Discounting Their Own Success: A Case for the Role of Implicit Stereotypic Attribution Bias in Women's STEM Outcomes

Click to increase image sizeClick to decrease image size

Acknowledgments

The stereotypic attribution bias research described in this article was supported by the National Science Foundation Gender in Science and Engineering Program (award no. HRD-0936434) awarded to the author.

Notes

Stereotypic attribution bias is assessed in terms of four subscores, representing “success internal difference” (the number of internal explanations given for male success behaviors subtracted from the number internal explanations given for female success behaviors), “success external difference” (the number of external explanations given for male success behaviors subtracted from the number of external explanations given for female success behaviors), “failure internal difference” (the number of internal explanations given for male failure behaviors subtracted from the number of internal explanations given for female failure behaviors), and “failure external difference” (the number of external explanations given for male failure behaviors subtracted from the number of external explanations given for female failure behaviors). The stereotypic attribution bias pattern is evinced in negative success internal difference scores, positive success external difference scores, positive failure internal difference scores, and negative failure external difference scores. Because stereotypic attribution bias was theorized to be evinced in this specific pattern of scores, “L” scores were computed from contrast weights as recommended by Furr and Rosenthal (2003), that is, by multiplying these difference score means by contrast weights assigned according to predicted patterns of attribution (–1, 1, 1, –1). A positive L score indicated the degree to which participants showed the stereotypic attribution bias pattern. Negative L scores indicated that the participant showed the opposite of the stereotypic attribution bias pattern. Analysis is conducted on L scores, which reflect the degree to which respondents showed the predicted pattern of stereotypic attribution bias.