Grading PolitiFact (Florida): Bill Nelson and provisional ballot rejection

To assess the truth for a numbers claim, the biggest factor is the underlying message.
--Bill Adair

The issue:

The fact checkers:

Aaron Sharockman:  writer, researcher
John Bartosek:  editor


Analysis:

I wonder if Sen. Bill Nelson (D-Fla.) knows that the most important aspect of a numbers claim is the underlying message?  Even more so, I wonder if PolitiFact staffers Aaron Sharockman and John Bartosek realize it.

Nelson used the provisional ballot stat to bolster his claim that a Republican bill affecting election procedures targeted women and young people.  That's the way PolitiFact tells the story, anyway.  Good luck finding a quotation from Nelson making that clear.

Trust PolitiFact.

PolitiFact:

Republicans say it's a way to eliminate possible voter fraud. But Democrats like Nelson say it's an (sic) way to suppress the votes of people most likely to change their names (women) and people most likely to move frequently (young people and college students). At a press conference with Democratic state legislators on May 2, Bay News 9 and Central Florida 13 reporter Troy Kinsey asked Nelson what was so wrong with people having to cast a provisional ballot.

"Look at history," Nelson answered. "Look at the provisional ballots in 2008. Fifty-five percent of them were discounted. Fifty-five percent of the people who cast a provisional ballot in the last presidential election -- their vote did not count."

PolitiFact editor Bill Adair says the underlying message is the most important aspect of a numbers claim.  Since history suggests that PolitiFact often fails to bother separating the underlying message from the literal numbers claim, let's pause to evaluate Nelson's statement.

First, we have a literal numbers claim, that 55 percent of provisional ballots--presumably in Florida--were not counted.

Second, though more importantly using Adair's guidelines, Nelson sends the message that having to use a provisional ballot serves to suppress votes.  We can't reasonably presume Nelson to mean he wants illegal votes protected, however.  If we take that course then things like age restrictions on voting likewise count as voter suppression efforts.  Nelson is reasonably taken as talking about suppression of legal votes.

PolitiFact's subsequent assay of the facts shows that Florida's provisional ballots were considered unevenly according to locale.  For example, Broward County rejected 94 percent of its provisional ballots while Duval County rejected a mere 19 percent.  The differences in the numbers prompt quite a few questions, but the PolitiFact assessment is sadly superficial, limited to a quotation of a Pew Center on the States report mentioning some differences in rejection criteria.  PolitiFact offers no suggestions as to the significance of the rejection criteria, moving quickly to focus on the literal numbers claim:

In total, Pew reported that 35,635 Florida voters submitted provisional ballots in the 2008 fall election (.42 percent of the total 8.4 million who cast ballots for president). Of the 35,635 ballots submitted, 17,314 were counted and 18,321 were not counted.

To line that up with Nelson's claim, that means 51.4 percent of the provisional ballots submitted were not counted for one reason or another.

As Nelson could justify rounding to 55 from 52.5, his number represents an inflation of only about 2 percent.  So the literal claim as to the numbers is close, and we can file the figures away to contribute to our evaluation of the PolitiMath theorem of mathematical accuracy.

PolitiFact:

State Department spokesman Chris Cate also provided us with provisional ballot statistics from the 2010 general election. They are worth noting because the percentage of provisional ballots ultimately discounted was cut in half when compared to 2008. In the 2010 fall election, local canvassing boards considered 13,181 provisional ballots. Canvassing boards wound up counting 9,790 of the votes and tossing out 3,391 provisional ballots -- or 25.7 percent of the total number of provisional ballots.

Nelson said that "55 percent of the people who cast a provisional ballot (in Florida) in the last presidential election -- their vote did not count." Two slight clarifications: Nelson is slightly high on the percentage claim, and he fails to note that the percentage of provisional ballots counted improved greatly in 2010. We rate this claim Mostly True.

It is worth noting, or rather emphasizing, that PolitiFact has implicitly detected an underlying message that figures in the final ruling.  After all, if the question was only the percentage of provisional ballots rejected in the 2008 election then the number of provisional ballots rejected in the 2010 election is entirely irrelevant.  Yet PolitiFact justified dropping Nelson from "True" down to "Mostly True" based in part on the 2010 numbers.

PolitiFact's attempt to evaluate the underlying message qualifies as severely underdeveloped, unfortunately.

As noted above, the reasonable version of Nelson's argument hinges on the number of legitimate votes lost as a result of the provisional ballot process.  PolitiFact fails to address the issue--a great yawning hole in the story.  The difference in the percentages of provisional ballot rejection, by itself, means nothing.  What matters is the extent to which the percentages reflect the loss of legitimate votes.  And that's quite the can of worms that PolitiFact leaves untouched.

It is a type of voter fraud to vote in the wrong precinct.

I'll repeat that to give it time to sink in.

It is a type of voter fraud to vote in the wrong precinct.

Precincts exist for a reason.  They are intended to help limit who can vote on what.  New York residents should not vote in Florida elections, for example.  For that reason, one of the statistics bandied about by PolitiFact (quoted from the Pew Center report) caused me some consternation:  "In Broward County, 100 percent of rejected provisional ballots were not counted because they were cast by voters not registered in the state."

Broward, one of Florida's most populous counties, was the one that rejected 94 percent of its provisional ballots (see above).

The Pew report, in truth, provides only a wisp of support for Nelson's underlying argument.  That wisp comes from the fact that poll workers sometimes make mistakes in handling provisional voters and their ballots bold emphasis added):

Finally, more than 27,000 or 6 percent of rejected ballots were disallowed because of various errors, including incomplete provisional ballot envelopes, missing or non-matching signatures on the provisional ballot applications, incomplete applications, and envelopes that contained no provisional ballots. While a number of these errors were likely committed by voters themselves, some were the result of administrative problems at the polls.

That 27,000 was a national figure, by the way.

It simply isn't clear, contrary to Nelson's argument, that rejecting provisional ballots is a bad thing.  We don't want New York residents voting in Florida even if they are women or college students.  Nor do we want to make it easy for people to vote in the wrong precinct.

Nelson's underlying argument is a red herring offered with PolitiFact's tacit approval.  The vast majority of rejected provisional ballots get rejected for good reason.  And for some reason PolitiFact did not seem to think it was important to Nelson's argument.

Go figure.



The grades:

Aaron Sharockman:  F
John Bartosek:  F


Afters:

It has been argued that provisional ballots should count as legitimate votes to the extent that the votes align with the choices available in a given precinct.  All Florida precincts vote for governor of the state, for example.  While the suggestion addresses, at least partly, the problem of fixing elections with imported votes, it produces an odd result.  Why should voters using the provisional ballot carry a privilege unavailable to voters who turned up at the correct precinct?

Bottom line, this controversy underscores the difference between the Republican and Democrat views of the voting process.  Republicans tend to try to restrict voting to those with at least some qualifications.  Democrats tend to want as many votes cast as possible.