From a recent Cleveland study on ‘04:
Almost 1,000 provisional ballots may have been wrongfully rejected because of registration problems alone. At least 944 rejected provisional ballots, mostly classified as “not registered”, were apparently mistakenly purged from the registration lists. Since this error was detected by only one type of search, which did not detect other voters who claimed similar errors, the true number of provisional ballots wrongfully rejected is likely to be higher.
We estimate that 2 out of every 5 provisional ballots that were rejected should have been accepted as legitimate. If we combine incorrectly purged provisional votes, projected votes rejected because of initial registration errors, provisional ballots lost through polling place misinformation and innocent errors filling out the provisional application, it appears that over 41% of rejected provisional ballots (or 14% of all provisional votes) may have been unnecessarily rejected.
We estimate that simply changing residence exposes voters to a 6% chance of being disenfranchised. Youth, the poor, and minorities are disproportionately affected. In fact, with respect to just provisional ballots, we found a two-fold increase in rejection rate in predominantly African-American compared to predominantly Caucasian precincts.
Full text:
http://www.clevelandvotes.org/news/reports/summary.htmlfrom Feb 2005-but important, in case you missed it.