals. Therefore, while workgroup member’s similarity of characteristics perceptions converge (as indicated by rwg(j) and ICC(1)), participants perceive less similarity of characteristics compared to similarity of goals. The correlation between shared goals and shared entitativity (r = .81, p < .001) is relatively high. The mean of prosocial behavior is close to 0. Most people did not respond to this question providing more evidence that this question was perceived as a voluntary part of the survey.
Descriptive Statistics for Workgroups.
Note. There are N = 71 groups and N = 347 individuals. For prosocial behavior, −1 is negative, 0 is neutral or no comment, and 1 is positive (prosocial). For leader role, 0 is no and 1 is yes. For tenure, 1 = Less than 1 month, 2 = 1 to 6 months, 3 = 6 months to 1 year, 4 = 1 to 3 years, 5 = more than 3 years. All other variables are a 1 to 7 scale from strongly disagree to strongly agree.
p < .05. **p < .01. ***p < .001.
The means, standard deviations, and correlations for the social groups are in Table 2. Social group members perceive that their groups are relatively high in entitativity (M = 5.98, SD = 0.54). The mean group-level similarity of characteristics, similarity of goals, and entitativity are similar in magnitude. Further, prosocial behavior is still relatively close to 0, indicating that it is also perceived as an extra-role behavior for social groups.
Descriptive Statistics for Two Social Groups.
Note. There are N = 86 groups and N = 396 individuals. For prosocial behavior, −1 is negative, 0 is neutral or no comment, and 1 is positive (prosocial). For leader role, 0 is no and 1 is yes. For tenure, 1 = less than 1 month, 2 = 1 to 6 months, 3 = 6 months to 1 year, 4 = 1 to 3 years, 5 = more than 3 years. All other variables are a 1 to 7 scale from strongly disagree to strongly agree.
p < .05. **p < .01. ***p < .001.
We ran null model analyses for both groups to determine if there was sufficient variance in the individual responses due to their group membership. For Level 1 (i.e., individual level) variables, ICC(1) is the total amount of variance in the variable attributable to group membership (Bliese, 2000). An ICC of .10 is considered a medium effect and .25 is considered a large effect (LeBreton & Senter, 2008). For workgroups, our ICC(1) values all fell within this range indicating that workgroup membership has an effect on individual responses (see Table 3). The ICC(1) values for the social groups are also in this range, except for satisfaction. Only a small amount of satisfaction’s variance (9%) is explained by group membership for the social groups.
Multilevel Path Analysis of Research Model for Workgroups.
Note. Entitativity (diff) is each individual’s difference in entitativity from their group’s mean level of entitativity. Positive entitativity (diff) mean the individual had higher entitativity than the group. Regression coefficients are unstandardized b. Pseudo R2 for Group is the amount is the amount of variance explained by group level entitativity.
p < .05. **p < .01. *** p < .001.
We used MPLUS to test the full model in one comprehensive analysis. Results are displayed in Figure 2 with the unstandardized coefficients for the workgroups in bold and the social groups italicized. Analysis details are presented in Tables 3 (workgroups) and 4 (social groups). We start with the workgroup analyses. At the group level, our data support Hypothesis 1a that, for workgroups, shared similarity of goals was positively related to shared entitativity (b = 0.69, p < .001). Shared similarity of characteristics was not related to shared entitativity in workgroups (b = 0.07, p = .60). Because of the strong correlations of these two variables, we wanted to rule out multicollinearity. The VIF for this path analysis is 2.38, below the 2.5 multicollinearity cut-off.
Results of cross-level analyses.
Regarding the cross-level relationships in the workgroups, our data support Hypothesis 2 through Hypothesis 5. Group-level entitativity positively relates to individual-level identification (b = 0.69, p < .001), individual-level satisfaction (b = 0.76, p < .001), individual-level commitment (b = 0.57, p < .001), and individual-level prosocial behavior (b = 0.19, p < .05). We calculated the pseudo-R2 in order to isolate the amount of variance in the individual variables attributed to shared entitativity. Pseudo-R2 tells us how much of the total variance is attributed to shared entitativity as opposed to just group membership (i.e., the ICC(1) calculations). Pseudo-R2 ranging from 2% to 4% are considered important in organizational research (Woznyj et al., 2017). Shared entitativity accounts for 2% of the total group variance of prosocial behavior and 4% of the total group variance of satisfaction, which is considered reasonable. Shared entitativity accounts for a substantial 11% and 12% of the total group variance for identification and commitment respectively.
We controlled for tenure, leadership role, and how much each individual’s entitativity varied from their group’s mean entitativity. Group tenure and leadership role positively related to identification and commitment. Participants’ individual entitativity difference had a relatively strong relationship with all of the individual variables, as it was expected that individual perceptions would affect individual attitudes and behaviors. Therefore, observing that shared entitativity has a statistically significant effect while simultaneously analyzing individual’s unique level of entitativity provides more evidence of importance of shared entitativity.
For the social groups, at the group level, our data support Hypothesis 1b and 1c; similarity of goals (b = 0.47, p < .001) and similarity of characteristics (b = 0.31, p < .001) are both positively related to entitativity (see Table 4 and Figure 2). For the cross-level analyses, our data support Hypotheses 2 through 5. Shared entitativity positively relates to identification (b = 0.83, p < .001), satisfaction (b = 0.25, p < .001), commitment (b = 0.70, p < .001), and prosocial behavior (b = 0.30, p < .01). Our pseudo-R2 calculations suggest that shared entitativity accounts for an important amount of the total variance for each individual-level variable (Woznyj et al., 2017): 10% for identification, 5% for satisfaction, 14% for commitment, and 3% for prosocial behavior.
Multilevel Path Analysis of Research Model for Social Groups.
Note. Entitativity (diff) is each individual’s difference in entitativity from their group’s mean level of entitativity. Positive entitativity (diff) mean the individual had higher entitativity than the group. Regression coefficients are unstandardized b.
p < .05. **p < .01. ***p < .001.
Regarding our control variables, group tenure positively relates to individual level identification, satisfaction, and commitment. Leadership role relates to commitment. Individual differences in entitativity positively relates to all four individual level variables. Therefore, shared entitativity plays an important role for individual attitudes and behavior in social groups as well as workgroups.
Additional analyses
To test the sensitivity of our calculated measure of individual-level prosocial behavior, we reran the analysis for each sample excluding the people who did not respond to the optional question. The results of this analysis show that, for workgroups, group-level entitativity positively relates to individual-level prosocial behavior (b = 0.38, p < .05), but for social groups, entitativity is unrelated to individual member’s prosocial behavior (b = −0.00, p = .96) after eliminating those who did not comment. This indicates that people who did not participate in the prosocial behavior in the social groups are important in understanding the relationship between shared entitativity and prosocial behavior for those who did.
Second, in both samples, identification and commitment have a very high amount of variance explained by group membership. We ran an additional analysis with shared identification and commitment at the same analytical level as shared entitativity (see Figures 3 and 4). Modification indices in the workgroup sample suggest a direct path from shared similarity of characteristics to shared identification, resulting in a significant path. Thus, for workgroups, while shared similarity of characteristics may not have an effect on entitativity in workgroups, it is related to shared identification. For the social groups, no modification indices were suggested. Shared entitativity is related to shared identification, commitment, and prosocial behavior.
Additional workgroup analyses.
Additional social group analyses.
Discussion
The contributions of this paper are to examine if entitativity is meaningfully shared by group members, to assess differences in how similarity relates to entitativity in work and social groups, and to assess the effects of shared entitativity on individual attitudes and behaviors. Our paper establishes that entitativity is meaningfully shared by group members and, further, that shared entitativity affects individuals above and beyond individual’s unique entitativity. Indeed, our results suggest that shared entitativity has a stronger effect on individual outcomes than is typically seen by group researchers (LeBreton & Senter, 2008; Woehr et al., 2015; Woznyj et al., 2017).
Therefore, theoretically, shared entitativity may provide insights into additional group dynamics. First, this research provides more support that entitativity is the foundation of successful groups. While we examined individual outcomes, the strength of shared entitativity may provide new avenues to understand intragroup processes and group level outcomes. A higher level of shared entitativity may help explain, for example, when conflict is productive within groups, the negative effects of group fault-lines, and the challenges of multi-team systems (Maltarich et al., 2021; Simons & Peterson, 2000; Zaccaro et al., 2020). As evidence continues to grow of the importance of entitativity within groups, we may further establish the importance of ensuring that all group members perceive that, indeed, they are in a group.
We also determined that shared entitativity develops through different paths in social groups and workgroups. For social groups, shared similarity of goals and similarity of characteristics relate to shared entitativity, likely due to the ease of determining similarity through deductive social categorization. For workgroups, shared similarity of goals relates to shared entitativity but shared similarity of characteristics does not. It is important to note that, for workgroups, similarity of characteristics is a group-level shared perception, but it is not related to shared entitativity. In our additional analyses, shared similarity of characteristics has a direct relationship to shared identification in workgroups.
Our theoretical argument is that, in groups with no clear, overarching social category, similarity develops through inductive processes during group members’ interactions with each other (cf., Jans et al., 2011, 2012). Some group members self-anchor their own characteristics to similar others while some group members perceive that is their shared uniqueness that makes their group “groupier.” Because our data show that, within a group, workgroup members converge upon similarity of characteristics, we believe that workgroup members use the same type of inductive process within groups. Between groups, however, the inductive self-categorization processes vary. For example, one group of bank tellers may self-anchor their similarity while another group of bank tellers may appreciate how each member makes unique contributions.
We propose that group members’ appreciation of their shared differences may happen later in the development of successful groups. As group members successfully interact and achieve group goals together over time, they learn about each other’s unique contributions. Each member perceives that they are appreciated for their unique contributions; they perceive and appreciate others for their unique contributions. It is, therefore, in established successful groups that group members turn more to an inductive self-categorization process that focuses on shared differences (cf., Jans et al., 2011, 2012). Longitudinal research on the development of successful groups and their evolving perceptions of each other will help us to better understand and theorize about these self-categorization processes, particularly in small, interacting groups.
We would also like to extend this particular argument to consider demographic diversity. Our study defined similarity of characteristics as similarity in values and beliefs. Yet, similarity of characteristics can also include demographic similarities and differences. We perceive a theoretical process involving shared similarity and shared entitativity that might explain the conflicting research on the effects of diversity on group performance (cf., Roberson, 2019). Early in their development, diverse groups may unite around similarity of goals to create shared entitativity. As the group successfully performs over time, inductive self-categorization processes appreciating shared group member differences creates a more robust shared entitativity based on similarity of goals and (dis)similarity of characteristics. It is in the later stage of successful group development that diverse groups appreciate their unique contributions. Future research testing inductive self-categorization processes across time may by particularly useful in demographically diverse groups.
Our research provides several important implications for group leaders, particularly for workgroup leaders. Group leaders should focus on new members’ socialization processes (Moreland et al., 2001) by highlighting and reinforcing shared goals. This should help new members develop enough entitativity to integrate into the group. In groups without clear overarching social categories, leaders should focus on creating shared characteristics for the members to perceive about their group. Deductive self-categorization processes create shared similarity more easily than inductive ones. As the group becomes stronger and members know each other better, the leader might then emphasize the unique contributions of each member to make a successful group. Leaders who can create shared similarity perceptions early in the group’s developmental process will create groups with greater “groupness.”
Limitations
One limitation of our research is that the design was cross-sectional. Longitudinal data would provide us with the ability to track changes in group dynamics as well as to study what is likely a recursive relationship of our model. A longitudinal analysis, however, would still not provide us with the ability to make causal claims within our model (Spector, 2019). One way we addressed this limitation was to include one behavioral measure (i.e., prosocial behavior) to complement the perceptual measures. While experiments would assist us with internal validity, the relationships, attitudes, and ongoing commitments present in our data would not be easily recreated in a one-time experiment in a lab. Therefore, our research allows us to test our model with real work and social groups, even acknowledging the inability to make cause claims.
Another limitation is that, by recruiting a variety of groups, instead of strategically sampling specific types of workgroups (e.g., those with and without a clear professional identity), we could not test in more detail the effects of inductive self-categorization. This, however, presents an opportunity for future research to determine and test when some workgroups can more easily develop shared similarity—or shared distinctiveness—to create a strong group.
Conclusion
While still important at the individual level, entitativity—the perception that a social unit is a group—is also relevant and influential at the group level. Workgroups and social groups may put different emphasis on which forms of similarity are important when they think about their groups. Workgroups emphasize similarity of goals, while social groups emphasize similarity of goals and similarity of characteristics. These differences are likely related to social categorization processes people enact when interpreting whether their groups are actually “groups.” It is important to recognize that group members share a level of perceived “groupness” with each other, and this shared perception affects not only their individual perceptions about the group, but also their behavior. Thus, entitativity is an important concept that helps us understand group processes and outcomes.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
