"Searching for equity: White normativity in online skin cancer images"

Abstract

In this paper, we examine the range of skin tones represented in publicly available online image search results through which non-medical audiences might seek information about skin cancer signs, symptoms, and risks. We use the Fitzpatrick scale, a numerical classification system grouping six human skin tones (or "phototypes") in dermatology, as a guide for analyzing the skin tones appearing in (n = 1600) Google image search results for search terms related to skin cancer. We find that light skin tones (1,2, and 3 on the Fitzpatrick scale) comprise the significant majority (roughly 96%) of those depicted in Google image searches of information about skin cancer signs and prevention; dark skin tones (4, 5, and 6 on the Fitzpatrick scale) appear with significantly less frequency (roughly 4%) in the same search results. Disparate representation of diverse skin tones-and, more specifically, omission of dark skin images-suggests that racial biases inflect the search results generated by seemingly race-neutral skin-cancer related search terms. This embedded racial bias privileges white normativity to the disadvantage of dark-skinned patients, who are most likely to be racially classified as Black.