Teachers’ Sensitivity to the Reliability of Information in Making Pedagogical Decisions

Models of teaching as a decision-making process have questioned whether teachers make judgments about student performance following Bayes’ theorem or simpler heuristic rules that may lead to systematic errors. Although research suggests that people are not Bayesian, Winkler argues that professional decisions may be Bayesian. To determine whether teachers’ estimates of student performance are consistent with Bayes’ rule, 164 subjects (teachers and non-teachers) were randomly assigned to one of 16 cells (2 4 design). Sixteen different stories about a fictitious student were constructed and presented to the subjects. These stories varied in terms of the reliability and valence of the information they contained. Subjects read the first section of the story and were asked to make initial estimates of the student’s future academic performance and preinstructional and interactive decisions. They then read the second section and were requested to revise their estimates and decisions based on the additional information. A comparison of the data with a path model representing Bayes’ rule showed that subjects’ estimates and preinstructional decisions roughly followed Bayes’ rule, while subjects may have found information about general ability less relevant to interactive instructional decisions.