Signal Detection Theory with Multilevel Models for Dyadic Judgments: A Practical Tutorial for Psychological Research

Researchers across psychology often interpret binary judgment data using raw accuracy scores that confound sensitivity with response bias and are influenced by unequal base rates. Signal detection theory (SDT) offers a principled alternative, yet most tutorials focus on single perceiver tasks and do not address the multilevel structure in interpersonal and behavioral research. This tutorial extends standard SDT applications by demonstrating how SDT parameters can be estimated within multilevel generalized linear models for dyadic and other crossed designs. The tutorial introduces SDT concepts, shows how to translate SDT parameters into probit mixed-model coefficients, and provides reproducible code for fitting these models with crossed perceiver and target effects. It also presents simulation-based guidance for study design. Finally, the tutorial includes a Shiny application for readers to explore these concepts interactively. Although demonstrated with romantic interest judgments, this workflow generalizes to any psychological domain involving binary decisions and multilevel data structures.