Validation of Basic Science Achievement Test Using R Software

Abstract

The study is aimed at validation of Basic Science Achievement test using innovative techniques (R software). The study employed an instrumentation and descriptive survey research design.  The Basic Achievement Test (BAT) instrument was made up of four components, which was used for generation of 98 items. Survey method was used to collect data from 567 Junior secondary school three (JSS3) students for trial testing. Simple random sampling was used to select public secondary schools in Port Harcourt metropolis. Four research questions guided the study. The study data was pretested and was found to meet the assumption of normality and no outlier before item analyses was done. Item analyses on the 98-item multiple choice test were done based on CTT technique alone since IRT assumption of unidimensionality was violated. Items from CTT analysis with rpbs≥0.20 and 0.30≤p≤0.80 were considered good items and selected. The R software was employed for all analyses carried out in the study while DIMPACK 1.0 was used to confirm test of unidimensionality. The reliability of the test was established through Cronbach Alpha, Split-half and Kuder Richardson 20 statistics using R, which produced a reliability coefficient of 0.87 for both Cronbach alpha and KR20 and 0.89 for Slit-half reliability. The result revealed that 55 items were considered good while, 43 items were marked for elimination. The findings show that BAT is both valid and reliable and thus, is recommended for measuring students' proficiency in Basic Science in public secondary schools in Port Harcourt metropolis.