The A() function calculates the A statistic, a nonparametric measure of effect size for two independent groups that’s also known as the probability of superiority (Ruscio, 2008), along with its standard error and a confidence interval constructed using bootstrap methods (Ruscio & Mullen, 2012). Optional arguments can be specified to calculate variants of the A statistic developed for other research designs (e.g., related samples, more than two independent groups or related samples; Ruscio & Gera, 2013). <DOI: 10.1037/1082-989X.13.1.19>. <DOI: 10.1080/00273171.2012.658329>. <DOI: 10.1080/00273171.2012.738184>.
Label | Latest Version |
---|---|
main | 3.0 |