Abstract
The present study shows that categorization of reward recipients into different entities affects distributive preferences by third-party non-recipient allocators. Rewards were allocated more equally to members of one group than to members of two dyads or to independent recipients. Moreover, allocators who were explicitly requested to allocate rewards justly were more egalitarian than those who were not requested to do so. More interestingly, rewards were allocated more equally between members in each of two dyads and between independent recipients, when a just allocation request was made, than when such a request was not made. This implies that a request for just allocation modifies the effects of recipient entity categorization toward more equal reward allocations.
Keywords
Positive and negative social resources of various kinds (such as goods, information, prestige, prison sentences, fines, and vacations) are frequently allocated to people by third parties. Third-party allocators are allocators who themselves are non-recipients (often referred to as authorities in the procedural justice literature; Lind & Tyler, 1988; Tyler & Lind, 1992). Third-party allocation is a distributive event and differs from exchange in that the latter is a bilateral process while the former is unilateral. Thus, self-interest is less likely to be a confounding variable in explanations of the third-party allocator’s motives and behavior. As third-party allocations continually occur in a variety of social contexts (at micro as well as macro levels) and affect most people’s lives, analyses of the factors affecting this type of social resource allocation preferences and behavior are theoretically and practically warranted.
Some studies in the tradition of equity theory were concerned with how equitably non-recipients allocate resources. The findings from this work suggest that just as people prefer equitable outcomes for themselves, they also want others to receive rewards and punishments in accord with their inputs (e.g., Baker, 1974; cf. “inequity aversion” advanced by Fehr & Schmidt, 1999). Further developments within the social psychological study of justice yielded a broader conceptualization of justice than the equity principle to accommodate additional rules of justice, such as need, equality, self-interest, and winner takes all (e.g., Deutsch, 1975; Kazemi & Törnblom, 2008; Reis, 1984). In general, this research attempted to clarify why people think and act according to different justice rules at different occasions, and why different persons think and act according to different justice rules in the same situation. A number of factors that appear to determine the way in which we define justice and that, in turn, affect resource allocation have been suggested in the theoretical and empirical literature (for a review and a categorization of these factors, see Törnblom, 1992).
The major objective of the study reported here was to investigate the effects of recipient entity categorization on third-party social resource allocation preferences, both when a just allocation is and is not requested. The present research combines insights from social categorization processes and distributive justice theory (Wenzel, 2002, 2004). Unlike research on social categorization and intergroup relations in the tradition of Tajfel and his associates (Tajfel, 1978), participants in the present study were not experimentally assigned to particular group membership on the basis of which their behavior toward their in-group and an out-group were studied. Our study features a simple manipulation of the recipient entity as one of three types: an aggregate of four independent individuals, two dyads, and a four-person group.
Why and how would recipient entity categorization affect allocation preferences of third-party non-recipient allocators? Turning to various justice theories for enlightenment does not seem to carry us very far. Therefore, we lean on insights from other areas of inquiry. A considerable amount of research on intergroup relations focuses on how individuals, as representatives of an in-group, behave toward their own group and an out-group. Studies by Tajfel (1978) suggest that a mere category differentiation among groups on seemingly trivial grounds (i.e., minimal groups) is sometimes sufficient to result in intergroup bias in the form of in-group favoritism and out-group discrimination. In the context of resource allocation, this tendency is manifested by a payoff strategy that promotes maximal difference in favor of the in-group (even when the allocator is not a recipient). The equitableness of the payoffs tends to matter less than the creation of an absolute difference between the in-group and the out-group. This seems to be true even when the in-group receives less, in absolute amounts, than would have been possible by using another strategy which would make payoffs to the groups more equal, or even equitable (cf., Eek & Gärling, 2008; Yamagishi, Mifune, Liu, & Pauling, 2008).
Research on social influence and conformity has shown that the degree of conformity increases with the number of influencing persons or entities. However, the crucial factor does not appear to be the size of the social entity (as expressed by the number of constituent individuals), but rather the size of opposition expressed in terms of the number of distinct and independent social entities (whether these be groups, individuals, or a mixture of both). This distinction between person and entity is crucial for understanding how we chose to manipulate one of our independent variables (i.e., recipient entity categorization). Arguing on the basis of a Gestalt principle of organization, where similar entities are grouped together in a person’s perception, Wilder (1977) suggested and empirically verified the proposition that four unrelated individuals would exert stronger influence than four individuals who are categorized as belonging to the same group, and that an increase in the number of groups (given the same total number of four persons, i.e., two dyads) would also result in greater influence. Thus, four entities yield more social influence than two, and two entities exert more influence than one (e.g., one group consisted of four individuals).
Following this line of reasoning, if the way a person categorizes, or perceives, others in terms of social entities affects the outcome of social influence, it seems reasonable to assume that it also would affect how a person allocates social resources to others who are categorized into different social entities. In the present study, participants who viewed videotaped recordings showing four actors taking part in a creativity test were asked to assess differences in performance among the actors. Three levels of recipient–actor entity categorization were used: one group of four actors (group), two dyads (dyad), and four separate individual actors (aggregate). Participants in all entity categorization conditions were then instructed to allocate rewards to each of the actors. Finally, the difference between the performances of the most and the least creative actor was calculated.
Furthermore, if the accomplishment of a particular purpose or goal is not specified for reward allocation, allocators are assumed to act in the direction of preserving the existing social arrangement among recipients. To this end, equality in reward distribution is likely to contribute to the maintenance of status similarity and cohesiveness within a recipient group. On the other hand, an aggregate of unrelated individuals is often perceived as differentiated along one or more status dimensions and competitively charged. Its preservation is likely to be aided by equitable reward distributions which are guided by the recipient’s performance. 1 Applying these arguments in the case of two or more groups and dyads, a third party would most likely allocate resources equally within but equitably between the dyads. Thus, when the accomplishment of some goal is not overtly requested (e.g., that the allocation must be just, maximally differentiating, promoting conflict, motivating productivity), the maintenance of the current recipient social entities is likely to be salient and guide resource distribution. In our study, as noted above, actors were asked to allocate rewards to recipients who they had witnessed to differ in terms of creativity.
In formal notation,
A second major focus of the present study is how recipient categorization affects third-party allocation when an explicit request for just allocation as a goal is made. The prediction of third-party non-recipient allocators’ choice of distributive rules in such a situation may be facilitated if the tacit (ground) and focal (figure/salient) perspectives that are likely to be activated are specified (cf., Wegner, 1982). In the absence of explicit guidelines concerning the required form of justice, it seems plausible that the allocator would act in congruence with his or her beliefs about what focal (salient) others would want him or her to do to create justice. As Mikula and Schwinger (1978) argued that the justice conception prevailing in the superordinated social system will be particularly influential, we assume that a reward allocator would opt for the socially and/or culturally appropriate and acceptable behavior, especially when she or he has no personal stake in the recipient entity’s outcomes. As data for the present study were collected in Sweden, the prevailing conception of justice in Sweden is described next.
In Sweden, equality appears to be the dominant justice ideal (e.g., Anderson, 1969; Tomasson, 1970). This value is fundamental to the policy statements of all political parties in the country, as well as to organizations concerned with equal rights for women, immigrants and other disadvantaged groups. The domestically powerful Swedish labor movement, as well as the stronghold of traditional social democratic values in general, provides considerable normative endorsement of egalitarian orientations, which have a great impact on its citizens. Factorial surveys conducted by Törnblom and Jonsson (1985, 1987), Törnblom, Jonsson, and Foa (1985) and Törnblom, Mühlhausen, and Jonsson (1991) yielded overwhelming endorsement for egalitarian distributions among Swedes.
Thus, an explicit request to allocate rewards justly is assumed to make the culturally appropriate justice conception focal and operative (making social categorization less salient; i.e., a tacit or ground issue). Given the equality-oriented ideology in Sweden, and a situation in which recipient categorization is non-salient, we predict the following:
In formal notation,
How might allocators act when two factors are salient rather than one (as in the two previous cases)? Our study includes also a focus on double salience (i.e., the interactive effect of recipient entity categorization and request/no request for justice). It seems reasonable to assume that social categorization of recipients as members of the same group, resulting in equality-oriented reward allocation, will, when combined with a request for just allocation, bring the allocation outcome even closer to full equality. Also, when just allocation is not explicitly requested, equity-oriented allocations are expected for recipients categorized as belonging to dyads and for separate individuals categorized as an aggregate. Put differently, comparing two situations, one in which just allocation is requested and in the other not, third parties will allocate rewards more equally between dyads and between independent recipients in the first than in the second situation. Thus, the combination of social entity categorization and request (Hypothesis 1 and Hypothesis 2) generates the following predictions:
In formal notation,
Method
Participants and Design
Seventy-five male (with an approximate Mage of 20) and 67 female Swedish undergraduates (with an approximate Mage of 25) viewed videotaped recordings showing 4 actors participating in a creativity test. The task was to find as many unusual usages as possible for a concrete building brick. Participants were assigned to one of six experimental conditions. A 3 (categorization: a 4-member group vs. 2 dyads vs. an aggregate of 4 independent individuals) × 2 (request: just allocation vs. no just allocation request) factorial design was used.
Procedure
Participants were made to believe that they were to participate in a large research program concerned with creativity. The alleged focus of this research was an examination of why people differ in their abilities to judge the creativity of others. To this end, participants were asked to watch four female actors on a video recording to find out whether or not they could perceive any differences among them in terms of creativity. The actors were viewed as they participated in a creativity test, attempting to think of as many unusual usages as they could for an ordinary brick. They had 2 min at their disposal. Each actor’s performance differed markedly from the others’ performances. The considerable gap between the most and the least creative actors was quite obvious on the video recording—the former listed 50% more unusual usages than the latter. The students acting on the video recording had been instructed how well they were to perform in the creativity test.
The categorization variable was experimentally manipulated through the Gestalt principle of proximity (i.e., how close they sat around a table). In the group condition, the individual actors sat very close to each other and interacted (i.e., talked to each other). In the dyad condition, the individual actors sat close to each other two by two and talked to each other in their respective dyads. In the aggregate condition, the individual actors were placed at different tables and thus did not engage in any interaction with each other. To reinforce the impact of the visual categorization manipulation, verbal information was also provided. Participants in the group condition were told that the four actors were members of the same group; in the two-dyad condition, they learned that the four actors came from two different groups/dyads (the least and the most creative actors were in different dyads); and in the aggregate condition, they were informed that the actors were four independent and unrelated individuals, unknown to each other. The ideas generated by the actors were identical across all six experimental conditions.
The request conditions were created through written manipulations in the questionnaire that was administered to the participants after they had watched the videotape. Half of the participants in each of the three conditions were explicitly asked to make a just allocation of rewards while the other half were merely asked to make an allocation. Each participant was randomly assigned to one of these conditions.
Dependent measure
A questionnaire was administered after the participants had viewed the videotaped recording asking how they would allocate rewards if they could give up to 50 Swedish kronor (approximately US$8) to each of the actors taking the creativity test. Following a procedure used in previous research (cf., Sabbagh & Malka, 2012), the difference between the amounts allotted to the most and to the least creative person was calculated. The two measures were examined in terms of how closely they approximated an equitable allocation (i.e., large differences between amounts) or an equal allocation (i.e., small differences). After completing the questionnaire, participants were thanked and debriefed.
Results
A 3 (categorization: group vs. two-dyad vs. aggregate) × 2 (request: just vs. no request) between-subjects analysis of variance revealed significant main effects of social categorization, F(2, 136) = 19.61, p < .001, η2 = .22, and request, F(1, 136) = 49.03, p < .001, η2 = .27. These main effects were further qualified by a significant two-way interaction, F(2, 136) = 9.52, p < .001, η2 = .12.
Lending support to Hypothesis 1, Bonferroni-corrected independent t-tests at p = .05 showed that third parties allocated more equally between actors belonging to the same group (one entity; M = 5.7, SD = 5.5) than between actors coming from two different dyads (two entities; M = 12.5, SD = 10.5), t(90) = −3.90, p < .001, and unrelated actors (i.e., aggregate/four entities; M = 14.0, SD = 7.7), t(94) = −6.20, p < .001. Allocations between actors coming from two dyads and actors as unrelated individuals were the same, t(94) < 1.
The main effect of request gives support to Hypothesis 2 in that participants were expected to allocate rewards more equally when requested to allocate as justly as possible (M = 6.7, SD = 5.5) than when no such request was made (M = 14.7, SD = 9.7). Thus, the mean differences between the amounts allocated to the most creative and the least creative actor were smaller in the former case than in the latter.
Hypothesis 3 predicted an interactive effect of social categorization and request on third-party allocations. This prediction was confirmed by the data. Means are given in Table 1. Specifically, comparing two situations, one in which just allocation was requested and in the other not, significant larger differences in allocations were observed in the latter than in the former case (i.e., more differentiated and less equal allocations). As revealed by Bonferroni-corrected independent t-tests at p = .05, this pattern was true when allocating among actors of two dyads (i.e., the dyad condition), t(44) = −6.27, p < .001, as well as allocating among separate individuals (i.e., the aggregate condition), t(48) = −3.89, p < .001. Thus, the effects of social entity categorization toward more differentiated allocations were increased under the no just request condition.
Means and Standard Deviations of Mean Differences Between Amounts Allocated to the Most and the Least Creative Persons When a Just Allocation Is and Is Not Requested.
Note. Standard deviations are indicated in parentheses.
However, no significant differences in allocations were yielded in the group condition, t(44) = −1.75, p = .088. This indicates that reward allocations among group members did not come closer to equality when just allocations were requested as compared with when such a request was not made.
Discussion
Studies on intergroup relations have shown that a mere categorization of individuals into in-groups and out-groups often produces bias against the out-group, when a non-recipient group member distributes rewards to members of his or her in-group and to members of out-groups. Our study suggests that even the allocations by a third-party non-recipient allocator (i.e., a person who is not a group member) may be affected by the way recipients are grouped and categorized. Participants were told that four strangers appearing on a video recording were either members of one group, members of two dyads, or four unrelated individuals. Thus, our recipient entity categorization variable included three levels: one entity (a group of four members), two entities (two dyads of two persons each), and four entities (four independent individuals). In other words, the distinction between the group, dyad, and aggregate conditions was contingent on how the four actors were arranged, a distinction which proved to yield different effects on reward allocations.
After watching each recipient-actor’s performance on a creativity test, participants were asked as third parties to allocate monetary rewards to each one of the four actors. The present study thus added to the design of previous allocation studies in the way recipients were categorized, the number of recipients (i.e., more than two), and the comparison between the conditions of presence and absence of explicit request for just allocations.
Researchers frequently make a distinction between situations in which the allocator is or is not a recipient of the rewards to be allocated. When the available amount of resources is limited, and the amount the recipient-allocator will receive is dependent on what she or he gives to the other recipient(s), the allocator often takes considerable interest in the situation of the other recipient(s). When the allocator’s share is at stake, a high degree of self-interest is expected (e.g., Johansson, 2005). This is true to a much lesser extent (if at all) in third-party allocations. In general, a third-party (i.e., non-recipient) allocator is assumed to have little (or no) self-interest vis-à-vis the recipient entity, as the allocation does not affect the allocator himself or herself.
It seems likely that the distinction between recipient and non-recipient allocators might not always be particularly useful or important. It is not difficult to think of occasions at which non-recipient allocators would be highly interested in and affected by the outcomes of others (e.g., when the affective relationship between allocator and recipient is close), or when recipient allocators would not exhibit self-interest (e.g., when the allocator sacrifices his or her share in favor of his or her needy child). Third-party recipients as well as non-recipient allocators may take great interest in the fate of others and forego self-interest, as when natural disasters evoke humane sentiments resulting in voluntary rescue actions or donations of money, food, and clothing. Thus, an alternative and, at times, more relevant distinction is made in terms of the allocator’s particular form of social awareness—regardless of whether she or he is a recipient or not (see Wegner, 1982). In other words, the identification of an allocator in terms of his or her recipient status may carry little (or no) significance in the absence of complementary information about the allocator’s activated form of social awareness. Attention by future researchers to issues like these would contribute to more precise predictions about allocation behavior.
Previous research has shown that we–they distinctions as well as a variety of recipient characteristics (e.g., performance, physical attributes, abilities, needs, conformity, effort) affect social resource allocation (cf., Leung & Bond, 1984; Ostrom, Carpenter, Sedikides, & Li, 1993). The results of our study suggest that recipients’ shares of resources also (at least partly) depend on the third-party allocator’s perceptions regarding the way in which the recipients are grouped or related to each other. The way recipients are structurally related to each other (e.g., as members of the same group in the group condition, different groups in the dyad condition, or as unrelated individuals in the aggregate condition) seems to influence their outcomes.
When just allocation was explicitly requested to guide reward allocation, allocators tended to respond to the request in each of the three entity conditions. Thus, as Table 1 shows, the differences in rewards provided to the most and the least creative person turned out to be smaller in each of the three entity conditions when justice was (as compared with when justice was not) requested. Participants in our study allocated rewards closer toward equality when justice was requested than when it was not. This held true for both the dyad and aggregate entity conditions. However, with regard to the group entity condition there was no significant difference between the degrees of equality of the rewards given to the most and the least creative recipients in the just and no just request conditions. A plausible explanation is that of a ceiling effect produced by the social categorization (i.e., group) manipulation. The reward difference between the recipients was so close to perfect equality that the remaining possible increment was not large enough to allow statistical significance.
Another interesting finding was that participants allocating rewards to dyads (i.e., two entities) were more influenced by the just allocation request than participants allocating to members of a group or an aggregate. Specifically, the differences in rewards provided to the most and the least creative person turned out to be much larger for the two dyads’ condition (14.3) as compared with the other two entity conditions (i.e., 2.0 for the group and 7.5 for the aggregate entity conditions; see Table 1) across the requested as well as not requested just allocation conditions. Additional studies need to be designed to explain this interesting pattern with regard to the dyad condition.
As we have seen, requests for just reward allocation modified the effects of social categorization of recipients on third-party allocations. Although most current studies do explicitly require participants to allocate justly, or assess the justice of a situation, several equity theory studies in the past did not. Participants were exposed to situations of inequitable pay, and the researcher then observed if and how they reacted to the inequitable treatment, but they were never asked whether or not they perceived the situation as just or unjust (e.g., Lane & Messé, 1971; Leventhal & Anderson, 1970; Leventhal & Lane, 1970; Mikula, 1974). More recent work within the area of procedural justice is also deficient in this respect. Indirect measures (i.e., measures of factors that are correlated with or contribute to the perception of justice) are assumed to be equivalent to direct measures of perceived justice (e.g., Colquitt, 2001; Tyler, Degoey, & Smith, 1996). These shortcomings are discussed in greater detail by Mikula (2005): “The lack of distinction between these different types of measures can lead to inappropriate assumptions about the validity of the measures, as well as conceptual flaws and inappropriate interpretations and conclusions” (p. 203). Thus, given that a request to act justly may make people act differently (for instance more equally, as in this study), than when they are not so requested, results from previous research may have to be interpreted with caution. Results from these two kinds of studies will have to be separately interpreted, as conclusions based on an undifferentiated mix of both is likely to be biased and, in the worst case, meaningless.
To analyze the practical implications of this study with regard to various phenomena (e.g., discrimination), a first important step is to distinguish among different types of situations and third parties. From the vantage point of a recipient, knowledge about the best way in which to identify and present himself or herself may make a difference in terms of the size of his or her share of allotted/entitled social resources. Thus, the strategic use of self-entity presentation opens up a line of interesting future research that may also complement theory and research within the area of self-presentation of which self-entity presentation is a special case.
Another practical implication of our findings is that performance may be judged more harshly in dyads, and even more so in aggregates (a collection of unrelated individuals). Put differently, it is more disadvantageous for high-performers and more advantageous for low-performers to be perceived as parts of a group (i.e., one entity) than as parts of a dyad (i.e., in our case two entities) or as parts of an aggregate (i.e., in our case four entities). Thus, our findings suggest that the differences in allocated rewards increase as the number of entities increase.
Justice is a fundamental and salient value in any society, and its forms may vary culturally. Through socialization even the most marginal member can hardly avoid being aware of and be affected by this powerful standard of moral conduct. As the culturally dominant ideology in Sweden generally seems to favor equality, participants in this study were hypothesized to be heavily guided by this norm in their resource distribution. Furthermore, even though participants in the group condition were all (at least partially) individuated through the differential performances shown by the video recordings, participants approached equality rather than equity when distributing rewards. This seems to counter the effect of individuation reported in other research (e.g., Wilder, 1978).
The results of this study may not be generalizable outside the Swedish context. Studies conducted in other cultures, particularly those in which equality is not as heavily emphasized as in Sweden, would likely generate different results. For instance, it seems reasonable to expect cultures characterized as high in power distance, masculinity, and individualism (see Hofstede, 1980, 2001) or authority ranking and market pricing (see Fiske, 1991) to be less equality oriented than cultures that are described as high in femininity and collectivism (Hofstede, 1980, 2001) or equality matching and communal sharing (Fiske, 1991).
As resource allocations occur within social, cultural, and historical contexts, further research along the lines of the present study should be conducted in other cultures and incorporate additional kinds of social entities as well as allocators, different types of performances, and additional types of social resources as rewards (see Törnblom & Kazemi, 2012, for classifications of social resources).
Footnotes
Acknowledgements
Thanks are due to Marilyn Brewer for her valuable comments on a previous version of this article.
This article is part of the special issue Leadership and the Group, Small Group Research, Volume 45(4).
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.