Restriction of range and correlation in outlier-prone distributions

Statistical theory indicates that restriction of the range of possible valu es of normally distributed variables, and many non-normal variables, reduce s correlations in unrestricted populations. Contrary to this typical outcom e, results of a simulation study show that range restriction sometimes incr eased the correlation between variables having outlier-prone distributions. This result occurred in the case of exponential and ex-Gaussian distributi ons, which are encountered in experimental studies involving response times . It did not occur in truncated versions of the same densities. Chance occu rrence of outliers in contaminated-normal, or mixed-normal, distributions r educed the correlation found between samples from uncontaminated population s. Conversely, detection and downweighting of outliers increased the magnit ude of sample correlations, and a similar result occurred far many other ou tlier-prone distributions. Practical implications of these findings are dis cussed.