The Effect of Categorizing a Continuous Independent Variable on Explained Variance

Abstract

Researchers, Particularly those in behavioral sciences, often categorize a continuous independent variable, in their research work. This is done for the main purpose of applying analysis of variance in testing their hypotheses. Such categorization is accompanied by loss of information which is reflected in a change in proportion of the variance in the dependent variable, accounted for, by the independent variable. To provide an empirical backing to this claim, a validated “teachers variables and the attainment of the goals of the Universal basic education scheme”, was administered on a random sample of 640 secondary school teachers in Ogoja Education zone of Cross River State,Nigeria. The resulting data were analyzed using simple and multiple linear regressions. The independent variables were then categorized using sample mean and standard deviation. Analysis of variance was then applied. The proportion of the variance accounted for by the independent variables, individually and collectively were then compared using Fishers Z-transformation test. The results show that differences in explained variances do exist in favour of linear regression analysis, though not significant. The implication of the findings in research design is discussed.