My co-author Dennis Simon and I studied the success of female candidates running for U.S. Congress for 10 years. Some miscoded data, however, suggested that we had overlooked a major outlet for women’s political participation: third parties. Traditionally, research on American elections excludes third-party candidates; when election data are coded, only the results for the two major parties are counted. But as we went further back in time, we kept finding more and more women who had run as third-party candidates and decided this phenomenon was worth investigating. However, we had no idea if there was any kind of relationship between the frequency of women running under third-party labels and women running as Democrats and Republicans: Did fluctuations in one precede the other? Or was there no relationship at all? Because we had time-series data—and competing theories about the possible relationships between them—we used a vector autoregression model and Granger causality tests to explore all the possible ways in which the proportions of women running as major-party candidates and third-party candidates might affect each other.