There are several sensitivity analysis procedures which have been considered by Tanaka and Okada (1989 a, b, c) to investigate the phenomena of how a small change of data affects the outcome of factor analysis. Among them are those for principal factor analysis (PFA), maximum likelihood factor analysis (MLFA) and least square factor analysis (LSFA). This motivates us to show that a similar technique can also be utilized to develop the sensitivity analysis in alpha factor analysis (AFA). Some examples are explained to illustrate the present procedure and a comparison is made in particular with the cases of PFA and MLFA.
This paper critically revicws two important concepts which should be considered even in the analysis of clinical trials. These are interaction and confounding. Both must often be a great burden for the interpretation. It may be more difficult to detect the existence of confounding than interaction. It is noticed that testing for the imbalance of baselinc factors is a part of the check for confounding. It illustrates a realistic situation in which both confounding and interaction are involved. Tests for interaction are shown as the basis of odds ratio, risk ratio and risk difference, with an empirical comparison. Finally, testing for qualitative interactions is discussed for a possiblc application.
To estimate the median effective dose (ED50), we models the dose-response curve by using generalized linear models and generalized additive models as their extension. The estimation of ED50 in generalized linear models has been known but not so for generalized additive models. We compare the result from nonparametric fitting of a generalized additive model with those of generalized linear models.
A Monte Carlo simulation study was performed, on the bases of a simple branching process model concerning proliferation of cancer cells, to investigate the relationship between the length of the tumor growth period and the characteristics of proliferation of cancer cells. This simulation suggested that even a slight change of the proliferation condition may cause a great difference in the length of the tumor growth and the morbidity of cancer as well.