ESTIMATION OF QUANTILES FOR THE SAMPLING DISTRIBUTION OF THE (R)WG WITHIN-GROUP AGREEMENT INDEX

The James, Demaree, and Wolf within-group agreement index (r(WG)) may be used to assess interrater agreement when two or more individuals (''judges'') provide ratings on one or more items (''targets''). Although the James et al. index is increasingly being used to assess whether the aggregation of data is justified for higher levels of analysis, the sampling distribution of r(WG) has not been delineated, and therefore it has not been possible to conduct tests of the statistical significance of obtained results. As a consequence of this current deficiency in knowledge about r(WG), a Monte Carlo simulation program (AGREESIM) was developed to estimate selected quantiles for the sampling distribution of r(WG) under any user-specified null distribution of judges' scores. The output is in the form of quantiles and their 95% confidence intervals (showing the precision of the quantile estimates).