The continuing inability of measures of personality to predict communicative or other types of behaviors has forced researchers to seek alternative ways of measuring. One alternative consists of a measure of four types of causal attribution in hypothetically crucial situations in five types of relationships of humans. The type of relationship in which communication occurs is assumed to indicate one particular causal perspective as most appropriate. An experiment was conducted to test the validity of the structure as measured and, of 12 predictions, 11 were statistically confirmed. A subsequent experiment was run to test the ability of the instrument to predict performance in four different types of communicative relationships from a knowledge of the causal perspective assumed to be the most appropriate for each. Of 11 predictors, 9 were statistically confirmed. (The evaluation instrument is appended.) (Author/MM)