MULTILEVEL ANALYSIS OF FACTORS PREDICTING SELF EFFICACY IN COMPUTER PROGRAMMING

This study used multilevel analysis to determine the predictive value of selected intrinsic factors (gender, computer ownership, mathematics background and computer experience) and institutional type (an extrinsic factor) on undergraduates’ Self Efficacy in Java Computer Programming (SEiJCP) in South – West, Nigeria. The study adopted a correlational design. Purposive Sampling was used to select 254 computer science undergraduates from four universities (three federal-owned and one state-owned) in south-west, Nigeria. Three research questions were answered. Two research instruments namely, Computer experience scale (r = 0.84) and Java Programming Self Efficacy Scale (JPSES, r = 0.96) were used to collect data. Data were analysed using descriptive statistics, and null and linear growth model (LGM) procedures. The intercorrelation coefficients among the extrinsic factor, intrinsic factors and SEiJCP were moderate. Null model shows that the variations in SEiJCP accounted for by insitutional level differences was 99.0%. The fixed part of the LGM of intrinsic factors showed that only mathematical backgroung contributed significantly (p < 0.05) to the prediction of SEiJCP. The random part of the LGM showed no significant contributions of the interactions of the intrinsic factors, to the prediction of SEiJCP. About 60.0% of the student level variation in SEiJCP is explained by the differences in intrinsic factors. The institution – level variable had large predictive value on programming self efficacy. Computer science departments should increase the number of mathematics courses in their curriculum.