This component describes our attempts to answer a question raised by the main research. Specifically, the alternative account that we proposed in the main manuscript was that the IAT's advantage over the explicit stereotypes in predicting important outcomes was due to the IAT's shared variance with personal attraction to science. From that hypothesis, it follows that controlling for shared variance with personal attraction would significantly reduce the advantage of the IAT over explicit stereotypes in predicting relevant outcomes. However, we did not find an established method to test whether that reduction is significant. In the pdf document, we describe the problem and our attempts to solve it. We also describe a few alternative analysis strategies for the comparison between the IAT and explicit stereotypes in predicting the outcome measures, before and after controlling for personal attraction. This component also includes our R script, and data files that include only the relevant variables (see other components for the raw data).