Mathematics self-beliefs influence students’ self-image, classroom behavior, engagement, and motivation. However, current research faces challenges due to overlapping constructs, which can lead to high correlations and ambiguous findings. For instance, while mathematics self-efficacy is often defined broadly in educational research, task-specific mathematics self-efficacy refers specifically to confidence in completing particular tasks. To enhance the credibility of predictions regarding mathematics achievement, it is crucial to distinguish between different types of general mathematics self-beliefs. This thesis aims to provide new insights into these beliefs by refining methodologies for their measurement.

An exploratory factor analysis of upper-secondary student data, utilizing both quantitative and qualitative reasoning, revealed a clear distinction between mathematics self-concept and generalized mathematics self-efficacy beliefs. A key difference between these two constructs is that the latter significantly influences student engagement. Additionally, the identified factor structure included two constructs related to mathematics anxiety. One noteworthy finding is that the worries of providing incorrect answers in the classroom significantly contribute to gender differences and diminish students’ agentic engagement. Overall, these findings underscore the importance of removing variables with substantial intercorrelations from the analysis, provided this procedure is supported by qualitative reasoning. The findings of this thesis support a pragmatic approach to investigating mathematics self-beliefs, ultimately offering new insights into the significant role of autonomy-supportive teaching in shaping students’ general mathematics self-beliefs.