Teachers’ self-reported and actual content-related TPACK – new results on their relation and gender differences

•Actual, contextualized knowledge relate to self-reports of content-related TPACK.•Actual TCK and to a lesser degree actual CK predict corresponding self-reports.•Self-reported PCK and TPCK are closer related to actual (T)CK than to actual (T)PCK.•Over all dimensions, no gender bias is found for self-reported knowledge.•Yet, self-reports of pedagogy-related knowledge seem to be biased by experience. Measuring Technological Pedagogical and Content Knowledge (TPACK) in context is still a pertinent issue, as previously rather decontextualized self-reports have been the predominant measure, while knowledge test instruments are scarce. Self-reports can be interpreted as general measures of performance-related self-beliefs. Still, due to the contextualized nature of TPACK and potential biases, their use as proxies for actual knowledge has been criticized. Self-reports may be especially gender biased as women often underestimate their performance in STEM subjects. Drawing on a sample of N = 161 mathematics in-service and pre-service teachers, we aim to analyze (i) the structure of the self-reported knowledge and (ii) the relationship between self-reported and contextualized actual knowledge. To this end, we used general TPACK self-reports and a test instrument that infers the amount of knowledge separately for each dimension from performance over multiple authentic demands that teachers encounter teaching secondary school mathematics. The current study shows that the TPACK self-beliefs can be separated and measured reliably. Although all self-beliefs show bivariate relations to corresponding actual knowledge dimensions, this changes for PCK and TPCK self-beliefs when other actual knowledge dimensions are controlled. We interpret these findings that TCK self-belief and to a lesser degree CK self-belief seem to be calibrated according to corresponding actual knowledge, while PCK and TPCK self-beliefs are primarily calibrated according to non-pedagogy-related actual knowledge. Lastly, we do not find gender biases, but a small gender effect with lower actual and self-reported knowledge for female teachers over all dimensions.