Treatment of the Fourfold Table by Partial Association and Partial Correlation as it Relates to Public Health Problems

Publisher Website
Google Scholar
Abstract
The treatment is considered of plural variables when the data fall into dichotomous classifications so that the analysis depends on several 4-fold tables. Two methods are described. The first consists, for the 4-fold table of each subuniverse, in determining the difference between proportions in terms of the standard error of the difference, averaging these unit deviates, and determining significance of the average, using 1/[image]N as the standard error of the mean, where N equals the number of subuniverses. In cases where the observations for the 4-fold tables were small, chi-squares were computed using Yates'' correction, the square roots extracted, and the sign adopted as if the unit deviate had been computed. The 2d method consisted of computations of product moment correlations and partial correlation coeffs. After the partials were computed they were transformed into Fisher''s z and the results reduced to unit deviates by dividing each z by its standard error. On the whole, agreement between results of the 2 methods seemed satisfactory.

This publication has 0 references indexed in Scilit: