Knowledge that Transforms

Biases in judgments and decision making that are commonly explained in terms of the judge's prior expectancies can originate in unbiased environmental learning. Applying a recently developed sampling approach to decision making to the assessment of student achievement in a simulated classroom, we investigate systematic biases in teacher's judgment of student achievement. What might appear to reflect teacher expectancies based on students' overall ability level, gender stereotypes, or naı̈ve behavioral theories can be explained by ordinary learning rules underlying decision making, such as increasing accuracy with increasing sample size of observations in a probabilistic environment. Ability (% correct answers) and motivation levels (% raising hands) of 16 students in eight subject matters were manipulated. In Experiment 1 the judged difference between smart and poor students increased with sample size, due to participation rate or unequal teacher attention. In Experiment 2, verification biases in hypothesis testing about the relative assets of boys and girls in language and science were shown to be independent of gender stereotypes, but mediated instead by differential size of stimulus samples. Experiment 3 showed that judges' sensitivity to environmental base rates can mimic expectancy biases resulting from scripted knowledge.