Wie sollten Rückmeldungen von quantitativ erfasstem Schülerfeedback (nicht) gestaltet werden?

Abstract

Data-based school-improvement depends on the correct and concise extraction of the information from the data through teachers. Such data are usually not provided in raw form, instead graphs are presented which encode transformed and aggregated data. Such graphs generally allow an effective extraction of the relevant information but simultaneously challenge teachers’ statistical literacy. The present study investigates this trade-off by design- based-research with private tutors. We triangulated perceived informativity and validity of interpretation (N = 83) about differently aggregated and encoded feedback on instructional quality (N = 9730) with log files from a feedback web application (N = 232). We found associations between dwelling time (on specific graphics) and the corresponding perceived informativity (positive, monotonous, quadratic) and perceived validity of interpretation (U- shaped, quadratic). Additionally there was evidence for a preference of very simple statistical coefficients and transformations (arithmetic mean, likert scale). The relevance of these results was discussed on the background of the present highly mathematics-oriented sample.

Making sense of evidence-based governance reforms: an exploratory analysis of teachers coping with the Austrian performance standard policy

“…The aggregation level of the reported results is particularly important: More profound data representation triggers individual diagnostic use of data and increases the willingness to reflect results (Breiter & Light, 2006). Including 'further questions' at the end of data feedback boosts teachers' examination of the results, however, without significant influences on classroom development actions (Merk et al, 2018).…”

“…A central barrier to data use may be the understanding of the statistical information. Teachers seem to focus primarily on simple information, such as arithmetic means, and do not take in more sophisticated information which may give impulses for instructional development, such as effect sizes or mean differences (Merk et al, 2018). Support persons for reading and interpreting data feedback ('feedback moderators') were used by about a third of Austrian schools.…”