Use of Contingency Tables to Investigate the Association of Gender, Type of Primary School, KCPE Results and KCSE Results in Mathematics

Abstract

In this study, the log-linear model with four categorical variables is used to investigate the association between the results of two stages in the Kenya Education system. The source of data was the 2010 KCPE mathematics results and the corresponding 2014 KCSE results in mathematics. The study sample consisted of 1161 candidates, 603 males and 558 females who graduated within the period under study (2010-2014). The instrument used for data collection was the records of candidates' results that made up the sample. In particular, factors such as gender of candidate, type of primary school attended, KCPE Grade and KCSE Grade in mathematics were examined for presence or absence of association. The first two variables had each two categories and the last two variables had each three categories. A four dimensional 2í—2í—3í—3 contingency table was considered. The null hypothesis of mutual independence among these variables was subjected to test. Data was analyzed using the R version 3.2.1 software. KCSE mathematics grade was found to be associated with the KCPE grade but not associated with the type of primary school or gender.