[Examples of pitfalls in statistical analysis--5: Analysis of a multiple factor model].

When we analyze the effects of multiple factors, it is inadequate to repeat simple comparisons of the groups divided by a factor. The reasons for this are twofold, i.e., the total error rate of the study grows higher in proportion to the number of comparisons, and we cannot evaluate the relationships between the factors. It is reasonable in such a situation to use multivariate analyses which are developed to analyze multiple factor models, by changing the analytic point of view. In this article, I present an example to illustrate several types of mistakes which occur when we simply compare the two groups divided by one factor for other multiple factors, and explain the evaluation with correlation analysis and factor analysis.