Amid claims of fraud during the 2020 election cycle, I scraped official Detroit precinct voting data to compare to see if vote totals followed Benford's law.
A pdf of the file I used is provided, as well as a csv file that was crudely generated by copy-and-pasting the pdf into a spread sheet. I then scraped the csv to get clear results for each district, then did the Benford analysis on it.
It's been noted by others that there is no clear indication that voting results ought to follow Benford's law (in particular, because the size of precincts is not exponentially distributed). Some more technical work has been done by other (see Colin Adams) that analyzes voting data modulo an integer < 10 and/or looks at the second digit, as opposed to the first (as Benford's Law is traditionally applied).
A pdf of the results are provided as well.