This note provides supplementary arguments for Billot, Gilboa, Samet, and Schmeidler (2005, hereafter BGSS). The result there is strengthened so that it also holds in situations where multiple cases are mapped to the same probability assessment, a possibility allowed but unaddressed by the original paper. Link to the article
WANG, F. (2019). Probabilities as Similarity-Weighted Frequencies: A Comment. Econometrica.