Only 32% of the Dutch primary schools do something with programming, despite its benefits on students’ skills, job opportunities, understanding of the world, and diversity. Primary school teachers not getting involved with programming education can be caused by a lack of belief in their teaching programming abilities. The belief in your own abilities towards a certain task is referred to as self-efficacy. We hypothesize that Dutch primary school teachers have a low self-efficacy towards programming education (partially) due to the teaching method used in programming education. Programming is often taught via discovery learning while Dutch primary school teachers use teacher-centered methods such as direct instruction more often. The usage of a certain teaching method can also influence the performance of the teacher. When teachers are able to use their existing knowledge and skills, then they have adaptive expertise. This research aims at answering the following question: What is the effect of teaching methods on the self-efficacy of Dutch primary school teachers towards programming education? The research question is answered through a mixed methods study consisting of a questionnaire, as part of a quantitative research, and an experiment, as part of a qualitative research. The questionnaire includes questions about teaching methods, self-efficacy, and demographics. In total 259 teachers participated within the questionnaire. We found that both direct instruction and discovery learning are well known by Dutch primary school teachers. Direct instruction is used by almost all participating teachers in their regular education. However, discovery learning is used by only 53% of the teachers in our sample. We also found that gender, grade, programming experience, and programming teaching experience result in differences in teachers’ self-efficacy towards programming education. Furthermore, teachers who expect to use discover learning in programming lessons have a different self-efficacy than teachers that do not expect this. An experiment is set-up in which ten Dutch primary school teachers give four programming lessons in Scratch while following either a direct instruction or discovery learning approach. Data is gathered on self-efficacy and adaptive expertise. We found that getting involved in programming education, independently of the teaching method, can increase the self-efficacy of teachers. However, the teachers in the discovery learning expressed more negative feelings towards the experience of teaching programming. The usage of discovery learning also seems to bring the risk of a decrease in self-efficacy. We think that adaptive expertise gives us more insights into this decrease. The teachers with a decreased self-efficacy showed higher levels of adaptive expertise. We think that this contributes to them having higher expectations of themselves. When these expectations are not met, a decrease in self-efficacy can occur. The research we have done indicate that the self-efficacy towards programming education of Dutch primary school teachers can increase while using either direct instruction or discovery learning. The experiment indicates that the usage of discovery learning can cause more negative feelings as well as the risk of a decrease in self-efficacy.