Multiple paired comparisons =========================== ---------- This project exploits a time-honored technique for perceptual scaling that was first developed in the early twentieth century by psychologist Louis Thurstone. The goal of this technique is to scale a set of stimuli with respect to their perceived similarity to a “target” stimulus. In simple terms, the technique involves first viewing (or experiencing via some sensory modality) the target stimulus. This is followed by viewing a small number of stimuli in pairs and making a judgment for each pair as to the stimulus that is more similar to the memory of the target stimulus. The product of this procedure is a scaled ordering of stimuli based on perceived similarity to the target. The protocol of perceptual scaling by multiple paired comparisons has a long history of use in real world applications, such as optometric examinations for refractive error, food and wine preference analysis, and a broad range of market research topics. But multiple paired comparisons have never been used in the context of eyewitness identification. This perceptual scaling procedure is now being used by our research team to address two significant problems in eyewitness identification. Lineup procedure ---------------- ---------- First of these problems concerns the best way to conduct a lineup procedure. The traditional lineup procedure is known as the “simultaneous lineup," in which all lineup faces (typically six) are viewed at the same time. A more recent procedure is known as “sequential lineup,” in which faces are presented one at a time. Our plan is to directly compare laboratory tests of the novel paired comparison lineup protocol to analogous tests of the well established simultaneous and sequential lineup protocols. We hypothesize that a lineup procedure based on Thurstonian perceptual scaling will yield improved eyewitness performance, relative to that obtained by either simultaneous or sequential lineup procedures. We thus predict that this new approach will not only lead to more accurate suspect identifications, but it will also provide a new scale for measuring witness correctness and confidence. Filler selection ---------------- ---------- First of these problems concerns the best way to conduct a lineup procedure. The traditional lineup procedure is known as the “simultaneous lineup," in which all lineup faces (typically 6) are viewed at the same time. A more recent procedure is known as “sequential lineup,” in which faces are presented one at a time. Our plan is to directly compare laboratory tests of the novel paired comparison lineup protocol to analogous tests of the well-established simultaneous and sequential lineup protocols. We hypothesize that a lineup procedure based on Thurstonian perceptual scaling will yield improved eyewitness performance, relative to that obtained by either simultaneous or sequential lineup procedures. We thus predict that this new approach will not only lead to more accurate suspect identifications, but it will also provide a new scale for measuring witness correctness and confidence. We seek an evidence-based method for filler selection, which is rooted in human judgments of the similarity of facial appearance. The perceptual scaling method described above is the vehicle by which we plan to achieve this. Specifically, we will ask human subjects to make paired comparison judgments of similarity to a “target” face. By this means we should be able to obtain, for a given target face, a scalar value of perceptual similarity to each additional face in a database. With these empirically assessed values of perceptual similarity in hand, we can assess eyewitness performance using fillers that are known perceptual distances from the suspect.