Application of generalizability theory in estimation of variance components in National Examination Council essay questions in English Language

Abstract

Application of generalizability theory in estimation of variance components in National Examination Council Questions in English language. Design is two facet fully crossed G – study and D – study. The population 723 senior secondary schools in Ikwerre Local Government Area. Simple random sampling technique via balloting to draw six schools out of fourteen senior secondary schools. Cluster sampling technique to sample 153 students. Instrument is 2018 November/ December National Examination Council English Language Essay Questions. Data obtained analysed using computer software SPSS through General Linear Model via variance components MINQUE method. Result the largest contribution to error is person by item by rater (∂2pir) (5.259) with percentage variance of 47: 221 and person by item (∂2pi) (4.253) percentage variance of 38.17; person by rater (∂2pr) (0.415) percentage variance of 3.72, person (∂2p) (0.398) percentage variance 3.57%,rater (∂2r) (0.80) percentage variance 7.18, item (∂2i) (0.024) percentage variance 0.21, item by rater (∂2ir) (1-0.008*) percentage variance -0.071. The universe score 5.5368, Relative error variance 5.1146, Absolute Error Variance 0.4222, G- study coefficient 0.5198 and Index of dependability 0.929 obtained. Recommendation, generalizability theory should be used in psychometric properties estimate to generate reliable test items.