The equal opportunity illusion: The effects of prejudice and power on information seeking, employee evaluation, task assignment, and estimates of employee success

Contents

Acknowledgements

Abstract

Introduction

Method

Results

Discussion

References

Author's Note

Tables..

Table 1. The effects of participant prejudice level and social influence strategy

on selection of leading questions, skills test assignment, final task assignment, employee evaluations and estimated success.

Appendices..

illustration not visible in this excerpt

Acknowledgments

I am especially grateful to Dr. Theresa K. Vescio for her continuous support and help throughout the completion and revision of this thesis.

Her knowledge, guidance and commitment of time as well as her contribution of ideas and different points of view helped me gain new insight and inspiration.

Abstract

This study tested the effects of individual variables (prejudice level) and situational factors (power instructions) on information seeking strategies, employee evaluation, estimation of likely success, and task assignment in an employer – employee, ethnicity relevant experimental design, with subjects always assigned the role of employer and an ostensibly other person (a same gender black individual depicted in a photograph) assigned the role of employee. Subjects (N=60) were categorized into groups that varied on power (exclusive or inclusive leadership instruction) and prejudice (quartile split of MRS scores). Participants were asked to select a subset of questions and tasks from various lists for the ostensibly other subject to answer. Participants at a later point in the experiment rated selected questions and tasks. At the end of the experiment the participants were asked to give a final employee evaluation and estimation of likely success for a future project.

Next to the attempt of replicating generally accepted and expected interrelations of power and prejudice with certain attention (information – seeking) strategies and the use of stereotypes and their effect on evaluation and estimation, one of the main focuses of this study is on the effects of the above variables on behavior (final task assignment).

Consistent with predictions participants with a low prejudice level assigned more valued tasks, focused more on strength of the employee and estimated greater employee success than did high prejudice participants. Also participants with inclusive leadership instructions assigned relatively more skill tests with supporting help and estimated greater employee success than participants with exclusive leadership instructions. Interaction - effects across the skills test- information seeking-, employee evaluation-, final task assignment-, and estimated success- variables showed that high prejudiced participants in the exclusive leadership style condition respond in stereotype consistent ways significantly more often than participants in the inclusive leadership condition and low prejudice participants.

The Equal Opportunity- Illusion: The effects of prejudice and power on information seeking, employee evaluation, task assignment, and estimates of employee success

The front page of the New York Times on Friday, March 30th 2001 read “California Census Confirms Whites Are In Minority – State now most diverse”. National projections further indicate that in the United States “persons of color will constitute a numerical majority sometime between the years 2030 and 2050” (D.W. Sue & D. Sue, 1999, quoting the U.S. Census Bureau, 1992; D.W. Sue et al., 1997).

Despite the fact that immigration factors and differential birth rates will soon lead to equivalent numeric representation among members of different ethnic groups in the United States (D. W. Sue & D. Sue, 1999), there are still large differences in the relative social status and social power conferred to members of different social groups. Members of dominant groups (e.g., White men) more often hold high social status, high power positions than do members of disadvantaged groups (e.g., women, people of color). For example, 95% of all senior management positions (vice president and above) and 85% of faculty members at institutions of higher education are held by white males (Sklar, 1999).

Similar differences in social status can be found among ethnic groups in societies around the world. As Sidanius, Levin, and Pratto (1998) note, most “societies tend to be stratified along at least one salient social dimension (e.g., along racial, ethnic, tribal, socioeconomic, religious lines)” and systems of social stratification tend to be remarkably stable. For example, the Indian caste system has remained intact for over 3,000 years, whereas White European Americans have dominated and oppressed African-Americans for hundreds of years (Smith, 1991). More importantly, some social stratification systems have endured for hundreds of years despite the fact that members of oppressed groups actually represented a numeric majority (e.g., the Apartheid regime remained in tact in South African until 1994).

Thus, it seems unlikely that a shift in the numeric representation of various ethnic groups in the United States will dismantle the well established and longstanding confound between social group membership and social power. Instead, repeating historical trends, it seems more likely that the distribution of power will remain stable regardless of the numeric representation of people from various ethnic groups; Whites will continue to be conferred power over individuals who belong to disadvantaged social groups.

Given these realities, much recent attention has focused on the relationship between power and stereotyping. The questions of primary interest in nearly all of this research lies on who stereotypes whom and why. Consistent with foundational assumptions of the more general models of impression formation (for instance Fiske & Neuberg, 1990), power and stereotyping researchers have generally assumed that stereotyping is a default tendency. It has been assumed that social perceivers process information in the least effortful and/or cognitively demanding way (e.g., category-based processing) unless motivated to do otherwise. Building upon this assumption, it has been suggested that low power individuals should individuate and carefully process information about high power individuals because outcomes of interest to low power individuals are highly dependent upon high power others. In contrast, it has been theorized that high power individuals should stereotype low power others either because they lack the cognitive resources and/or motivation to individuate low power others (e.g., stereotyping by default) or they do so in order to maintain status differences (e.g., stereotyping by design; see Goodwin et al., 2000).

Although this line of reasoning still enjoys wide recognition today, the narrowness of this focus overlooks the social influence dynamics of power relations. While it is true that the powerful are often less dependent on low power others for valued outcomes, the outcomes of the powerful are rarely completely independent of low power others. Outcomes for powerful people often depend on successful interactions with and successful influence of low power others. For example, a high power manager may want to achieve a certain level of sales within his department, but can he do so without successfully motivating and influencing the sales representatives who sell the goods? Likewise, the head of an academic department may want to produce an internationally visible program, but can she do so without inspiring a high degree of quality work from the faculty over whom she has power? And if a person in such a position of power ineffectively or inappropriately exerts power, might such attempts to influence others actually undermine one’s goals?

Taking the above considerations into account one goal of the present research was to more thoroughly consider the motives and social influence techniques of the powerful and to consider the possibility that stereotyping is not a default. Instead the present work was designed to examine the possibility that stereotyping is a consequence of particular kinds of power situations and results in interactions with some kinds of high power people. To provide a background for the consideration of this possibility, I first consider the way in which power has traditionally been conceptualised. I then review the contemporary power and stereotype literature, both noting the way in which the powerful may perceive and act in stereotypic ways that maintain the status quo and noting the limitations in the conceptualisations of power characteristic of most contemporary work.

Conceptualizations of Power

Historically, power has been conceptualised in terms of one person’s ability to exert social influence, or control, over another person’s fate (Copeland, 1994; French & Raven, 1959; Imai, 1993; Manz & Gioia, 1983; Thibaut & Kelley, 1959). For example, Weber (1947) defined power as “the probability that one actor within a social relationship will be in a position to carry out [his or her] own will despite resistance, regardless of the basis on which this probability rests “. Importantly, a person does not actually have to influence others to have power. The potential for person A to influence person B in meaningful psychological ways, regardless of whether that influence was actually exerted, confers power to person A (Raven & French, 1959; see also Copeland, 1994). Furthermore, the ability to influence others is not limited to interpersonal exchanges. One group of people may have power over another group of people, as Sachdev and Bourhis’s (1991) definition of power stresses (i.e., the degree of control that one group has over its own fate and the fate of outgroups).

Taking the above mentioned points into consideration I propose the following definition of power, which adheres to the historical conceptualisation of power as social influence, but at the same time acknowledges the importance of considering the group vs. individual as well as the potential vs. actual influence aspects, as mentioned in the above definitions.

The following definition of power is offered:

“Power is the relative, potential or actual unilateral outcome control of one person or group over its own or another person’s or group’s fate in a social relationship”.

Power and Stereotyping

Given the traditional and present emphasis on the social influence aspect of power, as well as social psychologists’ long standing interest in the issue of stereotyping and prejudice, much recent attention has focused on the relationship between power and stereotyping. It has been suggested that cultural stereotypes and the stereotyping of subordinates justify and maintain the status quo (e.g. Jost & Banaji, 1994; Sidanius & Pratto, 1999). Furthermore, it has been proposed that high power people both exhibit stereotype consistent perceptual biases and behave in ways that make it more likely that they will elicit evidence of stereotype confirmation from low power others. At a societal level it has also been suggested that stereotyping is a major contributor to the stratification and maintenance of power differentials in both modern societies (e.g. European Americans’ dominance over African Americans in the U.S.A.) and ancient societies (e.g. caste system in India). Simply stated, “stereotyping subordinate group members is the mechanism through which dominant groups maintain the status quo” (Jost & Banaji, 1994, 1998).

Perceptual and Behavioral Confirmation

There are two means by which the powerful can come to behave in ways that justify and maintain the status quo. First, high power people may form erroneous and stereotype consistent impressions of low power others, which has been referred to as perceptual confirmation. For example, high power (as compared to low power) individuals are perceived as more fair when using punishment strategies (Molm, Quist and Wiseley, 1994) and perceived to be more accurate judges of others when, in fact, they are less accurate judges (Ebenbach and Keltner, 1998). Beyond being more likely to perceive others in a stereotypic and inaccurate manner, high power people may also behave or interact in ways that elicit stereotypic behaviors from low power others, which is referred to as behavioral confirmation. For example, Word, Zanna and Cooper (1974) demonstrated, that when interviewer participants interacted with a Black confederate interviewee, participants sat further away from the confederate, conducted shorter interviews, made less eye contact and made more speech errors as compared to those in the White confederate interviewee condition. In a second study, confederate interviewers were trained to act either like the participants in the White confederate or Black confederate conditions of the first study. Trained confederates then interviewed participants. Demonstrating that the behaviors of the relatively powerful have meaningful consequences for those low in power, independent raters evaluated the performance of the interviewee participants. Participants who were treated like the Black confederate of Study 1 (e.g., less eye contact, more speech errors) were evaluated as performing more poorly than those treated like the White confederate of Study 1.

Importantly, as power, status, and group number increase, discriminatory behaviors have been found to increase (Sachdev & Bourhis, 1991). In fact, Operario and Fiske (1998) have suggested that merely having power, and thus control, motivates bias and the tendency to stereotype and discriminate against others. According to these authors, “racial oppression derives from a) power – the disproportionate ability of some individuals or groups to control other people’s outcomes; and b) prejudice – the universal tendency to favor the in- group over the out- group. Racism functions additively from asymmetrical power and racial prejudice” (p.49). Importantly, it has been suggested that this occurs both automatically, as soon as perceivers can place others within a category, and unintentionally, even among the most well-intentioned and egalitarian people“ (Operario and Fiske, 1998).

The vicious circle of perpetual power and stereotype reinforcement

The preceding consideration might have made one thing already clear: Both stereotypes (or cognitive representations of outgroups) and prejudice (or one’s negative feelings about outgroups) may initiate a vicious circle in which biased perceptions and biased feeling reinforce one another. More specifically, sometimes power may enhance the use of stereotypes, with the latter surfacing in perceptual and/or behavioral biases. At other times the reliance on stereotypes contributes to a stratification of existing power differentials.

Therefore, on top of one-way effects of power on stereotyping and vice versa, as described above, we find that power and the two forms of stereotyping (perceptual and behavioral) may also continuously reinforce each other. Various studies (e.g. Goodwin, Operario and Fiske, 1998) show that situational control and interpersonal dominance (power) are conditions that promote motives to stereotype, leading to cognitive and judgment biases that in turn cumulatively reinforce the status quo and thereby exaggerating preexisting power differentials.

Given that power has traditionally been defined in dynamic social influence terms and that there is ample evidence that the powerful perceive and behave toward low power others in ways that make it likely that low power others will confirm (or be perceived as confirming) expectations, much recent attention has been directed to attempts that explain the relationship between power and stereotyping. The most influential model of power and stereotyping has been Fiske’s (1993) power as control (PAC) model, which is reviewed below.

The Power as Control (PAC) Model

In their investigation of the interaction between power and stereotyping, Fiske and her colleagues argue that the powerless carefully attend to the powerful who control their outcomes in an effort to enhance prediction and control. As a result, powerless people form relatively complex and non-stereotypic impressions of others who have control over them. In contrast, the powerful, pay less attention, and thus are more vulnerable to stereotyping. Fiske and her colleagues’ (e.g., Fiske, 1993; see also Dépret & Fiske, 1999; Goodwin et al., 1998; Goodwin et al., 2000) model assumes three main reasons for perceivers’ increased (as compared to targets’) use of stereotypes. These reasons are: a) high power people have more demands on their attention (lack of cognitive resources) than do low power people, b) low power people are more dependent on high power others for outcomes than the reverse, and c) the kind of people that come to hold high power positions might tend to be high on certain individual difference variables such as social dominance. Importantly, this theorizing is mainly based on the assumption that cultural stereotypes are universally known and internalized, providing a source of the expectations that the powerful bring to bear on a situation.

Interestingly, this model considers individual factors to be only of minor importance, exerting merely indirect influence on stereotype use. According to the PAC model, individual variables, such as social dominance or prejudice, are linked to people in high power positions, which might have a tendency to be high on such variables, as the theory points out (see above). Whereas Fiske et al. in their discussion about why a certain group of individuals (powerful) stereotype others more than another group of individuals (powerless) regard individual factors only relevant in this indirect way, other authors, e.g. Christiansen, Kaplan and Jones (1999), do mention such an individual factor, namely prejudice. However, in the discussion of their findings the authors agree that participants' different prejudice levels were irrelevant for the way they represent certain groups and categories internally: Findings showed that evaluations by high-prejudice participants were more negative than those of low-prejudice participants only when the target applicant was described by a single negative stereotype. The authors concluded from these findings that both groups had similarly negative predispositions toward minorities, with those of more prejudiced individuals requiring less negative stereotypical information to be activated.

The most imperative evidence supporting Fiske’s (1993) assumptions seems to be Devine’s (1989) dual process model of racial prejudice. According to Devine’s model all White Americans are aware of and have internalized the dominant cultural stereotype of African Americans and these stereotypic representations can be activated automatically. More specifically, according to Devine, this automatic activation happens in both, high and low prejudice White individuals given the mere presence of a Black individual or mere activation of the category “Black”. This, so the model continues, would lead both high and low prejudiced perceivers to judge African American individuals in relation to the cultural stereotypes. Fiske’s reasoning relies on this model; Fiske assumes that stereotypes are accessible to all high power perceivers and those in high power positions are especially prone to rely on stereotype -based expectations that guide information seeking, behavior and inferences. While highly influential and having impetus on a great deal of research, the power as control (PAC) model equates outcome control (or one’s ability to have relative greater say in an outcome) with power. The theoretical statements of the PAC model and tests of that model omit the dynamic social influence components of power relations, as noted at the outset. Also problematic are three additional points. First, as noted previously, the theory assumes that stereotypes are universally internalized, i.e., stereotyping is a default tendency. Second, hypotheses derived from this model and tests of those hypotheses have focused simply on the descriptive question of “who stereotypes whom” omitting a consideration of “why” stereotyping occurs. Thirdly, Fiske’s primary tool of measuring stereotyping is attention. This indirect method by which stereotyping has been measured most predominantly does constitute a major problem as this focus on attention has lead to a neglect of other perhaps more suitable variables, like motivation, for instance. More will be said in this regard later (see “the aspect of motivation in stereotyping”, p.20). These three points are problematic in light of the fact that there is a growing literature showing that not all individuals internalize the cultural stereotypes and numerous situations have been shown to influence stereotyping. I will review the relevant literature illustrating the problems with these assumptions in the sections that follow.

Universally internalized stereotypes?

A common feature to several theories of racial prejudice is the notion that White Americans internalize the negative cultural stereotype of Black Americans (e.g., Devine, 1989; Gaertner & Dovidio, 1986). Presumably, as a result of shared socialization experiences, all Whites are aware of the cultural stereotype, which gets represented cognitively and can be activated automatically. From this perspective the difference between high and low prejudiced people lies not in the presence or absence of the cognitive representations (all people have stereotypic representations), but in the conscious control that is exerted in attempts to control stereotypic responding. It has been suggested that low prejudiced people “put on the brakes” and try to limit the influences of stereotypes in situations in which they have the resources and awareness to control stereotypic responding.

While there is no disputing the fact that White Americans tend to be universally knowledgeable of the cultural stereotypes of Black Americans, recent findings question the notion that such knowledge is similarly internalized by both high and low prejudiced people and automatically activated. Additionally, there is a growing body of evidence noting that situational factors importantly influence whether stereotypic perception and behavior ensues. All of these findings stand in contrast to the assumption that stereotypes are universally internalized, thus questioning a fundamental assumption upon which Fiske’s (1993) PAC model was derived.

Individual Differences in stereotyping

While many theories have assumed that all people have cognitive representations of the dominant cultural stereotype, which is automatically activated in the mere presence of an African American individual, recent research demonstrates that there are critical differences in the cognitive representations that high and low prejudiced Whites have of African Americans. Lepore and Brown (1997), for example, showed that the relations among categorization, stereotyping, and prejudice are more flexible than is often assumed. It was shown that differences in responses between high- and low- prejudice people occurred when a category, rather than stereotypic traits, was subliminally primed. When the category Blacks was subliminally primed high -prejudice participants formed a more negative and less positive impression of the target person than did participants in the no prime condition, with low -prejudice people tending in the opposite direction. But when valenced stereotype content was primed, high- and low- prejudice people increased negative ratings. These differences were found despite the fact that the two groups of high-and low prejudice people share the same knowledge of the stereotypes of Black people, as could be shown in an earlier study (study 1).

Thus, high-and low-prejudice people differ in their automatic responses to category activation, which suggests that they hold different representations because of their beliefs and despite their common stereotype knowledge. This directly opposes the notion of many authors that most White Americans hold negative stereotypes of Black Americans (e.g. Gaertner & Dovidio, 1986), which are activated automatically when confronted with a member of that category, as Devine’s (1989) “mere presence” -argument states (see page 12) or Fiske & Neuberg (1990) conclude.

Taking the above arguments of Lepore and Brown into consideration we can no longer talk about universally internalized stereotypes, but instead state that high – and low- prejudiced individuals have different stereotypes. According to Vescio and Biernat (1999), both high and low prejudiced people have internalized representations of group differences in social status (e.g., blacks lower social status, lower income, less educated), but high and low prejudice people differ in their explanation of such differences in social status. Whereas low-prejudiced individuals primarily regard situational components of stereotypes as the cause (discrimination, prejudice etc. lead to blacks’ lower social status), high-prejudiced individuals believe in a dispositional causation for those differences (e.g., blacks are lazy, unmotivated, unintelligent…). That is, only high prejudiced people have the traits consistent with the cultural stereotype. As a result both groups have different expectations, which eventually can be violated. Supporting evidence for this line of reasoning of differing stereotype content between high- and low prejudice people comes from various studies. Vescio & Biernat (1999), for example showed that individual differences in prejudice level affect conditions under which evaluations of outgroup targets are more extreme and stereotypic. The authors hypothesized that high prejudiced White perceivers, having internalized the negative cultural stereotype of Black Americans, would expect a lack of intellect as well as lack of motivation in Blacks. On the other hand, low prejudiced Whites, rejecting the cultural stereotype, were hypothesized to expect academic achievement and success. Violations of these expectations were assumed to lead to more extreme appraisals in such that evaluations become more extreme in the direction of the expectancy violation. The results showed precisely that. High prejudiced perceivers evaluated an expectancy violating high-quality lecture given by a Black professor more extremely favorable (i.e., rated much more favorably than a similar lecture administered by a White professor). In contrast, low prejudiced Whites evaluated an expectancy violating poor quality lecture given by a Black professor much more extremely and in a negative direction.

Another study, supporting the idea of differences between high- and low prejudiced individuals’ representations and explanations of stereotypes was offered by Larsen et al. 1980 (cited by Jones 1997, p.156). In this study, participants who scored high on the Modern Racism Scale (MRS; McConahay, 1986), unlike low scoring subjects, did not suppress the effects of stereotypes on their judgments when given a chance to do so. Moreover, when they received feedback suggesting they were biased toward either black males or gay men, those with negative attitudes (as measured by the MRS for Blacks or the Heterosexual Attitude toward Homosexuals [HATH] for gay men) felt no compunction about their biases, but those low in modern racism or anti-gay attitudes felt emotional distress when made aware of their biases.

Extending and investigating even further the idea of different stereotype representations, others have proposed (based on various studies investigating prejudices against Blacks, homosexuals, and fat people) that low prejudiced people might have such situational explanations as a result of humanitarian/egalitarian self-concepts, which are negatively associated with all measures of prejudice and discrimination. This self-concept, so the authors argue, does serves as a „prejudice antidote“, as was demonstrated by the association of Protestant Work Ethic (PWE) values with increased prejudice toward Blacks homosexuals, and fat people (e.g. Biernat, Vescio, Theno & Crandall, 1996).

The importance of inter-individual differences and their influence on the accessibility of and the willingness to use stereotypes is further highlighted by social dominance theorists (e.g., Sidanius & Pratto, 1999) and their proposition of a construct, which is called “Social Dominance Orientation” (SDO); an individual difference variable, defined as “the degree to which individuals desire and support group-based hierarchy and the domination of ‘inferior’ groups by ‘superior’ groups” (p.48). People high in SDO, for example more strongly endorse racial stereotypes, attributing the lower social status of racial outgroups to characteristics of individual group members.

Differences between high- and low- prejudiced people appear at an automatic level, supporting the idea of a flexible link between categorization and stereotyping. “It is endorsement, not knowledge that is likely to shape the representation in memory, strengthening the links between the category label and certain stereotypic features instead of others. Prejudice does not resemble a habit that has to be broken (referring to Devine’s assumption, 1989) but one that is, for some people, already broken” and “it is not knowledge which strengthens the links between category and stereotype, it is rather the endorsement” (Lepore and Brown, 1997). The contemporary findings contemplated above, argue against the universal internalization theory and clearly support the idea that high prejudice persons and low prejudice persons have different representations of groups. Moreover, this is the case because of their beliefs and despite their common knowledge about stereotypes!

Situational differences in stereotyping

Historically, there have been various approaches to the understanding the concept of prejudice (for an overview see Jones, 1997, p.143ff). Social psychology has mainly regarded prejudice as an individual-level phenomenon. Zanna (1994), for instance proposed four sources of influence on the development and occurrence of prejudice, all lying within the individual:

1) Stereotypical beliefs (the notion that typical members of the group possess certain characteristics or traits),
2) Symbolic beliefs (the notion that typical target group members violate cherished majority- or dominant-group traditions, customs, and values),
3) Emotions that are aroused by a member or members of the group, and
4) Past experiences with members of the group.

Other theories for explaining prejudice that also stress the individual – level of prejudice are 1) terror management theory (cf. Solomon, Greenberg and Pyszczynski, 1991; Greenberg, Pyszczynski et al., 1990), which regards prejudice as a terror management reaction to perceived threat and heightened awareness of one’s very own mortality, 2) ego-defensiveness theory (Dollard & Miller,1941), 3)Adorno at al.’s (1951) authoritarian personality theory, or 4)Altemeyer’s (1993) right wing authoritarian personality theory (in Jones, 1997, p.145ff), to name a few.

Although, this focus on the individual level is not surprising, conceiving that we generally speak of “a prejudiced person” or “individuals who are prejudiced”, there are, apart from individual differences, also situational variables, which also have strong effects on prejudiced behavior.

Thus, an often-neglected fact is that individuals usually do not exist in isolation. Human beings are social beings, socialized in communities with certain norms and values, constantly interacting with other individuals. In trying to understand the reasons for and influences on prejudice, society itself, in providing manifold situations and contexts for interactions, has to be considered and regarded as a major source of influence on prejudice. Sidanius and Pratto’s social dominance theory (SDT, 1999), for instance, by integrating models from various fields (see previous page), does take this important aspect into account and puts much emphasis on the interface of individual and society.

Thus, situational norms, as part of situational factors in general, influence stereotyping, prejudice and/or discriminatory behavior. Dovidio and Gaertner (2000) for instance, demonstrated very clearly how ambiguity as a single situational factor might offer an excuse for aversive racists to rationalize their discriminatory behavior on the basis of some other factor than race. Ambiguity, as the situational factor in this case, therefore, can be thought of as a window, facilitating to show individuals’ true attitudes and feelings.

Another example is the already mentioned study by Vescio and Biernat (1999), in which the authors argue that low prejudiced -and high prejudiced individuals’ different stereotype-based expectancies and their respective violations may affect own performance as well as judgment of a high status target. Situational variables in that study were ethnicity of the target person (lecturer) and quality of the lecture, which demonstrably affected perceiver’s judgment and performance. In sum, a consideration of situational factors is important for a deeper understanding of the concept and mechanisms of prejudice.

The aspect of motivation in stereotyping

As pointed out above, when reviewing Fiske's PAC model (see p.11), stereotyping is normally assessed via perception or behavior. This reliance on attention as the major means of measuring stereotyping has lead to a neglect of other, possibly more valid assessment techniques. While attention has been proposed the major measurement of stereotypes, the aspect of motivation has been completely omitted when talking about the intentional reliance on stereotypic information (stereotyping by design). As noted above, Fiske and her colleagues (Fiske, 1993; Dépret &Fiske, 1999; or Goodwin et al., 1998 and Goodwin et al., 2000) do talk about cognitive load as reasons for powerful individuals’ reliance on stereotypes, but omit the aspect of motivation when attending to some but not other information.

Other authors have tried to integrate both, attention and motivation variables in their discussion of stereotyping. Supporting the cognitive load – argument as a reason for stereotyping, Bodenhausen (1993) for instance, sees affect as a possible influence on our judgment processes in that arousal limits our cognitive capacity and therefore make us more susceptible to rely on simpler and more dominant processing strategies like stereotypes. On the other hand, and more in line with the motivational aspect argument, Bodenhausen hypothesizes, the second way affect influences the use of stereotypes to be by means of motivational influences. These, so the author argues, affect our willingness to use certain cognitive capacities and result in more or less use of stereotypes, depending on the emotion involved.

More support for a stronger inclusion of motivational factors into the discussion of stereotyping by design, comes from various other studies. Reviewing literature on dissonance theory (Festinger, 1957), for example, asks for caution in the use of attention as a measure for stereotyping. Many “selective exposure” experiments (see for example Frey, 1986), have demonstrated that individuals’ sometimes intentional focus on stereotype confirming information, thereby strengthening their already existing world views (stereotyping by design), also depends on motivational factors. Frey (1986) also showed very clearly that sometimes an individual's motivation to search for worldview consistent information might even be reversed and dissonant information be filtered out. This is the case when a) a very consistent and stable cognitive system exists, which could easily disregard the dissonant information, or b) a very weak and fragile cognitive system exists, which would collapse anyway in the long run and therefore is better to be changed and revised in order to create a new kind of consonance (Stroebe, Hewstone & Stephenson, 1997).

These arguments show very clearly that attention is sometimes, but not always, related to stereotype confirming behaviors and judgments. Moreover, these arguments highlight the importance of a stronger consideration of motives when talking about “stereotyping by design”. Thus, when talking about attention as a measure of stereotyping, one only looks at one possible measure, which might be applicable sometimes, but not always. In other words, what we attend to much depends on our motivation (see Bodenhausen above) and also on the status quo of our cognitive system (see Frey above), as well as situational variables that match existing stereotypes (Vescio, Snyder and Butz, 2003). Omitting motivational factors and merely differentiating stereotyping by design from stereotyping by default processes will only portray an incomplete and misleading picture of stereotyping.

Acknowledging these important facts, a different measure of stereotyping is suggested, namely the “information seeking strategy”, endorsed by powerful subjects. Furthermore, it is argued that one important variable contributing to the way employees are evaluated, tasks are assigned, and likelihoods of success are estimated (all dependent variables in this study), is the way information is collected by the powerful. Not so much attention/inattention to certain information, but endorsement (e.g. information seeking strategies), is regarded the missing part of the “stereotype- puzzle”. In Fiske’s research the critical dependent variable is attention (mostly the amount of time spent viewing stereotype consistent or inconsistent information about the target -group), which is taken as a measure of stereotyping, offering no direct measure of stereotyping or information seeking. In fact, empirical focus almost exclusively relies on the perceptions of high - and low - power individuals, neglecting the important question of consequences.

[...]