Slides at osf io3 unva A metaanalysis of

  • Slides: 27
Download presentation
Slides at osf. io/3 unva/ A meta-analysis of change in implicit bias and behavior

Slides at osf. io/3 unva/ A meta-analysis of change in implicit bias and behavior Calvin Lai, Ph. D. Assistant Professor Dept. of Psychological & Brain Sciences Washington University in St. Louis February 7, 2019 calvinlai@wustl. edu

Implicit Bias Predicts Behavior Vote choice (e. g. , Arcuri et al. , 2008;

Implicit Bias Predicts Behavior Vote choice (e. g. , Arcuri et al. , 2008; Galdi et al. , 2008) Job performance of minority employees (Glover et al. , 2017) Hiring discrimination (e. g. , Reuben et al. , 2014; Rooth, 2010) …and many other behaviors (e. g. , Alesina et al. , 2018; Green et al. , 2007; Greenwald et al. , 2009; Hofmann et al. , 2007; Kurdi et al. , 2013 Oswald et al. , 2013; Stout et al. , 2011)

Idea: Change implicit bias, prevent unwanted behavior!

Idea: Change implicit bias, prevent unwanted behavior!

A Brief History of Implicit Bias Change Stable & Rigid 1985 Bargh, 1999 Devine,

A Brief History of Implicit Bias Change Stable & Rigid 1985 Bargh, 1999 Devine, 1989 Wilson et al. , 2000 Malleable & Flexible 2001 Dasgupta & Greenwald, 2001 Blair, 2002 2018 Reviews Dasgupta, 2009 Gawronski & Sritharan, 2010 Lai, Hoffman, & Nosek, 2013 Lai & Banaji, 2019

A Meta-Analysis of Procedures to Change Implicit Measures Forscher*, Lai*, Axt, Ebersole, Herman, Devine,

A Meta-Analysis of Procedures to Change Implicit Measures Forscher*, Lai*, Axt, Ebersole, Herman, Devine, & Nosek, under revision * Co-lead-authors

Study Inclusion Criteria 1. Be an experiment 2. Include manipulation of implicit bias 3.

Study Inclusion Criteria 1. Be an experiment 2. Include manipulation of implicit bias 3. Measure a pre-existing implicit bias

Database 4, 908 articles Eligible 417 articles 593 studies Final 342 articles 492 studies

Database 4, 908 articles Eligible 417 articles 593 studies Final 342 articles 492 studies 571 independent samples 87, 418 participants

Change in implicit measure vs. a hypothetical control

Change in implicit measure vs. a hypothetical control

Change in implicit measure vs. a hypothetical control

Change in implicit measure vs. a hypothetical control

Change in implicit measure vs. a hypothetical control

Change in implicit measure vs. a hypothetical control

Takeaway: Implicit biases can be changed! Change in implicit measure vs. a hypothetical control

Takeaway: Implicit biases can be changed! Change in implicit measure vs. a hypothetical control

Does implicit bias play an explanatory role in changing explicit biases and behavior? Implicit

Does implicit bias play an explanatory role in changing explicit biases and behavior? Implicit Bias Manip. Explicit Bias / Behavior

No. Mediation Effects on Explicit Biases Mediation Effects on Behavioral Outcomes

No. Mediation Effects on Explicit Biases Mediation Effects on Behavioral Outcomes

What about potential moderators? • There was so little variation in the indirect effects,

What about potential moderators? • There was so little variation in the indirect effects, we originally had to fix the variation to zero to get the models to converge. • However, we tested the following moderators: Moderator Absolute vs. Relative Measurement Spontaneity Learning vs. Context Effects Self-Presentation Motives Does it matter?

Why doesn’t implicit bias change predict behavioral change? 1. Associations are causally inert 2.

Why doesn’t implicit bias change predict behavioral change? 1. Associations are causally inert 2. Measurement issues 3. Measurement correspondence 4. Implicit associations are a group-level construct

Associations are Causally Inert

Associations are Causally Inert

Associations are Causally Inert WEAKNESS: No parsimonious alternative account for behavior that implies automatic

Associations are Causally Inert WEAKNESS: No parsimonious alternative account for behavior that implies automatic evaluation Endowment Effect: Owning something makes you like it more Mere Exposure: Seeing something makes you like it more C>P IAT Effect: We’re faster to categorize concepts that are more closely linked together in memory Name-Letter Effect: Liking things that share our first initial more Halo Effect: Assuming beautiful people are awesome in many other ways

Measurement Issues: What are we actually changing? - Associations - Ingroup Favoritism (White +

Measurement Issues: What are we actually changing? - Associations - Ingroup Favoritism (White + Good/Bad) - Outgroup Hatred (Black + Good/Bad) - Overriding bias - Ability to detect a correct response - Guessing

9/18 Interventions Reduced Implicit Racial Preferences on the IAT D Summary Score Lai et

9/18 Interventions Reduced Implicit Racial Preferences on the IAT D Summary Score Lai et al. , 2014

6/18 Reduced Ingroup Favoritism 1/18 Reduced Outgroup Hatred Data from Lai et al. ,

6/18 Reduced Ingroup Favoritism 1/18 Reduced Outgroup Hatred Data from Lai et al. , 2014, 2016 Calanchini, Sherman, & Klauer, under review

Measurement Issues: Noisy behavior Future research should: - Assess more behavior - Assess behavior

Measurement Issues: Noisy behavior Future research should: - Assess more behavior - Assess behavior over time and across situations

Measurement Correspondence Global attitude “How much do you like Christianity? ” Behavior: Going to

Measurement Correspondence Global attitude “How much do you like Christianity? ” Behavior: Going to church next Sunday Attitude toward behavior “How much do you like attending church worship service? ” Attitude toward behavior in a given time, target, & context “How much do you like attending your church's worship service next Sunday?

Measurement Correspondence Global attitude • Interracial interactions • Vote choice • Willingness to date

Measurement Correspondence Global attitude • Interracial interactions • Vote choice • Willingness to date someone of another race • Hiring discrimination • . . and many more

Implicit Associations are a Group-Level Construct Payne et al. , 2017

Implicit Associations are a Group-Level Construct Payne et al. , 2017

Implicit Associations are a Group-Level Construct Payne et al. , 2017

Implicit Associations are a Group-Level Construct Payne et al. , 2017

Why doesn’t implicit bias change predict behavioral change? Possibilities: 1. Implicit associations are causally

Why doesn’t implicit bias change predict behavioral change? Possibilities: 1. Implicit associations are causally inert 2. Measurement issues 3. Measurement correspondence 4. Implicit associations are a group-level construct

Acknowledgements Collaborators Patrick Forscher Trish Devine Jordan Axt Brian Nosek Charlie Ebersole Michelle Herman

Acknowledgements Collaborators Patrick Forscher Trish Devine Jordan Axt Brian Nosek Charlie Ebersole Michelle Herman Slides at osf. io/3 unva/ DDGE-1315231 Email calvinlai@wustl. edu Website calvinklai. wordpress. com