Publication Type: Conference Paper/Unpublished Manuscript
Review Method: Peer Reviewed
Abstract: Although explicit racism has nearly vanished, and explicit sexism has reduced, many scholars worry that voters may still exhibit implicit sexist or racist voting behaviors. In other words, voters decide "as if" they were sexist or racist even though they may genuinely believe themselves to be unbiased. Yet, scholars differ on how to measure these implicit attitudes and behaviors. In September 2008, we conducted an online experiment to examine the relationship between (1) sexist voting decisions and (2) prominent measures of implicit like the Implicit Attitude Test (IAT), Affect Misattribution Procedure (AMP) and symbolic sexism survey questions. This paper makes three key contributions. First, we propose a precise definition of biased behavior and apply this definition to sexist voting decisions. Second, we test the predictive validity of prominent measures of implicit sexist attitudes. Third, we test two priming effects: whether measuring implicit attitude reduces sexist voting or not and conversely, whether measuring sexist voting reduces makes any difference. Implications for research design and analysis of implicit attitudes and behaviors are discussed, including ANES's decision to include the AMP in its 2008 study.