Erratum: Roberts & Verhoef (2016) (Journal of Language Evolution DOI: 10.1093/jole/lzz003-dummy)

In Roberts and Verhoef (2016), the paired change in ranking analysis (the final paragraph of the Results section, corresponding to Figure 3) contained a statistical error. This measure focused on authors who submitted abstracts to both the single-blind EvoLang 10 and the double-blind EvoLang 11, and examined how much their ranking improved from the first to the second. (e.g. if an author had the 10th highest ranked paper in EvoLang 10 and the third highest ranked paper in EvoLang 11, they would have a positive 'paired change in ranking', since they went from getting a lower-ranked evaluation to a higher-ranked evaluation.) In this measure, there was a crossover interaction with no main effect: i.e., for nonstudents, female authors' rankings improved more than male authors', whereas for students it was male authors who improved more than female authors. However, in the original article, a main effect was erroneously reported. This occurred because the data were analyzed using regression with dummycoding; the coefficient for 'Gender' was misinterpreted as representing a main effect of gender, but in dummycoded regression this coefficient actually only represented the simple effect of gender at the baseline level of student status (i.e. nonstudents).