[Associations between students' social background, mode of instruction and learning outcomes during the pandemic].

The COVID-19 pandemic has led to widespread school closures, which have been associated with massive changes in teaching and learning and student learning outcomes. While initial performance studies have examined learning losses due to the school closures, there are few findings to date that provide evidence on key predictors of the development of other student outcomes (stress experience, learning gain, learning time) during the school closures. This study approaches these questions by retrospectively predicting student self-assessments in these student outcomes for the periods before and during school closures by social background (academic household, language spoken at home) and instructional design (open vs. traditional). The multilevel regression models presented are motivated by theories of educational inequality and concepts of open instruction. In addition, the hypotheses tested are based on recent survey studies on COVID-19 and school. Results calculated on the basis of data from N  = 1566 upper secondary school students show higher values in the retrospectively assessed stress experience of students without German as their native language than for children with German as their native language during school closures. With regard to the retrospectively self-assessed learning gain and learning time, however, there are no correlations with social background (beyond the differences in the baseline values before COVID). With respect to the relations between open instruction and student outcomes, it appears that students from traditional instruction report statistically significant higher learning time during school closures than students in open instruction. Students in open instruction, however, do not rate their learning gain to be lower than traditionally taught students despite less learning time.