This paper is part of a series providing introductory explanations and demonstrations of various social science data analysis techniques. One-way analysis of variance using a fixed model is used to test the null hypothesis that national differences have no effect on subjective competence. Two-way analysis of variance enables the investigator to separate the effects of two nominal predictors, to test for the significance of each of their effects, and to estimate the proportion of overall variation due to each predictor's effect. Under experimental conditions, with uncorrelated predictors, this technique handles the problem easily. In most survey situations, however, predictors are correlated and it is difficult to separate each predictor's effects. Special problems include having one observation per cell, pooling nonsignificant sums of squares, and considering interaction among variables as nonparallel lines. In the present example, social and economic development had a stronger effect on subjective competence than did cultural differences. Tables, notes, and 19 references are included. For other volumes in this series, see NCJ 72010-14.