A recent paper claims:
Racial and gender stereotypes impact virtually every important life outcome, from job interviews to job placement, from housing to education, from police stops to prison terms. For example, Blacks in the United States are less likely to land job interviews than identical White applicants (Bertrand & Mullainathan, 2004) and to receive harsher sentences compared to White perpetrators of similar crimes (Klein, Petersilia, & Truner, 1990; Pettit & Western, 2004). Experimental evidence suggests that these disparities can be partly attributed to stereotypes of Blacks as more aggressive and less intelligent than Whites (Sommers, & Ellsworth, 2000). Gendered Races: Implications for Interracial Marriage, Leadership Selection, and Athletic Participation
Let’s unpack the statement:
Claim #1: Group stereotypes have a pervasive impact on individuals
Claim #2: Group stereotypes partially cause outcome differences between groups.
It will be noted that claim #1 does not imply claim #2.
Now, in defense of claim #1, the authors pick a few case studies and ignore the meta-analyses. Importantly, they neglect to quantify effect. As discussed in, “The Unbearable Accuracy of Stereotypes,” the effect of stereotypes on the individual level is small.
What Do People Do When They Judge Individuals?
People should primarily use individuating information, when it is available, rather than stereotypes when judging others. Do they? This area of research has been highly controversial, with many researchers emphasizing the power of stereotypes to bias judgments (Devine, 1995; Fiske & Neuberg, 1990; Fiske & Taylor, 1991; Jones, 1986; Jost & Kruglanski, 2002) and others emphasizing the relatively modest influence of stereotypes and the relatively large role of individuating information (Jussim et al, 1996; Kunda & Thagard, 1996).
Fortunately, literally hundreds of studies have now been performed that address this issue, and, even more fortunately, multiple meta-analyses have been performed summarizing their results. Table 3 presents the results from meta-analyses of studies assessing stereotype bias in many contexts. It shows that the effects of stereotypes on person judgments, averaged over hundreds of experiments, range from 0 to .25. The simple arithmetic mean of the effect sizes is .10, which is an overestimate, because the meta-analyses with more studies yielded systematically lower effect sizes (r = -.43 between effect size and number of studies). The few naturalistic studies of the role of stereotypes in biasing person perception have yielded similarly small effects (e.g., Clarke & Campbell, 1955; Jussim et al., 1996; Madon et al, 1998).
How small is an effect of r = .10? It is small according to Cohen’s (1988) heuristic categorization of effect sizes. It is among the smallest effects found in social psychology (Richard et al, 2003). An overall effect of .10 means that expectancies substantially influence social perceptions about 5% of the time (as per Rosenthal’s  binomial effect size display). This means that stereotypes do not influence perceptions 95% of the time.
In general, therefore, based on over 300 experimental studies and a smaller number of naturalistic studies, stereotypes have only very modest influences on person perception. Of course, there is always the possibility that researchers have not searched in the right places or in the right way for powerful stereotype biases in person perception. At minimum, however, the burden of proof (for the existence of widespread, powerful stereotype biases in person perception) has shifted to those emphasizing such powerful biases.
The existence of small stereotype effects, however, does not necessarily mean that people do generally rely heavily on individuating information. But the empirical evidence shows, in fact, that they do. The one meta-analysis that has addressed this issue found that the effect of individuating information on person perception was among the largest effects found in social psychology, r = .71 (Kunda & Thagard, 1996). In other words, people seem to be generally doing the right thing – relying on individuating information far more than stereotypes.
Weighted by number of studies, the correlation comes out to under r = 0.05, which is equivalent to an effect size, d, of under 0.1. This is not strong evidence in support of claim #1.
But what about claim #2?
Before discussing this we need to do some conceptual footwork. What is the implicit mental model that the authors are working from? It is this:
(a) group stereotypes have a significant individual level effect.
(b) individuals in a negatively stereotyped group will tend to be be negatively impacted relative to individuals in a positively stereotypes group.
(c), (a) and (b) will lead to mean group differences.
The first point to note is that, in this scenario, the magnitude of the group effect will depend on the magnitude of the individual level effect. Were the correlation between stereotypes and judgements of individuals zero, the effect on a group, by way of individual effects, would be zero. While the correlation between stereotypes and judgements of individuals, based on hundreds of studies, is not zero, it is very close to that. This puts a low ceiling on the possible group level impact of stereotypes.
The second and more important point is that the conclusion (c) does not follow. A group stereotype which has significant individual level effects will not necessarily cause mean group differences in excess of what exists. Specifically, the magnitude of the impact of a stereotype will be a function of (a), as we said, the effect on the individual level, (b) the accuracy of the stereotype, (c) the degree to which people take into account individual information, and (d) the degree to which actions taken on the basis of stereotypical information is commesurate with actions taken on the basis of non stereotypical information. Imagine the following scenario:
There’s a widely known perfectly accurate stereotype in country X that Group A is one standard deviation less Zachslebachlsegooglebop (z) than group B. The decisions of individual’s in country X are informed by this stereotype in proportion to their lack of knowledge of the specifics of individual cases. Actions taken on the basis of this stereotype do not differ from those taken on the basis of knowledge of individual differences within groups (i.e., a random individual from group B treats a randoms individual from group A like an individual from group B who is known to be one standard deviation less Z than the average of group B.)
In this scenario, the cumulative action of all individuals in country X, will not induce more difference, associated with the Zity, than would be induced by the Z differences themselves. And the cumulative action will not result in more Z differences than already exist. Individuals are treated according to their estimated level of Zity conditioned on group identity only insofar as no other information is know. Since the groups, in this scenario, do differ in Zity, insofar as individuals are treated conditioned on group, they are treated in proportion to the differences that are — and so no worse than they would be, on average, were perfectly accurate knowledge had.
(Of course, anti-stereotypists typically oppose such knowledge because they oppose the treating groups in accordance with actual differences (when in fact they oppose this). More generally, anti-stereotypists see group differences as a sort of societal bias — as something artificial viewed from the perspective of their ideal. Since accurate stereotypes index these “artificial” yet real differences, these stereotypes are viewed also as “biased” or false –regardless of empirical accuracy. Anti-stereotypists, of course, don’t make their position clear. Like others of their sort they are intellectually dishonest and perhaps evil.)