Resolving Contradictions of Predictive Validity of University Matriculation Examinations in Nigeria: A Meta - Analysis Approach

Abstract

The study examined the various means of solving contradictions of predictive studies of University Matriculation Examination  in Nigeria. The study used a sample size of 35 studies on predictive validity of University Matriculation Examination in Nigeria, which was purposively selected to have met the criteria for meta-analysis. Two null hypotheses were generated and tested at 0.05 level of significance. The data collected were tested using t-test statistics and fisher's Z-transformation at 0.05 level of significance. The findings of this study revealed that studies with sample size of 120 and above is significantly different from studies with sample size less than 120. Published journal articles are more significant than their unpublished studies like master's thesis and Ph.D. dissertations. Based on the findings of this study, it was recommended that researchers should not use sample size less than 120, the larger the sample size, the better the result for predictions. It was also recommended that unpublished studies should undergo editorial review to reduce some of the flaws in unpublished studies.

Keywords: Effect size, predictive validity, meta-analysis, sample size.