An Invitation to Multivariate Analysis: An Example About the Effect of Educational Attainment on Migration Propensities in Japan

To provide a strong motivation for students to learn multivariate statistics and the multivariate way of thinking, this paper uses an easily understandable example of ascertaining the effect of educational attainment on migration propensity in Japan. With cross-tabulations and a logistic model, we demonstrate the necessity of the multivariate approach by showing that the control for the effect of gender is indispensable for revealing the true effect of educational attainment. We further identify two conditions under which valid inference about the effect of a factor depends critically on the control for another factor. Finally, we identify an apparent contradiction between tabulated and logistic results and present a resolution to it.