"Evaluation of Sampling Methods for Scatterplots" claims that people perform dot-density comparisons best when using random sampling rather than other sampling approaches. This claim and other core conclusions of the article are not supported by the article’s empirical evidence. The article’s reported results and figures do not meet its own stated threshold of statistical significance, and the analyses are ill-suited for the research questions. Some of these issues are present in the article and could have been spotted by reviewers, whereas other issues were only noticeable because I requested the code and data after publication (which a reviewer would have been prohibited from demanding during the IEEE TVCG review process). A reanalysis calls into question whether any generalizable claims can be made from these results. This comment, its analysis code, and the original article’s data are freely available at http://osf.io/hsujr.