Reply to Healy et al.: Value of ex ante predictions and independent tests for assessing false-positive results

• Type of Material: Article
• Authors: Anthony Fowler, University of Chicago; Brendan Pablo Montagnes, Emory University
• Language: English
• Date: 2015-12-01
• Publisher: National Academy of Sciences
• Publication Version: Final Published Version
• Copyright Statement: 2015 PNAS
• Final Published Version (URL): http://dx.doi.org/10.1073/pnas.1520253112
• Title of Journal or Parent Work: Proceedings of the National Academy of Sciences of the United States of America
• Volume: 112
• Issue: 48
• Start Page: E6592
• End Page: E6592
• Keywords: Humans; Politics; Emotions; Decision Making; Government
• Research Categories: Political Science, General; Sociology, Public and Social Welfare

Abstract

In our study, we conclude that a published result that college football games influence elections is most likely a false-positive. In a letter to the editor, Healy et al., the authors of the original study, raise several objections. Healy et al.’s letter suggests that they have partly misread our paper as a criticism of their research design. We grant that their research design and identifying assumptions are sound but worry that their result is a chance false-positive, which can arise even with a perfect research design. We test several additional, independent hypotheses that should hold if college football games do indeed influence elections. We find no support for any of these independent hypotheses, leading us to conclude that the original result is most likely a false-positive.