Download Expectation States Theory: Formation of Performance Expectations and Status Hierarchies and more Lecture notes Social Theory in PDF only on Docsity!
CHAPTER 2
Expectation States Theory
SHELLEY J. CORRELL
CECILIA L. RIDGEWAY
INTRODUCTION
Women in work groups often feel that their ideas are ignored or mistakenly credited to one of their male coworkers. African Americans often say they feel that they have to per- form twice as well as their white counterparts to be given the same level of recognition. The ideas of people who talk more in a group are often judged to be more valuable than those offered by less talkative members. People with more prestigious jobs are more likely to be chosen leader of a group, such as a jury, even when their job has little, if anything, to do with the task at hand. Women are more likely than men in a group to be interrupted. Ideas often "sound better" when offered by someone perceived to be attractive. What all of these observations have in common is that some members of a group seem to have real advantages that are denied to others. They have more opportunities to speak, their ideas are taken more seriously, and they have more influence over other group members. In expectation states theory these hierarchies of evaluation, influence, and participation are referred to as the "power and prestige structure" or the "status structure" of the group. The theory seeks to explain how these inequitable structures emerge and are maintained, and how they are related to other aspects of inequality in society.
SHELLEY J. CORRELL • Department of Sociology, University of Wisconsin-Madison, 1180 Observatory Drive, Madison, Wisconsin 53706 CECILIA L. RIDGEWAY • Department of Sociology, Stanford University, 450 Serra Mall, Building 120, Room 160, Stanford, California 94305- Handbook of Social Psychology, edited by John Delamater. Kluwer Academic/Plenum Publishers, New York, 2003.
29
30 Shelley J. Correll and Cecilia L. Ridgeway
HISTORY
Expectation states theory began as an effort to explain some of the most striking findings of Robert F. Bales' (1950) influential early studies of interpersonal behavior in small groups (Berger, Conner, & Fisek, 1974; Berger & Zelditch, 1998, pp. 97-113). Bales (1950, 1970) recorded the interactions of homogeneous, initially leaderless decision-making groups of three to seven unacquainted Harvard sophomore males over mul- tiple hour-long sessions. Despite the initial lack of group structure and the social similarities of the members, inequalities in interaction developed quickly, stabilized over the first session, and then guided interaction thereafter. If inequalities emerge quickly in unstructured groups of social equals, Bales (1950) reasoned, status hierarchies are very likely in any group. The inequalities Bales observed consisted of four correlated behaviors: participation ini- tiated, opportunities given to participate, evaluations received, and influence over others. Bales (1970) found, for instance, that groups developed a most talkative member who talked considerably more than the others in the group. This most talkative person was also the one addressed most often by the others. The more a person talked, compared to the others, the more likely he was to be rated by others has having the best ideas and doing the most to guide and influence the group. The founders of expectation states theory, Joseph Berger, Bernard Cohen, Morris Zelditch, and colleagues, sought to explain why these correlated inequalities, labeled the group's "power and prestige" (i.e., status) structure, emerge together and how this happens even in a group of social equals. Berger and his colleagues were also influenced by two additional sets of early studies. One set demonstrated the power of status structures, once formed, to bias group members' evaluations of each other and their behavior in the group. Riecken (1958) showed that the same idea was rated as more valuable when it came from a talkative group member than from a less talkative one. Sherif, White, and Harvey (1955) demonstrated that group members over- estimate the performance of high status members and underestimate the performance of low status members. Whyte (1943), in his classic study of a street comer gang, showed that group members actually pressured one another to perform better or worse to keep their perform- ances in line with their status in the group. Another influential set of early studies demonstrated that when members of a goal- oriented group differed in socially significant ways, the interactional status structures that emerged tended to reflect the social status attached to each member's distinguishing charac- teristics. Strodtbeck, James, and Hawkins (1957), for instance, found that mock jury mem- bers' occupational status and gender predicted how active and influential they became, how competent and helpful they were judged to be by others, and how likely they were to be cho- sen foreman of the jury. Yet, the question left unanswered was how this occurred. These studies encouraged Berger and his colleagues to formulate expectation states the- ory as a theory of an underlying process that (1) accounts for the formation of interactional status structures and (2) can explain how these structures develop both in groups of social equals and in groups where people differ in socially significant ways (Berger et al., 1974; Berger, Fisek, Norman, & Zelditch, 1977; Berger & Zelditch, 1998). The way people's socially significant characteristics, such as race, gender, occupation, or age, shape their access to participation, influence, and positive evaluation is an important aspect of social stratifica- tion in society. As a consequence, although expectation states theory began by explaining sta- tus structures in homogeneous groups, its explanation of status structures among people with significant social differences has become the most highly developed and commonly used aspect of the theory.
32 Shelley J. Correll and Cecilia L. Ridgeway
physical attractiveness), (2) social rewards, and (3) patterns of behavior interchange between actors. We describe these three processes next along with empirical evidence in regard to them.
Status Characteristics and Performance Expectations
Perhaps one of the most important ways that actors develop differentiated performance expec- tations is by using socially significant attributes of individuals, called status characteristics, to anticipate the quality of their future task performances. Status characteristics are attributes on which people differ (e.g., gender, computer expertise) and for which there are widely held beliefs in the culture associating greater social worthiness and competence with one category of the attribute (men, computer expert) than another (women, computer novice). Status char- acteristics can be either specific or dijfuse. Specific status characteristics, such as computer expertise, carry cultural expectations for competence at limited, well-defined range of tasks and, consequently, only impact the formation of performance expectations in this limited range of settings. Diffuse status characteristics, on the other hand, carry very general expec- tations for competence, in addition to specific expectations for greater or lesser competence at particular tasks. They affect performance expectations across a wide range of settings.
Gender is an example of a diffuse status characteristic in the United States and else- where. Widely shared cultural beliefs about gender have been shown to include expectations that men are diffusely more competent at most things, as well as specific assumptions that men are better at some particular tasks (e.g., mechanical tasks) while women are better at oth- ers (e.g., nurturing tasks) (Conway, Pizzamiglio, & Mount, 1996; Wagner & Berger, 1997; Williams & Best, 1990). It is useful to compare the cultural beliefs that constitute a status characteristic to group stereotypes and to social identity based on group categorization. It is well known that mere categorization encourages beliefs that favor one's own category over another (Brewer & Brown, 1998; Mullen, Brown, & Smith, 1992; Tajfel, 1978). Status beliefs, in conteast to in- group favoritism, are social representations that consensually evaluate one category as more status worthy and competent than another. This means that rather than simply preferring one's own group, even those disadvantaged by a status belief accept, as a social fact, that the other group is socially evaluated as better than their own (lost & Burgess, 2000; Ridgeway, Boyle, Kuipers, & Robinson, 1998; Ridgeway & Erickson, 2000). As a set of evaluative beliefs about social categories, status beliefs form an element of many widely shared group stereotypes. Importantly, the status element of group stereotypes, if present, is fairly similar across stereotypes that otherwise differ dramatically in content (Conway et al., 1996; lost & Banaji, 1994). For instance, the stereotypes of gender, of race/ethnic categories, and of occupations differ enormously in specific content. But each of these stereotype sets has in cormnon a status element that associates greater worthiness and competence with one category of the distinction (men, whites, professionals) than another (women, people of color, blue-collar workers). Because of this similar status element, expec- tation states theory argues that otherwise very different social distinctions can have compara- ble effects on the organization of interactional status hierarchies. In discussing status beliefs, we should be clear that we are not endorsing the content of these beliefs. Nor are we suggesting that the self-fulfilling consequences of status beliefs are inevitable. Instead, it is our contention that reducing social inequalities in everyday contexts requires first acknowledging that status beliefs exist and then attempting to understand and expose the inequitable processes they prime. It is to that task that we now turn.
Expectation States Theory 33
STATUS CHARACTERISTICS THEORY, status characteristics theory is a formal subtheory of expectation states theory that seeks to explain how beliefs about status characteristics get translated into performance expectations, which in turn, shape the behaviors of individuals in a group (Berger et al., 1977; Webster & Foschi, 1988). Some refer to status characteristics the- ory as a theory of status generalization, which is the process of attributing specific abilities to individuals based on the status characteristics they posses. At the heart of the theory is a set of five assumptions that link beliefs about status to behavior (Balkwell, 1991; Berger et al., 1977). According to the salience assumption, for any attribute to affect performance expectations, it must be socially significant for the actors in the setting. A status characteristic is salient if it either differentiates actors, or if actors believe that the characteristic is relevant to completing the group's task. Consequently, situational goals and the way actors compare one another on the characteristic impact how and if a sta- tus characteristic affects performance expectations. The same characteristic (e.g., having a college degree) can advantage an actor in one setting (with a less educated group), have no impact in another (in a group where all have university degrees), and disadvantage the actor in a third setting (with a more educated group). Importantly, this implies that no status char- acteristic advantages or disadvantages an actor in all settings. Whether the status beliefs cul- turally available to actors shape performance expectations in any actual setting depends on the structure of the local setting itself. The second assumption is called the burden of proof assumption and concerns the way status characteristics that differentiate actors but are not initially relevant to the performance of the group's task impact the formation of performance expectations. Actors act as though the burden of proof rests with showing that a salient status characteristic should not be taken into account when forming performance expectations. All salient information is incorporated, unless something in the setting explicitly dissociates the status characteristic from the task. So, for example, if gender is salient in a setting it will differentiate the performance expecta- tions for men and women even though gender itself is not relevant to the task at hand. It is through the burden of proof process that diffuse status characteristics such as gender, age, race/ethnicity, and social class have modest but pervasive effects on the status hierarchies that emerge across a large range of settings in which they have no obvious task relevance. The sequencing assumption specifies what happens in the more complicated situation when actors either enter or leave an existing social setting. The main point is that no status or competence information is lost. The performance expectations that formed in one encounter carry over to the next encounter, even if the specific actors change. This assumption has been used to intervene in the status generalization process. For example, if a man observes a woman performing a task better than he does, this can positively impact the performance expectations he forms for women in future encounters (Pugh & Wahrman, 1983). The effect may wear off over time without a "booster" experience, however (Markovsky, Smith, & Berger, 1984).
The aggregation assumption explains how the status information associated with multi- ple characteristics is combined to form aggregated performance expectations. In actual groups, such as work groups or committees, people commonly differ from one another on several status characteristics at the same time, and often these multiple status characteristics generate inconsistent expectations for performance. For example, on a legal team, a member may be not only a Harvard trained lawyer, but also an African American woman. A distinc- tive advantage of status characteristics theory is it offers a procedure for making exact pre- dictions for the order of performance expectations actors will construct from a given set of salient consistent and inconsistent status characteristics. To continue with our example, if
Expectation States Theory 35
_D_* r+ _c'_*
FIGURE 2-2. Graph theoretic representation of two actors differing on one diffuse status characteristic.
Signed graphs, like the one in Figure 2-2, link actors to expected task outcomes (posi- tive or negative) through a series of paths. Since performance expectations are relative for each pair of actors in a setting, the structure represents the status situation for two actors, p (for self) and o (for other). Figure 2-2 depicts the relatively simple status situation where only one diffuse status characteristic, symbolized D, is salient in the setting. The positive sign attached to D for actor/? indicates that/? has the more valued state of the diffuse characteris- tic compared to actor o. For example, p might be a man interacting with a woman, o. A neg- ative dimensionality line connects the two states of D. Since the actors possess oppositely valued states of D, the characteristic D is salient in the setting. Proceeding to the right, the symbol F represents the expectation of an actor's general com- petence. Since actor p has the more valued state of D, the expectation for p's general compe- tence is high relative to actor o. Higher expectations for general competence lead to higher expectations for competence at the group's focal task. The symbol C* refers to the expectation for an actor's competence at a specific task. As the positive and negative signs attached to C* indicate, the expectation for competence at the focal task is higher for actor/? compared to actor o. This path exists because, as stated in the burden of proof assumption, a salient status charac- teristic is believed to be relevant unless it is somehow explicitly dissociated from the task at hand. T"" refers to a successful task outcome, and T' refers to an unsuccessful task outcome. There are two paths linking actor p to expectations about his future task perform- ance. The first is the path: p D+ F"" C^* f^ and the second path is: p Z)"" D~ F~ C" T~. Two important features of these paths are their lengths and their signs. Shorter paths have a greater impact on the magnitude of the expectation. Conceptually, as paths become longer it becomes harder for an actor to reason from the path to the task outcome. By simply counting the links between actor and task outcome, we deter- mine that the first path diagramed above has a length of 4, compared to a length of 5 for the second path. The sign of the paths are determined by the method commonly used with signed graphs: We multiply the signs of the path by the sign of the task outcome to which the path leads. Doing so for the two paths above indicates that both are positive. If we now apply the aggregation assumption, we first combine all like signed paths to com- pute the expectations for the positive and negative subset for actor p according to the formulas
e ; = { 1 - [ 1 - / ( 0 } - [! - / ( « ) } ; (la)
e ; = { i - [ i - / ( 0 ) •••[!-/(«)}; Ob) and then the aggregate expectation is represented by:
e = e+ - e. (2)
A similar calculation is made for actor o. Actorp's expectation advantage over actor o is sim- ply the difference between their individual expectations {e —e^.
36 Shelley J. Correll and Cecilia L. Ridgeway
Values for/(0 have been estimated empirically (Berger et al., 1977). Fisek, Norman, and Nelson-Kilger, (1992) have also derived a functional form for f(i), which fits existing data well:
/(0 = l-exp(2.6182-'). (3)
In Figure 2-2, actor/? has two positive paths, one of length 4 and one of length 5, and no negative paths. Therefore, equation (la) becomes:
e^= { l - [ l - / ( 4 ) ] [ l - / ( 5 ) ] - 0 ). (4a)
Likewise, actor o has two negative paths, one of length 4 and one of length 5 and no positive paths, making equation (lb)
e „ = 0 - { l - [ l - / ( 4 ) ] [ l - / ( 5 ) ] }. (4b)
Using Fisek et al.'s derivation (equation [3] above),/(4) = 0.1358 and/fS) = 0.0542. Substituting these values into equations (4a) and (4b), e = 0.1827 and e^ = ~0.1827, making the expectation advantage of actor p over actor o as 0.3653.
EMPIRICAL EVIDENCE. Status characteristics theory, and expectation states theory more generally, have been subjected to rigorous empirical evaluation, which has generated consid- erable evidence in support of the theory. Most of this evidence has come from social psycho- logical experiments. Experiments afford the researcher the ability to isolate and manipulate variables of key theoretical interest, while controlling for potentially confounding factors. As such, experiments produce data that can more clearly establish the extent to which a change in an independent variable caused a change in the dependent variable, rather than being the result of some confounding or spurious factor. The conceptual advances within status characteristics theory can largely be attributed to the reliance of researchers on a standardized experimental setting. This setting consists of a set of standardized procedures for introducing manipulations and operationalizations of key theoretical variables (e.g., status characteristics), assessing the effects of the independent vari- ables on the dependent variable, which is usually a measure of social influence, and employ- ing manipulations to achieve the scope conditions under which the theory is argued to hold (Troyer, 2001). By holding these aspects of the setting constant across studies whenever pos- sible, the results that are produced can be compared across studies, which allows researchers to build on the results of others with confidence. The standardized setting begins by instructing research participants that they are parti- cipating in a study designed to evaluate a "newly discovered skill." They are told that they will participate in a decision-making task with a "partner."* The task will evaluate their abihty in regard to the skill. Several different "abilities" are commonly evaluated, including "contrast sensitivity ability," "meaning insight ability," and "spatial judgment ability." Participants axe told that these skills are unrelated to known abilities, such as mathematical competence or artistic ability. These instructions and the use of a task associated with a fictitious ability are
*Quotes around phrases in this section indicate that the phrase represents an experimental deception. For example, the phrase "newly discovered skill" is communicated to the research participant. In actuality, the skills are usually fictitious. Likewise, "partners" are often computer programs, unbeknownst to the subject.
38 Shelley J. Correll and Cecilia L. Ridgeway
differential impact of status characteristics based on their relevance to the task leads to some distinctive predictions of the theory. For instance, the theory predicts that in a mixed sex group with a gender-neutral task, men will have an advantage over women in participation and influence. If the task is a masculine typed one, men's advantage over women in these behaviors will be even greater. But if the task is a feminine typed one, women will have a modest advantage over men in participation and influence. A large body of research sup- ports this pattern of behavioral inequalities in mixed sex contexts (for reviews, see Ridgeway, 2001a; Ridgeway & Smith-Lovin, 1999). Experiments further confirm that people form influence hierarchies as if they were com- bining consistent and inconsistent status information, as predicted by the aggregation assump- tion (Webster & Driskell, 1978; Zelditich, Lauderdale, & Stublarec, 1980). There is evidence as well for the inconsistency effect. Recall that the addition of another status characteristic in a situation is argued to have a greater marginal impact on the status hierarchy if it is inconsis- tent, rather than consistent with other salient status information (Berger, Norman, Balkwell, & Smith, 1992; Norman, Smith, & Berger, 1988). Berger et al. (1992) compared the ability of subset combining to account for the interactional hierarchies participants in experiments formed from sets of consistent and inconsistent status information with three other informa- tion processing principles. They found that subset combining provided the best fit for the data. In a broader evaluation of status characteristic theory's ability to predict group status struc- tures with its graph theoretic model of salience, relevance, and aggregation, Fisek et al. (1992) compared theoretical predictions to data from 24 experiments, reporting a good fit.
Rewards and Performance Expectations
Recall that expectation states theory posits three processes by which differentiated perform- ance expectations emerge (see Figure 2-1). We have discussed at length the impact of salient status characteristics. We now turn to the other processes, beginning with the impact of socially valued rewards. The theory argues that when a socially valued reward is distributed unequally among mem- bers of a group, the actors will infer performance expectations from their reward differences (Berger, Fisek, Norman, & Wagner, 1985). In this way, the differential distribution of rewards, like status characteristics, can actually create a status hierarchy among actors or modify positions in an existing hierarchy. In an experimental test of this argument. Cook (1975) showed that when a third party gave differential rewards to group members who had no other basis for evaluating their performances on a shared task, the members used the reward differences to infer ability dif- ferences. Harrod (1980) and Stewart and Moore (1992) showed that allocating differential pay levels to participants in an experiment created corresponding influence hierarchies among them during interaction. These results highlight how the power or good luck represented in the unequal possession of rewards generates status distinctions that are considered legitimate by those in the setting. By creating performance expectations, the unequal rewards appear to be "deserved" and, thus, jusfly bring respect, deference, and influence. Unequal rewards, according to the theory, combine with other factors, such as salient status characteristics, to determine the aggregated per- formance expectations that shape the behavioral status order in the setting. In established hierarchies, actors' expectations for rewards in a task setting are interde- pendent with their expectations for performance and, consequently, with their positions in the status steucture (Berger et al., 1985; Cook, 1975). It is a common observation in established hierarchies that valued rewards (pay, a comer office) tend to be distributed in accordance with
Expectation States Theory 39
rank and help maintain the relative power of those ranks (Homans, 1961). Because of the interdependence of performance and reward expectations, the theory predicts that when a sta- tus characteristic is salient in a setting, those disadvantaged by it will implicitly expect lower levels of rewards for themselves than will those advantaged by the characteristic. Research on women's lower sense of entitlement to rewards compared to men supports this prediction (Bylsma & Major, 1992; Jost, 1997; Major, McFarlin, & Gagnon, 1984).
Behavioral Interchange Patterns and Performance Expectations
In addition to status characteristics and rewards, a third factor that can have independent effects on performance expectations is the behavioral interchange pattern that develops among two or more actors (Fisek, Berger, & Norman, 1991; Skvoretz & Fararo, 1996). Such a pattern occurs between two or more actors when one engages in assertive, higher status behaviors (e.g., initi- ating speech, making a task suggestion, resisting change in the face of disagreement) that are responded to with deferential, lower status behaviors by the other actor(s) (e.g., hesitating to speak, positively evaluating the other's suggestion, changing to agree with the other). The more frequently these types of patterns are repeated between the actors, the more likely the actors are to view the behavioral patterns as cultural status typifications, which are shared beliefs about typical high-status-low-status, "leader-follower" behaviors. Following the common assumption that people speak up more confidently about things at which they are more expert, salient sta- tus typifications induce actors to assume that the more assertive actor is more competent at the task than the more deferential actor, creating differential performance expectations for them. In support of this argument, a variety of assertive verbal and nonverbal cues including taking a seat at the head of the table, having an upright, relaxed posture, speaking up without hesitation in a firm, confident tone, and maintaining more eye contact while speaking than listening have been shown in the United States to make an actor's ideas "sound better" and increase influence (for reviews see Dovidio & Ellyson 1985; Ridgeway, 1987; Ridgeway, Berger, & Smith, 1985). Behavior interchange patterns shape performance expectations most powerfully among those actors in a group who are equals in both their external status characteristics and their reward levels, such as between two women in a mixed sex group (Fisek et al., 1991). Behavioral interchange patterns are the means by which expectation states theory accounts for the devel- opment of status structures in homogeneous groups like those studied by Bales (1950, 1970). When actors differ in status characteristics, the differentiated performance expectations created by the status characteristics shape the actors' verbal and nonverbal assertiveness. Consequently, differences in status characteristics shape behavioral interchange patterns, as several studies have shown (Dovidio, Brown, Heltman, Ellyson, & Keating, 1988; Ridgeway et al., 1985; Smith-Lovin & Brody, 1989). In a clear demonstration of expectation states the- ory's predictions in this regard, Dovidio et al. (1988) showed that when mixed sex dyads shifted from a gender neutral task, where the man had a status advantage, to a feminine typed task, where the woman had a status advantage, the actors' participation rates and assertive nonverbal behaviors reversed from a pattern favoring the man to one favoring the woman. Thus, between actors who already differ on status characteristics, behavior interchange pat- terns often add litde new information to the existing order of performance expectations.
Fisek et al. (1991) used the graph-theoretical methods described earlier to develop a model of how behavior interchange patterns combine with status characteristics and rewards to create an aggregated order of performance expectations for actors in the setting, which impacts the status structure of the group. They evaluated this model's ability to account for participation
Expectation States Theory 41
work: "Our parents' advice was true: We really do have to work twice as hard [as whites] to be considered half as good" (p. 58). Carter describes a common observation by members of low status groups: Due to status beliefs that disadvantage them, they must actually perform at higher levels than members of high status groups to be judged as equally competent. More generally, the level of performance required for inferring ability varies with the status char- acteristics individuals possess. In an extension of expectation states theory, Foschi (1989, 2000) incorporates insights from the psychological literature on attribution to account for these kinds of observations. She intro- duces "standards" as the mechanism by which actors attribute performance to ability. Foschi regards standards as a function of salient diffuse status characteristics that create differential per- formance expectations for actors. According to double standards theory, these differential performance expectations activate the use of different standards for attributing ability. When lower status individuals perform well at the group's task, their performances are critically scrutinized since a good performance is inconsistent with what was expected based on their position in the group's status hierarchy. When higher status individuals perform equally as well, their perform- ances are consistent with status-based expectations and are, therefore, less scrutinized. Thus, those possessing the more valued state of a status characteristic are judged by a more lenient standard than are those with the more devalued state. As a result, equal task performances are more likely to be judged as indicative of ability when performed by a higher status member of the group. The evidence supporting double standards theory ranges from accounts and descriptions, to results from surveys and experiments (for a review see Foschi, 2000). For example, in one experiment subjects in mixed sex dyads were informed that the group's task was one on which men generally perform better (Foschi, 1996). After completing this task, subjects were told that they scored in the mid range and either slightly higher or slightly lower than their opposite-sex partners. Subjects were then asked to estimate what percentage of questions the higher performing subject would need to have answered correctly in order to determine that s/he possessed task ability. As predicted, subjects set a significantly higher standard for ability when the better performer was a woman rather than a man. Biernat and Kobrynowicz (1997) report similar results for race as well as gender. As with expectation states theory more generally, the predictions of double standards theory are dependent on features of the setting. For example, when gender is salient in the set- ting, the theory predicts that men will be held to a more lenient standard than women either when men are thought to be better at the task at hand or, according to the burden of proof assumption, when gender differentiates people in a setting but is not specifically linked to the task. If the setting is instead one where women are thought to be better at the task, the theory predicts that women would be judged by a more lenient ability standard. Double standard theory shows that in addition to being given fewer opportunities to par- ticipate initially in the group, when lower status members do participate, their performances are evaluated by a stricter standard. This makes it difficult for competent performances by lower status members to be noticed as such, which further reduces their ability to achieve high status in the group.
Second Order Expectations
Status hierarchies have been shown to emerge in collectively oriented task groups because actors in the group develop differentiated performance expectations for themselves and their group mates. The performance expectations described in expectation states theory axe. first
42 Shelley J. Correll and Cecilia L. Ridgeway
order expectations: they are the personal expectations an actor, p, holds for self and other, a. However, it is likely that the expectations actor p believes are held by others in the group also influence the emerging status structure. This idea has its roots in the long standing insight from social psychology that our perceptions of others' expectations influence our sense of self and our behavior in interaction (Cooley, 1902; Goffman, 1959; Mead, 1934). Recent theoret- ical elaborations in expectation states theory have sought to explain how these beliefs about others' expectations—called second order expectations —influence the power and prestige order of groups (Moore, 1985; Troyer & Younts, 1997; Webster & Whitmeyer, 1999).
Second order expectations refer, more specifically, to what an actor, p, believes that another in the situation, o, thinks about/?'s and o's relative abilities (Moore, 1985; Webster & Whitmeyer, 1999). Since people generally overestimate the extent to which others see things as they do (Marks & Miller, 1987), actors usually presume their own self-other expectations are shared by those in the situation and act on them accordingly (Troyer & Younts, 1997; Zelditch & Floyd, 1998). In this situation, second order expectations provide no new infor- mation. However, when second order expectations are communicated and they either conflict with first order expectations or are expressed when an actor has no self-other (first order) expectations, they will likely influence the first order expectations of actors in the setting and, consequently, the status structure of the group. Consistent with these ideas Moore (1985) found that when participants in an experiment with no information about their competence compared to a partner heard their partner's views about their relative competence levels, these second order expectations shaped the first order expectations participants formed for themselves compared to the partner. Troyer and Younts (1997) showed that when group members receive second order expectations that conflict with their own first order expectations, they combine the information in the two sets of expecta- tions to create aggregate, revised performance expectations that become the basis for their interaction in the group. They also found that in some instances, second order expectations actually had more influence than first order expectations in guiding interaction. Drawing on previous research, Webster and Whitmeyer (1999) propose that the impact of another's second order expectations on p's own expectations is a function of the perform- ance expectations/J holds for that other. Second order expectations communicated by an actor held in high regard will have a stronger impact than will expectations imputed by a less well regarded actor. Webster and Whitmeyer (1999) update expectation states theory's graph- theoretic model to show how second order expectations combine with all other salient status information to create the aggregate performance expectations upon which group members enact their status structure. While social psychologists have long believed that our perceptions of others' expecta- tions are important in making sense of self and guiding interaction, the incorporation of this insight into expectation states theory makes it possible to generate precise predictions about the relative impact of first and second order expectations in various settings. Consequently, this body of theoretical and empirical work not only represents an important elaboration of expectation states theory, but it also provides a systematic and empirically supported account of one of the key insights of social psychology.
Legitimacy
Empirical evaluations of expectation states theory have clearly demonstrated that individuals who posses a diffuse status characteristic that is devalued in society experience interactional
44 Shelley J. Correll and Cecilia L. Ridgeway
persuasion to exercise dominant, directive power, group members, as the theory predicts, were significantly more likely to comply with status advantaged leader and to resist the mer- itocratic leader. Thus, group members were more likely to treat the diffuse status advantaged leaders as legitimate.
Status Construction Theory
Distinguishing characteristics such as occupation or race become status characteristics in a society when widely shared status beliefs develop that associate greater status worthiness and competence with those in one category of the characteristic than in another category. One of the ways that expectation states theory has broadened its focus in recent years has been to ask how such status beliefs develop. As we have seen, status beliefs play an essential role in connecting the status organiza- tion of society as a whole with the status experiences of individuals. Yet, sociology has little systematic knowledge about how these beliefs develop, are maintained, or change. Weber ([1921] 1946) suggested many years ago that social groups commonly acquire an economic advantage first before acquiring high status in society. Yet even this observation fails to explain how a purely economic advantage is transformed into shared cultural beliefs about social status. There are probably many ways that widely shared status beliefs form in societies. Status construction theory, however, asks whether the insights of expectation states theory can be used to explain at least some of these processes (Ridgeway, 1991, 2001b). Since expectation states theory has shown that status beliefs are at play in goal-oriented encounters among peo- ple, status construction theory asks if these same encounters might be a potent forum for the development and spread of new status beliefs or the maintenance or change of existing status beliefs. Status construction theory begins with a simple suggestion. When people who differ on a socially recognized characteristic interact in regard to a shared goal, a status hierarchy will emerge among them as it does in almost all goal-oriented encounters. There is a chance, how- ever, that the participants will associate the relative status each is accorded in this hierarchy with the characteristic that differentiates them, and form a fledgling status belief about the characteristic. Whether these fledgling status beliefs are supported in future encounters and become stable status beliefs depends on the nature of the beliefs other people in other encounters are also forming about the same characteristic. If there is some factor that gives people in one cat- egory of the characteristic (call them As) a systematic advantage in gaining influence and esteem in encounters with people in another category of the characteristic (call them Bs), then the majority of encounters between As and Bs will induce their participants to form status beliefs that As are more worthy and competent than J5S. Since more people develop status beliefs favoring As rather than Bs under such circumstances, people who hold beliefs favor- ing As are more likely to have their beliefs supported in future encounters than are those who hold contrary beliefs. Also, when people who form a status belief in one encounter act on it in a subsequent encounter between As and Bs, there is a chance that they will "teach" their status belief to the others present by treating those others either deferentially or assertively according to the belief. In this way, the initial small advantage for status beliefs favoring As rather than Bs is likely to spread and grow among people in the society. Under many circumstances, argues
Expectation States Theory 45
Status construction theory, the eventual result will be widely shared status beliefs that As are more worthy and competent than Bs. Computer simulations of this process by which status beliefs spread through society suggest that, if people do form beliefs in encounters as the the- ory argues, then widely shared status beliefs would indeed be a logical result under many societal conditions (Ridgeway & Balkwell, 1997). One factor that could give As an advantage in gaining influence and esteem in encoun- ters with Bs is an economic advantage, as Weber suggested. As we have seen, differences in socially valued rewards such as pay or wealth tend to create corresponding differences in per- formance expectations that, in turn, create differences in influence and esteem in goal- oriented encounters. Therefore, if more As become economically advantaged in society than 6s, As will have a systematic advantage in gaining influence and esteem in the majority of encounters between As and 5s. As a result, widely shared status beliefs favoring As over i5s are likely to develop in the society. In this way, an economic advantage is transformed into cultural beliefs about the status of social groups. To test whether people form status beliefs in this way, Ridgeway and colleagues (1998) told participants in an experiment that their partners differed from them in "personal response style." They were also told that they would be paid either more or less than their partners. While working on a decision task with their partners, influence hierarchies developed that corresponded to pay differences. After two such experiences, participants formed beliefs that "most people" see the typical person in the better paid response style group as more respected, more competent, more leader-like, higher status, but not as likeable as the typical person from the less well paid response style group. In other words, participants formed status beliefs favoring the economically advantaged response style group. Importantly, these status beliefs were consensual in that people from the less well paid group also agreed that most people see those from the better paid group as more respected and competent than those from their own group. Economic advantages are one factor that can bias the development of status hierarchies between people who differ on a socially significant characteristic and cause status beliefs to form about the characteristic. Other factors, such as control of technology or valuable information (e.g., computer literacy), could have this effect as well, as long as these factors systematically bias the development of status hierarchies among people who differ on a char- acteristic. Webster and Hysom (1998), for instance, show how society's moral evaluations of homosexuality systematically bias the development of influence hierarchies between homo- sexuals and heterosexuals and foster status beliefs that disadvantage homosexuals in percep- tions of worthiness and competence. For widely shared status beliefs to develop in society, however, it is important not only that people form beliefs from their encounters, but also that they "teach" the beliefs to others by treating those others according to the beliefs in subsequent encounters. To examine this, participants in another experiment were again told that they differed from their partners in response style (Ridgeway & Erickson, 2000). While working on a task, the partners, who were confederates, treated the participants as if they held status beliefs about the difference by acting deferentially or assertively, causing influence hierarchies to form. After two such experiences, participants developed status beliefs about the response style groups that corre- sponded to their partner's treatment of them, confirming that status beliefs can be spread by acting on those beliefs. An additional experiment showed that third party participants who witnessed someone different from them defer to or assert influence over someone similar to them also acquired corresponding status beliefs, suggesting that encounters spread status beliefs widely (Ridgeway & Erickson, 2000).
Expectation States Theory 47
reason why the theory has limited its scope to collectively oriented task groups is that in these groups individuals find it necessary to make relative anticipations of the likely task compe- tence of group members. Importantly, the logic of the theory does not specifically require collective orientation as much as it requires individuals to consider themselves relative to another. Erickson (1998) has argued that whenever situational demands pressure actors to assess their task competence relative to others on a socially valid task, status processes should occur. While collectively oriented task groups readily create this pressure, settings where indi- viduals engage in socially significant evaluative tasks, even if individually, also represent a setting where individuals are pressured to make relative assessments of their expected competence. Why is this so? Individual evaluative tasks can provide the pressure to make relative assessments of competence in situations where actors know they will receive a socially important and socially valid performance evaluation. The use of evaluative tasks to rank individuals' per- formances is socially valid in the Weberian sense; that is, individuals expect others to accept the ranking as legitimate and, consequently, orient their behavior toward this expectation (see Weber [1918] 1968, pp. 31-33). The anticipation of this ranking creates a pressure for actors to assess their task competence relative to others who they imagine are also being or have been evaluated. This coordination of rank position requires evaluating oneself in relation to the social environment. However, the standards for what constitutes a competent performance are not usually clearly defined beforehand, and others' precise scores are rarely known. In this uncertain environment, salient status characteristics are available to influence performance expectations, as they do in collective task situations. Through the process of status general- ization, individuals develop performance expectations for themselves that are consistent with their state on the salient status characteristic (Correll, 2001b; Erickson, 1998). Assuming that a status characteristic is indeed salient in an individual evaluative setting, three theoretical predictions are implied. First, those with the more devalued state of the char- acteristic will perform less well on the task compared to those with the more valued state of the characteristic (cf. Lovaglia et al., 1998; Steele, 1997). Second, controlling for actual task performance, those with the more devalued state will evaluate their own task performance as less indicative of ability compared with the evaluations of those with the more valued state. Finally, when others evaluate the ability of high and low status actors, the same performance will be judged as more indicative of ability for high status actors (cf Foschi et al., 1994). In an experiment designed to meet Erickson's (1998) revised scope conditions and test the second of these predictions, Correll (2001b) compared how male and female subjects rated their competence at a "newly discovered ability" after taking a test purportedly designed to measure this ability. To make the test socially valid, participants were informed that the test was being considered for use in screening applicants for graduate school admissions. To make gender salient and task relevant, subjects in half of the conditions were told that men usually score higher on tests of the ability. To specifically disassociate gender from the task in the other conditions, subjects there were told that there is no gender difference in test scores. All subjects received the same slightly above average scores for their performance. In the first condition, where subjects had been told that males score higher on tests like the one they had just taken, male subjects rated their task ability significantly higher than female subjects did even though all subjects had received identical scores. In the gender irrelevant condition, no gender difference was found in how subjects rated their task ability. Since this experiment was specifically designed to meet the expanded scope conditions laid out by Erickson (1998), it provides the most convincing evidence to date that status processes occur in individual eval- uative settings, settings that lack collective orientation.
48 Shelley J. Correll and Cecilia L. Ridgeway
Extending the scope conditions to include individual evaluative settings is an important advancement since this setting is both very common and highly consequential in its impact on educational and occupational attainment. It includes most standardized test settings, including those that are used to determine college, graduate school, and professional school admissions and those used for certification in a wide range of professional occupations. Expectation states theory has generated empirically supported propositions about how pre- existing inequalities are reproduced in collectively oriented task groups. This newer work in individual evaluative settings indicates that the impact of status processes on the reproduction of inequality is even more far reaching.
CONCLUSION
Expectation states theory is, in many ways, a textbook example of a theoretical research pro- gram. It is deductive, programmatic, formalized mathematically, cumulative, precise, and pre- dictive; and its propositions have been subjected to rigorous evaluation. More importantly, however, it is a theory that illuminates core issues in social psychology and sociology more broadly. It is fundamentally a "macro-micro-macro" explanation about one way that cate- gorical inequality is reproduced in society. Cultural beliefs about social categories at the macro level impact behavior and evaluation at the individual level, which acts to reproduce status structures that are consistent with pre-existing macro-level beliefs. Status structures in groups can be thought of as the building blocks of more macro-level structural inequalities in society. For example, to the extent that status processes make it less likely for women in work groups to emerge or be accepted as leaders, in the aggregate we will observe that more men than women hold leadership positions in organizations, a stratification pattern that is reproduced at least partially by the way macro-level beliefs impact individual behaviors and evaluations. By focusing on the role of differentiated performance expectations, expectation states theory provides a unifying explanation for how reward structures, behavioral patterns, and macro-level beliefs about a diverse array of social categories produce similar effects on the organization of interactional status hierarchies, the building blocks of societal stratification. It helps us understand how inequitable structures emerge in these smaller structures, which increases our understanding of the emergence and reproduction of inequality in society more generally.
REFERENCES
Aries, E. (1996). Men and women in interaction: Reconsidering the differences. New York; Oxford. Bales, R. F. (1950). Interaction process analysis: A method for the study of small groups. Reading, MA; Addison- Wesley. Bales, R. F. (1970). Personality and interpersonal behavior New York; Holt, Rinehart, and Winston. Balkwell, J. W. (1991). From expectations to behavior; An improved postulate for expectation states theory. American Sociological Review, 56, 355-369. Berger, J., Conner, T. L., & Fisek, M. H. (Eds.). (1974). Expectation states theory: A theoretical research program. Cambridge, MA; Winthrop. Berger, J., Fisek, M. H., Norman, R. Z., & Wagner, D. G. (1985). The formation of reward expectations in status situa- tions. In J. Berger & M. Zelditch (Eds.), Status, rewards, and influence (pp. 215-261). San Francisco; Jossey-Bass. Berger, J., Fisek, M. H., Norman, R., & Zelditch, M. (1977). Status characteristics and social interaction. New York; Elsevier.
