Predicting Implicit and Explicit Racial Bias from Past Interracial Experience University of Washington Department of Psychology Research Festival May 27, 2015 Michael P. McDonald Advised by Tony Greenwald
Measuring Implicit Racial Attitude African American OR Pleasant European American OR Unpleasant Pleasure Glorious Wonderful Laughter Joy Happy Peace Love Agony Hurt Awful Nasty Terrible Failure Evil Horrible
Measuring Implicit Racial Attitude African American OR Unpleasant European American OR Pleasant Agony Hurt Awful Nasty Terrible Failure Evil Horrible Pleasure Glorious Wonderful Laughter Joy Happy Peace Love
Scoring IAT Data Latencies to classify opposing pairs are used to compute a D score Negative indicates Black preference Positive indicates White preference Zero indicates neutral (no preference)
Project Implicit Publicly accessible website Variety of IATs (Age, Race, Gay/Straight) 30,000-50,000 Race IATs per month Self-selection Internet access, higher SES, more liberal Can provide basis for laboratory testable hypotheses
Explicit Attitudes: Likert Preference
Explicit Attitudes: Thermometer Ratings
Demographics
Individual Differences on Implicit & Explicit Measures
Whites show more implicit than explicit in-group (pro-White) bias . . . Mean Explicit = +.41 . . . Mean Implicit = +.83
Black+Mixed respondents show more explicit Black preference but almost no implicit preference _ _ Mean Explicit = -.56 _ _ Mean Implicit = +.11
Implicit attitude distributions overlap much more than explicit attitude distributions _ _ Mean (Black) . . . Mean (White) _ _ Mean (Black) . . . Mean (White)
Questionnaire View
Environment Items Appropriate descriptions of the neighborhood I lived longest in before age 10 are (select all that apply) racially diverse ethnically diverse minority Black majority Black minority White majority White Neighborhood I lived longest in from 10 to 18 Teachers I had in elementary and middle school Classmates in elementary and middle school Teachers I had in high school Classmates in high school Notice that these items lack qualification (strong, close, etc.)
Relationship Items I have a parent who is (select all that apply): Asian Black Hispanic Native American Pacific Islander White close family member of my own generation close family member younger than my generation strong childhood friendship strong post-childhood friendship romantic relationship highly admired preschool or elementary school teacher highly admired middle or high school teacher
Processing Multiple Response Data Items become binary responses “Parent” becomes Parent_Black, Parent_White, etc. Combined items were fashioned based on responses: Parent item = Parent_White – Parent_Black Repeated with all Relationship & Environment Items
Processing IAT Data Inclusion criteria: No more than 10% of responses < 300ms No more than 40% error rate Must be either US citizen or US resident
Findings and conclusions
Previous interracial experience predicts explicit attitudes better than implicit attitudes
Relationship items are more powerful predictors of attitudes, especially in Black+Mixed respondents
Familial relationships and friendships are more effective predictors, followed by school relationships
Next Steps Collect more data Select and retain most effective items Examine data from other race & ethnic groups Archive data on Open Science Framework & make publicly available
Acknowledgements With great appreciation for: Tony Greenwald Frank Kaiyuan Xu Teri Kerby Daniel Fredinburg † The Project Implicit Team