Numeric Anchors and Biased Estimation: A Logistic Regression Approach

AbstractAnchoring bias describes the tendency for numerical judgments to be influenced by an initial reference value. While anchoring effects are commonly demonstrated through comparisons of mean estimates across conditions, fewer studies model anchoring as a probabilistic outcome at the individual-response level. The present study investigated anchoring bias using a between-subjects design with low- and high-anchor conditions and operationalized bias as a binary outcome based on the proximity of estimates to the anchor value (±50%). Participants provided a numerical estimate following exposure to an anchor and reported their confidence in that estimate. Binary logistic regression was used to examine whether anchor value and self-reported confidence predicted the probability of a biased response. The analysis did not reveal a statistically significant association between anchor value and bias classification, nor between confidence and bias classification, under the specified model. These results illustrate how operational definitions and modeling choices influence the detection of anchoring effects and demonstrate the application of logistic regression to the study of judgment biases.