The issue of error in man-machine communication is already reported, based on experimental data. The experiment on a personal computer was conducted for the examination of human catastrophic behaviors. After the previous report, the analyses of variance were carried out for those experimental outcomes in order to clarify the factors which have some effect on the phenomenon concerned. The purpose here is to describe those analyses in detail. Seven factors are examined, and the results of the analyses show three of them affect the response of a subject.