“…Given larger data samples are susceptible to progressively smaller p-values, the robustness of this effect was tested on a smaller subsample using a sample size that will not reduce the p-value beyond what is consistent with the topic/discipline (Lin, Lucas, & Schmueli, 2013). Based on past work in social media activism (e.g., Foster et al, 2019;Kende et al, 2016;Wilkins et al, 2019), n = 300 was chosen so that the p-value would not be driven below p=.01. Consistent with the full sample data, the indirect effects of tweeting on BCA through gender identity alone, B = 0.01, SE = .03, %95 CI [−0.04, 0.08], and of tweeting on BCA through collective action intention alone, B = 0.05, SE = .09, % 95 CI [−0.14, 0.24], were not significant, but the serial mediation indirect effect was again significant, B = 0.07, SE = .03, 95% CI [0.02, 0.14].…”