| Abstract: | Methods appropriate to the analysis of longitudinal-panel data are not necessarily appropriate to the analysis of developmental data, because of the sequential order that developmental data are assumed to have. This paper compares, using data from a study of social categorization ability, the usual cross-sectional test for cumulative order (Guttman scaling) with a longitudinal test for sequential order using matrices. This matrix-based test is then compared to the analysis of differences in proportions. It is shown that the matrix-based test is more appropriate to the analysis of developmental data than either longitudinal difference-of-proportion analysis or the Guttman scale. |
