# Estimates of the number who were incorrectly purged from Florida's list of eligible voters in 2000

In 2000, many voters were incorrectly removed from Florida's list of eligible voters. (Notoriously, if you happened to have a name that was similar to a convicted felon's, there was a good chance you'd be removed. For example, Willie D. Whiting Jr was removed because his name was similar to a convicted felon named Willie J. Whiting.)

Much has been written about this, but the estimates I've come across vary wildly and are usually crude, back-of-the-envelope estimates. Either that or they simply give estimates without citing where the estimates came from. Examples:

Of the 3,258 names on the original list, therefore, the county concluded that more than 15 percent were in error. If that ratio held statewide, no fewer than 7,000 voters were incorrectly targeted for removal from voting rosters.

Salon's estimate at least has some hint of math and thought put into it, though I don't know where the 7,000 number comes from. Presumably it comes from 7,000/0.15 ≈ 46,667 total names on "scrub" lists. But I don't see any similar figure quoted in the article.

The next three either have no sources cited or if they do and you trace the sources, you usually end up at some newspaper article where some journalist simply claimed such-and-such a number.

Bill Nelson reportedly claimed:

In the 2000 Florida election, at least 1,100 eligible voters were wrongly dropped from voting rolls in an attempt to purge a list of felons

Nelson precisely cites a Palm Beach Post analysis that put the total number of people wrongfully dropped from the voter rolls in 2000 at about 1,100. Other estimates of wrongful targets are much higher, and experts about the 2000 election tend to go with those. Trouble is, we don’t have exact figures.

In the infamous Florida purge of 2000—for which conservative estimates place the number of wrongfully purged voters close to 12,000—Florida registrants were purged from the rolls if 80 percent of the letters of their last names were the same as those of persons with criminal convictions.

In NAACP v. Harris, a class-action, Blacks argued that over 57,000 Black voters were wrongfully purged from Florida's voting lists and that Black communities were provided with defective voting mechanisms. The case settled. However, had their vote not been disenfranchised the presidential election would have resulted in a victory of Albert Gore.

This last estimate is the most egregious. It gives by far the largest figure I've seen ("over 57,000"). It cites as a source the NAACP v. Harris complaint (PDF), but if you read that complaint, nowhere can you find that 57,000 figure!

Are there any credible (academic) estimates as to how many such voters exactly were incorrectly removed?

Also, bonus questions:

• What were their demographic characteristics?
• How would they have voted?
• What is your problem with the estimates you cited, and how would you expect us to be able to provide better ones? Jul 2, 2017 at 7:59
• @Avi: I have now added explanations as to why they are poor. I can easily imagine methods that would yield more reliable estimates, but these would require more work than simply one journalist quoting another journalist who quoted another journalist. For example, go through one-by-one the list of all purged voters and investigate whether each was or was not indeed a convicted felon (or had other legitimate reasons for being on the list). Very simple, but would obviously take a lot of time.
– user2212
Jul 3, 2017 at 0:52