Intrinsic and Extrinsic Factors as Predictors of Self Efficacy and Achievement in Programming Among Computer Science Undergraduates in South-Western Nigeria

Abstract

Java programming language is recent, dynamic and relevant in contemporary organisations. However, literature shows that learning computer programming using Java poses problems to many computer science undergraduates. Low self-efficacy in computer programming has been identified as one of the factors responsible for the observed problems which students encounter while learning programming. Little attention has been paid by researchers towards isolating factors that are likely to influence programming self-efficacy and achievement. This study, therefore examined the predictive value of intrinsic (gender, computer experience, mathematics background, computer ownership, locus of control, background in C++ and number of programming courses) and extrinsic (institution type) factors in undergraduates‟ self-efficacy and achievement in Java computer programming (SEiJCP and AiJCP) in south-western Nigeria. The study adopted a correlational design. Purposive sampling technique was used to select 254 computer science undergraduates from four universities (three federal and one state). Only universities offering Java and C++ programming languages in their curriculum participated in the study. Five research instruments namely: Computer Experience Scale (r = 0.84), Java Programming Self-Efficacy Scale (r = 0.96), Java Programming Achievement Test (r = 0.70), Levenson Locus of Control Scale (r = 0.88) and Computer Background Questionnaire with four subscales: C++ background (r=0.87), computer ownership (r=0.90), mathematics background (r=0.84) and computer experience (r=0.79) were used to collect data. Data were analysed using descriptive statistics, t-test and Multilevel Analysis Procedures (MLAP). For MLAP, Null and Linear Growth Models were examined. Null model showed that 91.0% of the variations in SEiJCP were due to institutional level differences. Students in the state university obtained higher scores in SEiJCP (𝑥 =173.97; SD =26.39) than their colleagues in the Federal (𝑥 =128.05; SD =44.57). The mean difference in SEiJCP scores between students in Federal and state universities was statistically significant (t = 7.57, df=252, p<0.05). The Null model also showed that 82.0% of the variations in AiJCP were due to institutional level differences. However, there was no significant difference in the mean score in AiJCP between students in the state university (𝑥 =22.92; SD =11.78) and their colleagues in Federal universities (𝑥 =20.54; SD =18.72). The Linear Growth Model (LGM) showed that only the number of programming courses significantly predicted SEiJCP (β=1.15), while gender, computer ownership, mathematics background, C++ background, computer experience and locus of control did not contribute significantly to the prediction. LGM showed that none of the intrinsic factors contributed to the prediction in AiJCP. The number of programming courses significantly predicted self-efficacy in Java computer programming while institutional type significantly predicted both self-efficacy and achievement in it among computer science undergraduates in south-western Nigeria. Computer science departments in Federal and state universities should increase the number of programming courses in their curriculum. The Federal universities should also organise tutorial classes in all programming courses in order to improve self-efficacy in them