Correlation Analysis Containing Discrete Variables Used in Certain Meteorological Problems

Abstract

In the meteorological field, correlation analysis has been performed for cases containing several continuous variables only. On the other hand, AITKEN, A. C. and GONIN, H. T. (1935) previously studied the moment generating function in the fourfold sampling procedure. Moreover TATE, R. F. (1954), OLKIN, I. and TATE, R. F. (1960) recently gained the several interesting results concerning the correlation analysis approach in cases where both continuous and discrete variables are simultaneously taken into consideration. Therefore in this paper, the author firsly derived several supplementary results from the correlational expressions gained by OLKIN and TATE and drafted a diagram to serve statistical large sample test of sample correlation coefficient, and secondly contrived to draft statistical diagrams of large sample test to correlation coefficient through formal expressions of bi-variate binomial distribution in reference to the past studies mentioned above. Lastly, some possible application examples are shown once at least in the meteorological field, to explain the author's contributions to such statistical fields. © 1966, Japan Meteorological Agency. All rights reserved.