Forthcoming Publications

Semi-supervised stereotyping: Untested stereotypic assumptions reinforce stereotypes.
William T. L. Cox and Patricia G. Devine
In Preparation

Stereotypes provide social perceivers with expectations about others, and those expectations are rarely put to the test (e.g., someone may infer that a Black man is dangerous and avoid him, thus never figuring out whether he was dangerous). Because perceivers most often assume their untested assumptions are correct, their belief in that stereotype, and their reliance upon it, increases. Across 3 experiments, we show that untested stereotypic assumptions reinforce stereotypes as much as confirmed stereotypic assumptions, leading to higher rates of stereotyping over time.

Poster reporting this work, SPSP 2015.

Breaking free of Lippmann’s Ghost: Modern advances in theories of stereotypes and stereotyping.
William T. L. Cox and Patricia G. Devine
In Preparation

Researchers are very often imprecise about their model of stereotyping or even fail to specify the model they use, assuming that “everyone knows” what a stereotype is. Many stereotyping and prejudice researchers seem to tacitly adopt a vague notion of stereotypes that resembles Lippmann’s original model, despite theoretical and empirical advances that favor other models. We argue that, in this way, the ghost of Lippmann’s model retains an unwarranted, often unnoticed influence, leaving the literature with unclear notions about what a stereotype is. Conceptual fuzziness about the nature of stereotypes and stereotyping impedes progress on many core issues of interest, including, for example, how stereotypes are activated, how they function, how they can be changed, and how they relate to prejudice and discrimination. We identify a number of implications, predictions, and new questions that are revealed by breaking free from Lippmann’s ghost and embracing modern models that offer greater theoretical clarity and precision.

Multiple determinants of prejudicial vs. nonprejudicial behavior
William T. L. Cox and Patricia G. Devine
In Preparation

Recent debates in the prejudice literature have centered around whether implicit or explicit measures are the “better” predictors of prejudicial behavior. I argue, however, that this debate is misplaced. I modeled consequential real-world behaviors in the lab across 5 studies, assessing the extent to which several key measures independently or jointly predict prejudicial versus nonprejudicial behavior. Interactions between explicit measures of internal and external motivation to respond without prejudice (Plant & Devine 1998) predicted race and gender biases behavior well across studies, opposing the oft-overgeneralized conclusion that “explicit measures” predict behavior poorly. Implicit bias, as measured by the Implicit Association Test (IAT), predicted some outcomes, but it accounted for less variance than the explicit motivation measures. As such, the present work provides evidence that tempers claims made by IAT opponents and IAT advocates alike. I argue for abandoning past paradigms that focus narrowly on one construct, measure, or type of measure as the primary determinant of intergroup bias. I recommend broader, theory-driven, integrative approaches that seek to understand the multiple intrapersonal, interpersonal, and situational factors that independently and jointly result in prejudicial or nonprejudicial behavior.

Gender- and race- stereotyped power roles in gay and straight Internet pornography.
William T. L. Cox, Carlie M. Allison, Janet S. Hyde, and Patricia G. Devine
In preparation

This article reports a media content analysis that tests an intergroup dynamics account of aggression and unsafe sexual behaviors in pornography. We hypothesized that the more social group differences there were between the pornography actors (e.g., gender, race, age), the more aggression and unsafe sexual behaviors should arise. We coded 200 pornography videos that varied in the extent to which the actors crossed group boundaries (e.g., in gay, same-race videos the actors cross no group boundaries, in straight, interracial videos, the actors cross two group boundaries). Hypothesis-blind coders coded the videos for unsafe sexual activities and a variety of negative and positive sex acts, language, and behavior. Results largely supported our hypotheses, such that videos that crossed an intergroup barrier (race, gender, or both) had more aggression and unsafe behaviors. Our rich, nuanced dataset is publicly available and can be used to test many other questions of interest.

Poster presenting this work, from SPSP 2013