Priming Effects and Performance Expectations in Mixed-Sex Task Groups

Abstract

I report the results of a laboratory experiment in which I examine the relationship between cognitive categorization processes and status-organizing processes, focusing on how seemingly irrelevant information becomes relevant to the informational structure of a task situation. In phase one, participants completed a task in which they were primed with photographs of women occupying either stereotypical or counter-stereotypical roles. In phase two, participants, along with a partner, completed a collective decision-making task. The two experimental phases were ostensibly unrelated from the participants’ point of view. Results indicate that priming manipulations significantly affected patterns of influence in mixed-sex groups. These effects were driven primarily by altering expectations of female group members.

Keywords

status characteristics priming gender expectation states stereotypes

Status characteristics theory (SCT) represents a well-supported explanation for the maintenance of social inequalities, focusing on the status-related behavior of actors in small, task-based groups. SCT contends that the structure of a collective task group often creates conditions in which group members draw on cultural beliefs about occupants of various social categories to assess the relative value of individuals’ contributions to the task, thereby importing macro-level status beliefs into micro-social interaction; these status-organizing processes are embedded in the “informational structure” of the interaction (Berger et al. 1977).

Considerable research indicates that the overt characteristics of social stimuli represent only part of the equation by which individuals categorize others. Research on priming effects suggests that individuals tend to categorize objects according to the construct that is most cognitively accessible and that activating a category in one setting increases the likelihood that the category will be applied in subsequent unrelated settings (Bargh, Chen, and Burrows 1996).

Recognizing that construct accessibility shapes the categorization process, construct accessibility may also shape status-organizing processes. If increasing the accessibility of a cognitive construct increases the likelihood of its application in future settings, the potential exists to manipulate the informational structure of an interaction. I report results from a laboratory experiment that examines how structurally based inequalities can be altered by priming stereotypical and counter-stereotypical gender beliefs.

Background

In categorizing other people, individuals employ a variety of physical cues, including but not limited to a person’s sex and race/ethnicity. The categorization process exerts substantial influence over how individuals evaluate and behave toward others. In his foundational work on perception, Bruner (1957) contends that perceptions of any stimulus vary depending on the perceiver’s “readiness” to categorize different properties of that stimulus. When individuals encounter a stimulus, they are likely to interpret that stimulus using the most accessible cognitive construct, often rejecting more plausible explanations (Bruner 1957). Psychologists refer to this phenomenon as cognitive priming.

As cognitive constructs, stereotypes are susceptible to priming. Priming category-relevant descriptors can facilitate the application of stereotypes to subsequently encountered stimuli (Shih, Pettinsky, and Ambady 1999). Evidence suggests that racial and gender stereotypes are cognitively available to all individuals within a given culture, regardless of personal endorsement of the beliefs (Blair and Banaji 1996; Devine 1989). Once activated, negative stereotypes shape perceptions of broad categories of people, advantaging some categories and disadvantaging others, even when individuals consciously reject the content of the stereotype.

Just as priming can increase the likelihood of stereotype application, priming with category-conflicting descriptors may inhibit the application of a stereotype. Blair and Banaji (1996) suggest that priming counter-stereotypic concepts interferes with the accessibility of the stereotypic category. Subsequent studies have shown that non-consciously activating stereotypes can have either positive or negative effects on task performance, depending on the nature of the stereotype (Shih et al. 1999).

Status Characteristics Theory

SCT examines structural inequality by exploring status-organizing processes, which are “process[es] by which differences in cognitions and evaluations of individuals . . . become the basis of differences in the stable and observable features of social interaction” (Berger et al. 1977:3). Berger and colleagues (1977) outline four stages involved in status-organizing processes: (1) cognitive and evaluative information associated with status characteristics becomes available to the actors, (2) actors “connect” the available information to features of the immediate interaction, (3) these connections promote the formation of a status hierarchy, and (4) once formed, the status hierarchy shapes behavior within the group. Relevant to the current study, higher status actors tend to be more influential over group decisions. Substantial research supports the contention that gender functions as a diffuse status characteristic (cf. Pugh and Wahrman 1983; Rashotte and Webster 2005; but see also Riordan 1983; Stewart 1988).¹

SCT and priming research share some common ground concerning the cognitive bases of social evaluation, pointing to avenues through which the two perspectives can be employed together. Following SCT, we recognize that gender is a salient characteristic in mixed-sex groups. Based on the priming literature, we can also acknowledge that the effects of gender as a status characteristic should vary in magnitude across situations depending on the relative accessibility of status-related constructs. Increasing the accessibility of status-consistent information should facilitate cognitive access to gender-based stereotypes, increasing behaviors that reflect stereotypic beliefs. Conversely, activating status-inconsistent information should impede access to stereotypes, and behaviors that reflect stereotypic beliefs should decrease.

Theoretical Models

SCT’s graph-theoretic model allows researchers to hypothesize the magnitude of status differentiation within a group with a great degree of specificity. Longer paths connecting actors to task outcomes require more cognitive effort to infer connections between actors and abilities; shorter paths produce stronger expectations, and longer paths produce weaker expectations (see Berger et al. 1977). I offer two theoretical models reflecting differing predictions about how category accessibility shapes expectations and influence in task groups. The two models reflect different assumptions about how group members incorporate information into the structure of the interaction.

Baseline model

I have argued that the relative accessibility of status-related cognitive constructs should affect the magnitude of expectations that individuals form for themselves and their interaction partners. It is possible that the traditional formulation of SCT sufficiently explains the magnitude of expectations in any collective task setting. Two possibilities exist: (1) status-based performance expectations are unaffected by category accessibility, or (2) SCT already adequately captures any such effects. To rule out this possibility, I assess how well the traditional model explains patterns of influence when status-related constructs are primed prior to group interaction.

Because I focus on groups in which gender is the only salient status characteristic, I model expectations as arising from a single diffuse characteristic. When actors are differentiated according to a single diffuse status characteristic, group members are connected to task outcomes through two lines (one for self and one for interaction partner). In this case, SCT models expectations using a graph with paths of lengths four and five, producing moderate status differences. If priming effects have no influence on expectations or if such effects are already accounted for in the traditional model, modeling expectations as a single diffuse characteristic should best account for behavior.

Priming model

Path lengths represent cognitive connections between group members and task outcomes: longer paths require more cognitive effort, and shorter paths require less. To model the effects of priming on expectations, I assume that more accessible constructs require less cognitive effort to infer connections, producing shorter paths between actors and task outcomes. When primes interfere with the categorization process, the need for increased cognitive effort should produce longer paths. Priming actors with status-consistent information should increase the magnitude of status differences, and priming actors with status-inconsistent information should reduce those differences.

When members of mixed-sex groups are primed with status exemplars, patterns of influence should vary based on the accessibility of status beliefs. When exemplars activate stereotype-consistent beliefs, inequality should increase. When exemplars present stereotype-inconsistent beliefs, inequality should decrease. I predict that in explaining patterns of influence, the priming model—which explicitly models the strengthening effects of stereotype-consistent primes and the weakening effects of stereotype-inconsistent primes—is preferable to the baseline SCT model, which does not directly account for prior activation of status-related cognitive constructs. Although the priming task I employ falls outside of SCT’s formal collective task scope, Lovaglia et al. (1998) demonstrate that status-based performance expectations can arise in situations lacking a collective focus.

Methods

I present results from a laboratory experiment testing how priming shapes gender-based performance expectations. The experiment employs a between-subjects design employing participant gender, status differences (status-equal versus status-differentiated), and priming effects (high-status exemplar versus low-status exemplar versus no priming manipulation) as independent variables.

Participants

A total of 269 volunteers from undergraduate courses at a large midwestern university participated. Participants completed an online prescreening survey. To reduce suspicion about the connection between experimental tasks, the prescreening survey website informed volunteers that they were registering to participate in two separate studies: (1) a study exploring attitudes toward media portrayals of women and (2) a study examining decision making in computer-mediated interactions. Volunteers received extra credit in one course.

Procedure

The experimenter greeted each participant and escorted him or her to a private room to complete the study in isolation. Participants signed an informed consent document that notified them that they were participating in two separate tasks. The experimenter reinforced that the two tasks represented separate and unrelated studies, which is consistent with studies examining priming effects. Participants were randomly assigned to experimental conditions. All task instructions were administered via computer.

Phase one. In phase one, participants completed a “photograph rating task,” ostensibly assessing attitudes about media depictions of women. The true purpose of the task was to administer priming stimuli. To avoid raising suspicions about any connection between the two experimental tasks, participants were informed only that the task was designed to examine “how people feel about various visual portrayals of women in the media.” Participants viewed a series of ten photographs depicting women as either low status actors or high status actors. Photographs conveyed a number of status cues, based on the woman’s setting, attire, and behavior. Low status primes depicted women engaged in gender-stereotypical behaviors, such as housework or low-wage domestic labor. Women in high status images were dressed in attire suggesting professional status and inhabiting stereotypically high-paying jobs that require high levels of education. Each photograph depicted only one person.²

Participants viewed photographs one at a time for five seconds each against a black background on a computer monitor. Pretests revealed that five seconds was sufficient for participants to process the status-relevant aspects of the photographs. After viewing each photograph, participants provided ratings of the status exemplar’s goodness, potency, and liveliness. Participants were instructed to make quick judgments about the women in the photographs and that ratings should be based on their “initial gut reactions.” To reduce pressures to provide socially desirable responses, I reminded participants that all of their ratings were confidential. Participants’ ratings provided a manipulation check to ensure stimulus photographs activated the intended constructs. All photographs were pretested to ensure the strongest possible priming manipulation.

Phase two

Next, participants were instructed that they would participate in a second study involving interaction with a partner via a computer network to complete a “contrast sensitivity” test. During the contrast sensitivity task, the computer displayed a series of slides depicting pairs of black and white boxes. Participants were instructed to determine which of the two boxes contained more white area. In truth, both contained an equal amount of white space, though participants were led to believe that there was a single correct answer for each trial.

Each contrast sensitivity trial consisted of three stages: (1) participants recorded an initial decision, (2) the program informed participants of their partner’s initial decision, and (3) after learning their partner’s decision, participants were given the option to either stay with their initial decision or to change their initial choice. This process was repeated over 25 trials, and partner feedback was manipulated so that 20 critical trials generated disagreement between the participant’s initial choice and the partner’s initial choice. When actors changed their initial choice in response to partner feedback, we assume influence has occurred.

Status manipulation

I manipulated participants’ relative status by providing fictitious demographic information. Participant gender served as the basis for status differentiation, and I conveyed status-relevant information through a fabricated exchange of “background information” between partners, in which participants learned their partner’s name. Female participants were informed that they were interacting with a male undergraduate named “Michael,” and male participants were informed they were interacting with a female undergraduate named “Jessica.” Participants in “status equal” conditions did not receive any information regarding their partner.

Lastly, participants completed a postexperimental questionnaire designed to ensure that the interaction met SCT’s scope conditions and answered questions assessing their subjective evaluations of their interaction partners. Participants were debriefed about all deceptive aspects of the study. Figure 1 presents an overview of the study design (six status-equal control conditions and six experimental conditions).

Figure 1.

Overview of Study Design: Male and Female Participants per Condition

Dependent variables

Employing participants’ contrast sensitivity data, I utilize trial-level data, examining the status-influence relationship using each participant’s unique pattern of “stay” and “change” responses across critical trials. In comparing models, I employ condition-level count data for the number of “stay” responses each participant reports. This involves aggregating trial-level data for each participant.

Independent variables

I employ theoretically derived expectation advantage values to model each participant’s status (dis)advantage relative to his or her partner. Based on the two theoretical models described previously, I derived expectation advantage values using Balkwell’s a priori path functions (expectation advantage values and path functions available upon request), which I use to predict openness to influence.

Analytic Strategy

My analyses follow a two-step process. First, I employ generalized estimating equations (GEE) to analyze trial-level data (see Kalkhoff 2005). GEE is ideal for these data for two reasons: (1) the dependent variable (“stay” responses) represents a dichotomous outcome measured over 20 trials, and (2) the effects of cognitive primes can decay rapidly. Examining the interaction between trial and expectation advantage allows me to assess whether priming influences earlier trials more strongly than later trials. This latter point is particularly important, given that I argue that priming effects shape performance expectations that are formed early in interaction. GEE analyses provide statistics for planned contrasts between conditions as well as an assessment of overall goodness of fit for both theoretical models.

Second, I compare my priming model against the baseline model using C-tests for non-nested models (Weakliem 1992). The C-test provides an assessment of the extent to which the priming model explains variance in openness to influence beyond that which is explained by the baseline model.

Results

I report results from 240 participants (120 female, 120 male). I exclude data from 29 participants (10.78 percent) from analysis for various reasons. The average SCT experiment excludes approximately 15 percent of participants (Dippong 2012), and my percentage is well below that mark. I excluded eight participants who expressed clear suspicions concerning the contrast sensitivity task, five participants who misunderstood important task instructions, and one participant based on equipment malfunctions. I excluded fifteen participants for failure to meet SCT’s scope conditions (nine for insufficient collective orientation and six for insufficient task orientation).

I employed a post-experiment interview to assess whether or not participants perceived that the two phases of the study were related. Bargh and Chartrand (2000) suggest that in probing participants’ suspicions about the relationship between priming tasks and behavioral measures, suspicion rates above 5 percent should raise concerns. I thoroughly questioned each participant using Bargh and Chartrand’s (2000) “funneled” debriefing procedures and found no instances of suspicion regarding the connection between tasks. And based on photograph rating data, I determined that the priming manipulation activated the expected cognitive constructs for all participants. Table 1 presents descriptive statistics for the analytic sample as well as the mean number of “stay” responses for all conditions involving a salient status characteristic.

Table 1.

Sample Descriptive Statistics and Condition Means

Variable	Mean
Age	19.53
Sex
Male	.50
Female	.50
Race/Ethnicity
White (non-Hispanic)	.80
Black/African American	.08
Other	.12
Condition	Mean “stay” count
Male participants
Higher status	11.95
Higher status, inconsistent prime	11.10
Higher status, consistent prime	11.80
Female participants
Lower status	10.70
Lower status, inconsistent prime	11.60
Lower status consistent prime	8.80

Table 2 presents GEE analyses examining the main effects of condition, status, and priming on trial-level data, as well as the interaction between status and prime. The overall effect of condition on openness to influence is significant (Wald chi-square = 22.76, df = 7, p = .002), suggesting that further analysis is warranted. Examining the effects of status and priming separately reveals that while the status manipulation produced a marginal result (Wald chi-square = 4.71, df = 2, p = .095), the effects of the priming manipulation were substantial (Wald chi-square = 6.61, df = 2, p = .037), as well as the status by priming interaction (Wald chi-square = 6.65, df = 1, p = .009).

Table 2.

Planned Contrasts, Main Effects, and Interactions

Contrasts	df	Wald Chi-Square	p
Male versus female: inconsistent prime		.37	.543
Male versus female: consistent prime		10.74	.001
Male consistent versus male inconsistent		.28	.600
Female consistent versus female inconsistent		10.79	.001
Main effects and interactions^a
Condition	7	22.76	.002
Status	2	4.71	.095
Prime	2	6.61	.037
Status × prime	1	6.65	.009

Estimates include control conditions with no salient status characteristics.

Planned contrasts between conditions further demonstrate the effects of priming on openness to influence. Differences across priming levels appear to be driven almost entirely by changes in openness to influence among female participants. Males primed with stereotype-inconsistent images do not differ from those primed with stereotype-consistent images (Wald chi-square = .28, p = .600). Females, however, differ significantly based on priming condition (Wald chi-square = 10.79, p = .001). Thus, I will compare theoretical models using my complete sample and conduct auxiliary analyses estimating separate models for male and female participants.

Comparing Models

Table 3 presents GEE results comparing the two theoretical models. Expectation advantage significantly predicts patterns of influence in both models, and trial number has no significant effect on the relationship between expectation advantage and influence. These results suggest that status-based expectations remained stable across the entire interaction: despite the passing of time, the differentiating effects of priming and of status positions did not decay.

Table 3.

Generalized Estimating Equations Results for Two Models

	df	Wald Chi-Square	p
Model 1: baseline status characteristics theory model
Expectation advantage	1	4.81	.029
Advantage × trial	19	24.81	.167
Model 2: priming effects model
Expectation advantage	1	10.85	.001
Advantage × trial	19	18.86	.466

Although initial results provide support for SCT in general, on a strictly accept or reject basis, GEE analyses provide no indication that either model can be discarded outright. This result is not entirely surprising, given the degree of correlation between models. Compared to the baseline model, the priming model does not indicate a different relationship between expectation advantage values and influence but whether priming strengthens or weakens expectations.

Because both models provide an acceptable fit to the data, it is necessary to compare models to assess which best accounts for variation in influence. To compare models, I employ C-tests (Weakliem 1992). Melamed (2013:225) states,

The C test requires three steps: (1) Estimate the predicted cell counts under two different models, and let f refer to the cell counts under the null model and let g refer to the cell counts under the alternative model; (2) Define a variable h = g – f, and then regress y – f on h using 1/f as a weight (where y refers to the observed cell counts); and, (3) Divide the coefficient for h by its standard error to obtain the C-statistic.

A significant C-value indicates that the alternative model is preferable to a null (baseline) model.

Following the procedures described previously, I compare the priming model to the baseline model, producing a value of C = 2.361; p = .02. In line with my predictions, I conclude that the priming model provides a significant improvement over the baseline model.

Auxiliary Analyses

Because results from planned comparisons suggest that male and female participants responded to the priming manipulation differently, I conducted further analysis, estimating GEE models separately for male and female conditions. In Table 4, it is evident that for the priming model, expectation advantage significantly predicts openness to influence in female participants (p < .001), but for male participants, expectation advantage does not significantly predict influence. These results show that differences in openness to influence across priming conditions were driven almost entirely by female participants’ responses to the priming manipulation. I discuss possible reasons for this difference in the following section.

Table 4.

Comparing Model Fit across Male and Female Conditions^a

	df	Wald Chi-Square	p
Model 2: priming effects model
Expectation advantage male	1	.67	.412
Expectation advantage female	1	15.21	<.001

Unable to estimate baseline status characteristics theory model due to a lack of degrees of freedom.

Turning again to C-tests, I compare models across female conditions to determine if either model is preferred. Analyses demonstrate that the priming model is a significant improvement over the baseline (C = 2.640; p =.01). Again in line with my predictions, I conclude that the priming model is preferable to the baseline model.

Discussion

Cultural beliefs about gender are deeply entrenched, and because they often shape behavior automatically at a nonconscious level, they can provide an appearance of innate differences, and their effects are difficult to overcome (Ridgeway 2011). Because status beliefs about the unequal abilities of occupants of certain social categories originate at the micro-social level and then diffuse throughout a society (Ridgeway 2011), it is theoretically possible to change cultural beliefs by invoking equality in micro-social interaction.

The present study began by asking if structurally based inequalities can be reduced by manipulating the relative accessibility of gender stereotypes. Analyses of two theoretical models examining the possible contribution of priming effects to status-organizing processes within the SCT framework reveal the promising finding that priming females with high status female exemplars can reduce the degree of status inequality in mixed-sex groups. At the same time, results pertaining to priming females with low status female exemplars suggest a possible mechanism by which disadvantaging gender beliefs activated outside the group setting are imported into group interactions. By exploring the role of construct accessibility in the formation of status hierarchies, the current study points to new avenues for status interventions.

The observed differences between male and female participants suggest different processes by which male and female participants processed status-relevant information. Female participants incorporated priming effects into the informational structure of the interaction by rendering status differences more or less relevant to the group task. In essence, females behaved as if the primes contributed additional information to the structure of the interaction beyond that which could be deduced by simply knowing that they were interacting with a male partner. C-tests indicate that the priming model provided a significant improvement over the baseline model, suggesting that priming females with high status exemplars created weaker inferences about the relationship between gender and task abilities.

Conversely, priming effects were effectively nonexistent for males; the high status primes were insufficient to activate beliefs that were inconsistent with existing beliefs or to interfere with the categorization process. And in the case of low status primes, because they were already consistent with existing gender beliefs, they provided males with no further relevant information. In other words, knowing that they were interacting with a female partner was sufficient to develop expectations, and the prior activation of gender-related status beliefs did not alter their perceptions of their partner.

The most noticeable difference between the experiences of male and female participants in the present study is that female participants were exposed to primes that conveyed self-relevant information while males were exposed to primes with other-relevant information. Because all participants rated photographs of women, the possibility exists that employing photographs of males would produce different results. One potential explanation for the differences reported here is that the self-relevance of the priming stimuli shaped cognitive and behavioral responses. Although the present data are not sufficient to determine if this is in fact the case, a brief theoretical discussion may help illuminate the underlying processes and lay groundwork for future research.

A substantial amount of research has demonstrated that individuals are particularly attuned to self-relevant aspects of social situations. For example, Shih and colleagues (2002) argue that when making social judgments about others, actors attend more closely to self-relevant traits. Thus, females and males likely engaged the priming photographs with different levels of cognitive processing, in which the self-relevant nature of the stimuli for females rendered the information contained in the photographs more salient.

The role of self-relevance in processing social information suggests that an effective approach to altering the cognitive accessibility of gender beliefs may require targeting males and females differently with self-relevant information. Shih and colleagues (2002) argue that individuals who are the target of a prime (in the case of the present study, females) respond to subtle stereotype activations while non-targets require a very blatant activation. More generally, females tend to possess more complex gender schemas (Markus et al. 1982), and as such, males tend to pay attention to information relevant to their own sex (Martin 1993). If self-relevance is driving differences in reactions to primes, then male and female participants should respond only to stimuli that prime beliefs about their own gender. If gender schematicity is driving differences, it may be possible to control expectations for males and females using stimulus photographs of males. Research is currently underway that will allow me to differentiate the effects of self-relevance and gender schematicity on reactions to gender-based priming.

In many ways, SCT and research on priming effects deal with parallel processes. While research on priming effects has shed light on the ways in which people categorize and evaluate each other, SCT has developed around a sociological framework centering on how cultural beliefs about social categories provide a basis for inequality at a micro-social level. Given the potential for the explanatory capacity of SCT to be enhanced by developing an integrated account for the construction and maintenance of status hierarchies that incorporates categorization processes and status-organizing processes, the most direct contribution of the present research is in opening a line of communication between the two research traditions.

A second and more noteworthy contribution of the present study is that it points to new avenues of status intervention. If the relative accessibility of status-disadvantaging beliefs can be altered by changing visual cues within a situation, this provides a way forward for reducing inequality at the micro-social level. By incorporating priming effects into SCT, the current study opens the door to exploring status-organizing processes with an eye toward bringing in and accounting for new sources of information. Further research is needed to elaborate the utility of the present study for explaining and addressing issues of gender-based inequality.

Footnotes

Acknowledgements

I would like to thank Will Kalkhoff, Stan Gregory, Richard Serpe, Carla Goar, Val Callanan, and Murray Webster as well as three anonymous reviewers for helpful feedback on previous drafts of this paper.

Editors’ Note

Both editors serve as editor-in-chief on all manuscripts. In this article, Jan E. Stets served as the editor-in-chief because Richard T. Serpe had a conflict of interest.

1

describe mechanisms by which individuals combine multiple pieces of status information.

2

The decision to employ only stimulus photographs depicting females was based on the overarching goal of the present study. One approach to reducing gender-based inequality is to decrease expectations for males. The aim of this study is to develop a status intervention to increase equality by increasing expectations for female group members.

Bio

Joseph Dippong is an assistant professor of sociology at University of North Carolina, Charlotte. His general research interests include status and power processes in small groups, social psychological explanations for crime and delinquency, and the interplay between social structure and biology. He is currently conducting research on the role of vocal accommodation in communicating status-based performance expectations in cooperative and competitive task groups.

References

Bargh

John A.

Chartrand

Tanya L.

2000. “Studying the Mind in the Middle: A Practical Guide to Priming and Automaticity Research.” Pp. 253–85 in Handbook of Research Methods in Social Psychology, edited by Reis

Judd

New York: Cambridge University Press.

Bargh

John A.

Chen

Mark

Burrows

Lara

. 1996. “Automaticity of Social Behavior: Direct Effects of Trait Construct and Stereotype Activation on Action.” Journal of Personality and Social Psychology 71(2):230–44.

Berger

Joseph

Fisek

M. Hamit

Norman

Robert Z.

Zelditch

Morris

Jr . 1977. Status Characteristics and Social Interaction: An Expectation-States Approach. New York: Elsevier.

Berger

Joseph

Norman

Robert Z.

Balkwell

James W.

Smith

Roy F.

1992. “Status Inconsistency in Task Situations: A Test of Four Status Processing Principles.” American Sociological Review 57(6):843–55.

Blair

Irene V.

Banaji

Mahzarin R.

1996. “Automatic and Controlled Processes in Stereotype Priming.” Journal of Personality and Social Psychology 70(6):1142–63.

Bruner

Jerome S.

1957. “On Perceptual Readiness.” Psychological Review 64(2):123–52.

Devine

Patricia G.

1989. “Stereotypes and Prejudice: Their Automatic and Controlled Components.” Journal of Personality and Social Psychology 56(1):5–18.

Dippong

Joseph

. 2012. “The Effects of Scope Condition-Based Participant Exclusion on Experimental Outcomes in Expectation States Research: A Meta-Analysis.” Social Science Research 41(2):359–71.

Kalkhoff

Will

. 2005. “Collective Validation in Multi-Actor Task Groups: The Effects of Status Differentiation.” Social Psychology Quarterly 68(1):57–74.

10.

Lovaglia

Michael J.

Lucas

Jeffrey W.

Houser

Jeffrey A.

Thye

Shane R.

Markovsky

Barry

1998. “Status Processes and Mental Ability Test Scores.” American Journal of Sociology 104(1):195–228.

11.

Markus

Hazel

Crane

Marie

Bernstein

Stan

Siladi

Michael

. 1982. “Self-Schemas and Gender.” Journal of Personality and Social Psychology 42(1):38–50.

12.

Martin

Carol L.

1993. “New Directions for Investigating Children’s Gender Knowledge.” Developmental Review 13(2):184–204.

13.

Melamed

David

. 2013. “Do Magnitudes of Difference on Status Characteristics Matter for Small Group Inequalities?” Social Science Research 42(1):217–29.

14.

Pugh

M. D.

Wahrman

Ralph

. 1983. “Neutralizing Sexism in Mixed-Sex Groups: Do Women Have to Be Better Than Men?” American Journal of Sociology 88(4):746–62.

15.

Rashotte

Lisa Slattery

Webster

Murray

Jr . 2005. “Gender Status Beliefs.” Social Science Research 34(3):618–33.

16.

Ridgeway

Cecilia L.

2011. Framed by Gender: How Gender Inequality Persists in the Modern World. New York: Oxford University Press.

17.

Riordan

Cornelius

. 1983. “Sex as a General Status Characteristic.” Social Psychology Quarterly 46(3):261–67.

18.

Shih

Margaret

Ambady

Nalini

Richeson

Jennifer A.

Fujita

Kentaro

. 2002. “Stereotype Performance Boosts: The Impact of Self-Relevance and the Manner of Stereotype Activation.” Journal of Personality and Social Psychology 83(3):638–47.

19.

Shih

Margaret

Pittinsky

Todd L.

Ambady

Nalini

. 1999. “Stereotype Susceptibility: Identity Salience and Shifts in Quantitative Performance.” Psychological Science 10(1):80–83.

20.

Stewart

Penny

. 1988. “Women and Men in Groups: A Status Characteristic Approach to Interaction.” Pp. 69–85 in Status Generalization: New Theory and Research, edited by Webster

Jr. Foschi

Stanford, CA: Stanford University Press.

21.

Weakliem

David L.

1992. “Comparing Non-Nested Models for Contingency Tables.” Sociological Methodology 22:147–78.