The self-explanation effect (SEE) has great potential in strengthening student learning outcomes. With adequate instructional guidance, this potential is even higher. Our study aimed to examine to what extent direct instruction (DI) and indirect instruction (II) in science, technology, engineering, arts, and mathematics (STEAM) education can strengthen different types of student engagement, as well as whether and to what extent the SEE can support instructional STEAM learning and enhance student engagement. The research was conducted through a quasi-experimental design. The data were collected using an instrument–the engagement scale and were analyzed using confirmatory factor analysis (CFA), one-way ANOVA analysis, and paired sample t-test. The study involved 103 students aged 9-10 years. The results of CFA showed that the adapted version of the instrument for measuring four types of student engagement–emotional, behavioral, cognitive, and agentic–represents a valid and reliable measure for these purposes in primary education. In addition the results of one-way ANOVA analysis and t-test revealed that the usage of DI and II proved to be successful in enhancing all four types of student engagement in the performance of STEAM activities, whereby the combination of the usage of DI with the SEE was singled out as the most successful strategy. As a significant practical implication, this research underlines the need to create appropriate conditions for introducing the SEE in the teaching of STEAM.