Abstract
A covariance structure analysis approach to the study of parameter trends is outlined. By using the program RAMONA, we illustrate the method by fitting a corresponding confirmatory factor analysis model to correlational data from a study involving several psychometric tests and fluid intelligence tasks. Multiple-restriction tests and 1-tailed, single-restriction tests are carried out to ascertain the trend in complexity degrees. Confidence intervals of associated normed differences in noncentrality parameters are used as a complementary means of examining hypotheses of equal complexity, and statistical power of the tests is evaluated.