Dealing With Data From Real Witnesses: Methodological and Analytical Considerations
Abstract
Laboratory studies are the cornerstone of eyewitness identification research, as they allow for the isolation and precise manipulation of variables under pristine conditions. However, these pristine conditions also limit the confidence with which we can generalize the findings of laboratory research to the experience of real witnesses. Furthermore, ethical and pragmatic constraints make it difficult to answer some research questions in a laboratory environment. For these reasons and others, researchers may wish to examine data from real witnesses. This type of research has enormous benefits and has provided valuable insights. However, it also presents several methodological and statistical challenges. In this chapter, we outline different types of research that have been conducted with real witnesses, including field experiments and analyses of archival data. We show how this type of research can be important and informative. We then consider some of the special challenges inherent in field data, including the absence of ground truth surrounding the guilt or innocence of the suspect, correlations between predictors, and non-independence of data points. Through examples, we show how misleading conclusions can be drawn if these factors are not properly considered. We end with recommendations for any researchers wishing to conduct a field experiment or archival analysis of eyewitness identification data.