Skip to main content
Figure 8 | EPJ Data Science

Figure 8

From: Gendered behavior as a disadvantage in open source software development

Figure 8

Point estimates, with 95 percent CIs, for variables related to gender (variables are listed on the vertical axis). Panel (a) shows coefficients from count models of zero-inflated negative binomial models predicting success (the number of stars received), while panel (b) shows log odds ratios from logit models predicting survival over a one year period following our data collection. Labels of five specifications (identical for success and survival models) are shown in the legend. The first model enters gender variables and controls, the second enters controls and categorical gender behavior classes from the decision tree analysis, the third enters controls and 23 variables recording programming language use. The fourth is identical to the first, but with data with 5 percent gender swaps, and the fifth is with 10 percent gender swaps. For the fourth and fifth models confidence intervals show the 2.5–97.5 inter-quantile range from 100 simulated datasets

Back to article page