Diagnosegenauigkeit von Deutschlehrkräften in der Grundschule: Eine Frage des Antwortformats?

Teachers' ability to accurately assess students' performance and item difficulties is essential for making adequate educational decisions in the classroom. So far, it is unknown how the accuracy of teachers' judgments is also related to the item response format that is used for assessing students' performances. Using data of state-wide reading proficiency tests in grade 3 (VERA) from the years 2012 to 2016 (N = 973 elementary school classes), we tested whether teachers' judgment accuracy was comparable for items with closed response formats (multiple-choice [MC]) and items with open response formats (constructed response [CR]). Teachers tended to overestimate students' performance for CR items and to underestimate students' performance to a similar extent for MC items. In four out of five years, the correlation between teachers' judgments and students' performances was significantly larger for CR items than for MC items. Thus, response formats should be taken into account when it comes to the accuracy of teachers' judgments. The findings are discussed with regard to MC-specific item properties and possible judgment processes.