Search for: All records

Psychological science can benefit from and contribute to emerging approaches from the computing and information sciences driven by the availability of real-world data and advances in sensing and computing. We focus on one such approach, machine-learned computational models (MLCMs)—computer programs learned from data, typically with human supervision. We introduce MLCMs and discuss how they contrast with traditional computational models and assessment in the psychological sciences. Examples of MLCMs from cognitive and affective science, neuroscience, education, organizational psychology, and personality and social psychology are provided. We consider the accuracy and generalizability of MLCM-based measures, cautioning researchers to consider the underlying context and intended use when interpreting their performance. We conclude that in addition to known data privacy and security concerns, the use of MLCMs entails a reconceptualization of fairness, bias, interpretability, and responsible use.

 

Given significant concerns about fairness and bias in the use of artificial intelligence (AI) and machine learning (ML) for psychological assessment, we provide a conceptual framework for investigating and mitigating machine-learning measurement bias (MLMB) from a psychometric perspective. MLMB is defined as differential functioning of the trained ML model between subgroups. MLMB manifests empirically when a trained ML model produces different predicted score levels for different subgroups (e.g., race, gender) despite them having the same ground-truth levels for the underlying construct of interest (e.g., personality) and/or when the model yields differential predictive accuracies across the subgroups. Because the development of ML models involves both data and algorithms, both biased data and algorithm-training bias are potential sources of MLMB. Data bias can occur in the form of nonequivalence between subgroups in the ground truth, platform-based construct, behavioral expression, and/or feature computing. Algorithm-training bias can occur when algorithms are developed with nonequivalence in the relation between extracted features and ground truth (i.e., algorithm features are differentially used, weighted, or transformed between subgroups). We explain how these potential sources of bias may manifest during ML model development and share initial ideas for mitigating them, including recognizing that new statistical and algorithmic procedures need to be developed. We also discuss how this framework clarifies MLMB but does not reduce the complexity of the issue.