Racial Bias in Forensic Decision Making. Comment on Yim, A.-D.; Passalacqua, N.V. A Systematic Review and Meta-Analysis of the Effects of Race in the Criminal Justice System with Respect to Forensic Science Decision Making: Implications for Forensic Anthropology. Humans 2023, 3, 203–218

A “Systematic Review and Meta-Analysis of the Effects of Race in the Criminal Justice System” [1] may confuse its readers and thus a clarification of this review article on “Cognitive Bias in Forensic Pathology Decisions” is required [2].

The review [1] deals with racial bias, an especially important issue when considering bias by experts (e.g., forensic experts analyzing evidence, medical experts diagnosing patients, etc.). The review contributes to the existing literature on bias and its sources [3], which is important for developing ways in which to address them (e.g., [4]).

As the review notes, there has been “intense public discourse that followed the publication” (see coverage also by Science magazine [5]) and, indeed, the review references nine Letters-to-the-Editor (references 107–115 in [1]). However, the review fails to provide a balanced view, as it does not also reference the responses to those Letters-to-the-Editor, thus providing the readers with a one-sided view.

Furthermore, while many of the referenced Letters-to-the-Editor were not only factually and demonstrably incorrect (as explicated in the unreferenced Responses), the Letters were an emotionally defensive pushback reaction to the first study that examined and revealed bias in forensic pathology decisions. As the Editor-in-Chief of the journal himself states in a Preface Letter (which is also not cited or referenced in the review), “It was not disappointing to me that this article received such scrutiny; however, the “emotional” nature of some of the letters was” (p. 2539 [6]).

It is also important to make clear that the article in question [2] is actually comprised from two studies. The first was a study of 1024 death certificates issued over a 10-year period in the state of Nevada for children under the age of six (see the Table in the review [1]). However, the statement made in the review that the study “serves as a primary example of how confounding factors may be conflated as racial bias [107–115], but due to research design, the “true effects” of racial bias are difficult to parse out” [1], does NOT relate to this study, but to the second study in the article [2].

Indeed, their reference to the Letters [107–115] regarding the confounding factors relate to the second study, not to the study of 1024 death certificates shown in the Table. This can be confusing to the readers and it is important to note and clarify that the issue of confounding factors relates to the second study in the article (not the first study that is presented in the review Table [1]).

The second study in the article [2] did indeed use two factors in the experimental conditions. It is this study that compared the manner of death decisions when examining the same medical information, but in one condition within a context of a black toddler brought to the hospital by the mother’s boyfriend, and in the other condition within a context of a white toddler brought to the hospital by the grandmother. While in the former condition the manner of death determination was considerably more ‘homicide’ (in contrast to ‘accidental’ death), in the latter condition it was more determined as ‘accidental’ death (in contrast to ‘homicide’).

Given that the medical information provided in both conditions was identical, there was a clear bias in the pathology decisions. However, given both factors (the race of the toddler and the caregiver) were manipulated, the study was not able to determine which of these factors (or both) biased the decisions. This is clearly stated in the study [2] and in the responses to the Letters.

However, the confounds related to the second study do not apply to the first study [2], where racial bias was not confounded with any other factor, and the review does not make this sufficiently clear. Furthermore, as a side note, it is also worth mentioning that the manipulation of race was performed in the second study explicitly with regard to one factor (the toddler), but was actually also implicit in the second factor (the caregiver): the grandmother of a white toddler is more likely to be white.

Studying racial bias in expert decision making, and in forensic decision making in particular, is a challenging task. The Letters and the Responses after the publication of the first article that examined bias in forensic pathology decisions [2] reveal some of these challenges. Nevertheless, since it is an important topic to research, it is therefore worth pursuing.