Race and ethnicity data for first, middle, and last names

We provide datasets that that estimate the racial distributions associated with first, middle, and last names in the United States. The datasets cover five racial categories: White, Black, Hispanic, Asian, and Other. The provided data are computed from the voter files of six Southern states -- Alabama, Florida, Georgia, Louisiana, North Carolina, and South Carolina -- that collect race and ethnicity data upon registration. We include seven voter files per state, sourced between 2018 and 2021 from L2, Inc. Together, these states have approximately 36MM individuals who provide self-reported race and ethnicity. The last name datasets includes 338K surnames, while the middle name dictionaries contains 126K middle names and the first name datasets includes 136K first names. For each type of name, we provide a dataset of P(race | name) probabilities and P(name | race) probabilities. We include only names that appear at least 25 times across the 42 (= 7 voter files * 6 states) voter files in our dataset. These data are closely related to the the dataset: \"Name Dictionaries for \"wru\" R Package\", https://doi.org/10.7910/DVN/7TRYAC. These are the probabilities used in the latest iteration of the \"WRU\" package (Khanna et al., 2022) to make probabilistic predictions about the race of individuals, given their names and geolocations.