Date of Award
Master of Arts (MA)
First Advisor or Mentor
In this study, we were interested in investigating if the Betaface facial analysis program reliably predicts eyewitness lineup choosing behavior. If face analysis programs are as good or better than human judgements, using them could be a reliably more efficient, reproducible, and equitable basis for choosing fillers and evaluating lineup fairness. We collected 27 datasets from eyewitness researchers and analyzed them to produce Betaface similarity values, which measured the similarity between all the photos in each array. We compared these Betaface data to the identification data from the original studies. Our analysis of the arrays via Betaface yielded data with a fairly high degree of GT-IT, GT-filler, and IT-filler similarity across arrays, which implies that the arrays are quite fair. There is no evidence to show that Betaface can reliably predict identification choosing behavior. To find a clearer relationship in Betaface values and identification rates, we would require data from studies that are attentive to systematically manipulating similarities in the selection of the fillers and IT. Manipulating these variables independently would yield non-correlated measures; without these manipulations, the lineup construction variables in the current dataset display too little variability to permit detection of possible Betaface-identification relationships.
Kane, Phoebe, "An Archival Exploration of Lineup Fairness in Eyewitness Research" (2023). CUNY Academic Works.