Vibe Coding in Programming Education: Examining Programming Self-Efficacy and Prompt Engineering Competence
Description
This study employed an exploratory convergent mixed-methods design (Creswell & Plano Clark, 2017) to examine the effects of Vibe Coding activities on students’ programming self-efficacy and prompt engineering competence. The quantitative component followed a one-group pretest–posttest design, in which participants completed measures of programming self-efficacy and prompt engineering competence before and after the intervention. To complement self-reported data, students’ final software projects and AI interaction logs were objectively evaluated using a researcher-developed Prompt Engineering Performance Rubric (PEPR) assessing code quality and architecture, UI/UX design, application performance, and prompt quality. The qualitative component aimed to explore students’ experiences with AI-assisted programming through reflective journals, AI interaction logs, and software development artifacts. These data were analyzed to identify patterns related to prompt refinement, problem-solving strategies, and human–AI collaboration. Integrating quantitative and qualitative findings enabled methodological triangulation and provided a more comprehensive understanding of the educational impact of Vibe Coding. Three instruments were employed for data collection. Programming self-efficacy was measured using the Computer Programming Self-Efficacy Scale (CPSES) (Tsai et al., 2019), while prompt engineering competence was assessed using the Prompt Engineering Competence Scale (PECS) (Gibreel & Arpaci, 2025). Both instruments demonstrated high internal consistency in the current study (CPSES α = .95/.91; PECS α = .94/.85 for pre-/post-tests). Additionally, the researcher-developed PEPR, reviewed by three domain experts, provided an objective assessment of students’ programming performance and AI prompting skills, allowing triangulation with the self-report measures.