Impact of Scratch on the achievements of first-year computer science students in programming in some Nigerian polytechnics

Abstract

To support the advancement of modern civilisation, our institutions of higher learning must produce the right pool of professionals, who can develop innovative software. However, the teaching and learning of the first programming language (CS1) remains a great challenge for most educators and novice computer students. Indicators such as failure and attrition rates, and CS1 student engagement, continue to show that conventional pedagogy does not adequately meet the needs of some beginning CS students. For its ease in introducing novices to programming, Scratch—a visual programming environment following the constructionism philosophy of Seymour Papert—is now employed even in some higher education CS1 classes with mixed evidence of its impact. Scratch captures the constructionist agenda by its slogan: “Imagine, Program, Share.” Therefore, this study explored the impart of using a constructionist Scratch programming pedagogy on higher education CS1 students’ achievements. This study also sought to compare the impacts of the two CS1 modes: the conventional class - involving textual programming language, lectures and labs, and the constructionist Scratch inquiry-based programming class. It further aims to discover if gender, academic level, age, prior programming, and visual artistic abilities moderate the effects of programming pedagogy on students’ achievements. To realize the study’s aims, the study employed a quasi-experimental pretest-posttest nonequivalent groups design, involving four intact CS1 classes of polytechnic students (N = 418) in north-central Nigeria. The investigation was conducted in phases: a pilot (n = 236) and main (n=182) studies lasting two academic sessions, with each study comprising one experimental and one control group. In each session, learning in both modes lasted for six weeks. In both studies, purposive sampling was employed to select institutions, and selected institutions were randomly assigned to treatment groups. Instruments employed included CS1 Student Profile Questionnaire (CSPROQ) and Introductory Programming Achievement Test (IPAT). To strengthen the research design, I employed Coarsened Exact Matching (CEM) algorithm—after conducting a priori power analysis—to generate matched random samples of cases from both studies. Thus, research data employed in the analysis include: from the pilot, 41 cases in each treatment group; from the main study, 42 cases in each treatment group. Descriptive and inferential statistics were employed to find answers to research questions and test the research hypothesis. Data from both studies satisfied the requirements for statistical tests employed, i.e., t-test and ANCOVA. The alpha level used in testing hypotheses was p = 0.05. The dependent variable is the IPAT post-test score, while the independent variables are treatment, gender, age, academic achievement level, prior programming, and prior visual art. The covariate was the IPAT pretest score. Statistical analyses were conducted using SPSS version 23. The t-test results from both pilot and main studies indicated that, both programming pedagogies had significant effects on student IPAT scores, although the effect of the constructionist Scratch intervention was higher. Results from the one-way ANCOVA analysis of both pilot and main study data—while controlling for students’ IPAT pretest scores—yielded the same outcome: There was significant main effect of treatment on students’ IPAT posttest scores, although the impact was moderate. Controlling for pre test scores, analysis of the main studies data yielded no significant main effects of: gender, age, academic level, prior programming and prior visual artistic ability. The result from the main study also reveals no interaction effect of treatment, gender, academic level, age, prior programming, and prior artistic ability. While the quality of CS1 students’ performance in each session varies as their IPAT achievements show, yet the results of this research revealed a consistent pattern: Students in the constructionist Scratch class outperformed those in the conventional class, although the impart was moderate. This finding implies college students without prior programming experience can perform better in a class following a constructionist Scratch programming pedagogy. The study recommends the use of Scratch, following a constructionist pedagogy with first-year students in colleges, especially those without prior background in programming