Abstract
An aligned rank test for a nonparametric analysis of the two way interaction. Research problems that require nonparametric analysis of the interaction frequently arise in the behavioral sciences. There is, however, a lack of available procedures in commonly used statistical packages such as SAS or SPSS. The purpose of the present study is to review the fundamentals of the aligned rank test for two widely used two-way research designs in psychology. The procedure has shown good properties in nonnormal distributions in terms of Type I Error control and statistical power. It is easily conducted using common statistical packages. It is applied to two case studies which result in gender differences in linguistic abilities in blind children not revealed by other procedures. Limitations of the aligned rank test have been observed in situations of covariance heterogeneity across groups, and also with sample sizes smaller than ten participants per condition. Its adequate use after model diagnostics can, however, increase sensitivity to detect behavioral effects of theoretical or practical interest.
