Study of LR models developed for physical traits

Forensic investigations often rely on the use of physical traits to search for the potential person of interest. Physical trait data is collected from victim testimonies, obtained from video-camaras or photos surrounding the crime scene, among others. This data is used to search across databases in order to define potential persons of interest to be further analyzed. Recently, some models have been developed that allow computing likelihood ratios (LR) for physical traits. Rarer characteristics can provide higher LRs in identifications. Here, we exhaustively study the expected LRs considering different characteristics. We aim to provide the forensic practitioners a notion about which values are expected under different scenarios, for example, when the person of interest is the searched one, or if it is not.