We expect politicians to exaggerate. But when they start making stuff up, they have to be called on it.

Last Wednesday, Al Schmidt, the sole Republican on the three member City Commission that runs elections in Philadelphia, released a report on "voting irregularities" in the April 2012 primary.

The report was praised by Harrisburg Republicans who said it showed Philadelphia is a swarming nest of voter fraud and that's why we need the new voter ID law. They exaggerated the report's findings, but that's par for the course in politics.

 

Then I saw the fundraising appeal from State Republican Chairman Robert Gleason, which began with this sentence: "Yesterday, Philadelphia officials released an alarming report showing that hundreds of cases of voter fraud had occurred during the past election."

In fact, the report said no such thing. Schmidt, in a lengthy news conference Wednesday, said no such thing. More on what the report does say in a moment.

I called Schmidt Friday, and asked him six times if Gleason's statement was accurate. He never directly answered the question (You can hear our conversation by playing the audio above).

When I spoke to Gleason, he pretty much admitted the report didn't document hundreds of cases of fraud.

"Only time will tell as to how much fraud there was, if it is fraud or if it is irregularities, or just a sloppy system," he said.

I asked if he'd overstated things in his fundraising letter.

"No, I don't think so," he said. "Because you know what? Sometimes you say, 'well they're innocent until proven guilty.' Well there's a lot of people that think you're guilty until proven innocent. So I don't have any problem with saying that in a fundraising letter."

You can read the full text of Gleason's fundraising appeal here.

Philadelphia - land of the voting dead?

I take this subject seriously, because over the years I've covered a lot of Philadelphia elections and looked into a lot of allegations of election tampering.

Except for one famous case involving absentee ballots in the early 90's, which led to changes in state law on procedures for absentee voting, most of them haven't amounted to much.

One veteran of Philly elections used to tell me that votes are stolen in every election, in Philadelphia and most everywhere – just not very many. I think Schmidt's findings are consistent with that impression.

I can hear you saying one illegal vote is too many, and you're right.

But it's an imperfect world, and people who study voting issues seriously know there's a tension between ballot security and ballot access, and both are important. You want to make sure elections are clean, and you want to make it easy for qualified citizens to vote.

The tension comes up again and again, in voting technology, early voting, internet voting and other issues.

When we take steps to expand the franchise, we should consider whether they weaken ballot security, and vice versa.

And from everything I've seen, the Pennsylvania voter ID law will only prevent an exceedingly rare kind of fraud, which is impersonating another voter.

And in fact, Schmidt's report bears that out. It doesn't identify a single case of voter impersonation in the 2012 primary. And, with one partial exception, none of the potentially illegal practices it exposes in that election would be prevented by requiring ID at the polls.

What's really in the Schmidt report

Schmidt's report does identify some cases of apparently illegal conduct: one woman who appears to have voted twice in two different voting divisions; one case of voter impersonation (but not in the 2012 primary, which the report focused on); 23 cases of un-registered people convincing poll workers to let them sign "voter slips" in violation of procedure and cast machine votes; one polling place in the Northeast where six more votes were tallied on machines than voters who signed in (curiously, the extra votes were all cast in the Republican primary in a predominantly Democratic division); and seven non-US citizens voting over the last 10 years.

These are small numbers, compared to about 170,000 votes cast in the primary.

But Gleason and others say this may be just the tip of the iceberg. Most of these findings came from a small sample of voting divisions. Expand it across all 1,687 divisions, they say, and you may have a mountain of fraud.

The problem with that argument is that the sample in the report wasn't randomly selected. The 14 divisions examined were those flagged in a comparison between state and city records (more on that in a moment).

So while it's possible there are great pools of slime in the rest of the voting divisions, it may also be that we've seen the worst here, and it's far from hundreds of cases of fraud.

In one category of irregularity, the use of "voter slips" to cast machine ballots, the report looked at all of them citywide, because they're easy to examine. So the 23 unregistered voters who slipped through that crack represent the citywide total – again, 23 out of roughly 170,000 votes cast.

Hundreds of cases – of something

So where are Gleason and others getting this "hundreds of cases of fraud" thing?

The report says there are 434 voting divisions in the city that appear to have at least one more vote on the machines than voters who showed up on election day. In all, this amounts 779 possible "overvotes" citywide.

People have noticed such discrepancies before, and there's a simple and innocent explanation that may account for most of the overvotes.

This involves a comparison of two different sets of records, one compiled in the city and another kept at the state.

The city (and every other county) tracks every voter who shows up on election day, and at some point after the election, workers scan bar codes associated with each voter to create a state record of who voted.

The state database is called SURE, for Statewide Uniform Registry of Electors. So it's entirely possible, as Schmidt noted Wednesday, that in scanning tens of thousands of documents, a few can be missed.

If you're scanning the records of 121 voters who cast ballots in a polling place and you miss a couple, you'll have a discrepancy: state records will show 119 voted in that division, but machines recorded 121 votes.

This is important: In such a circumstance, there was no illegal vote, and there was no impact on the outcome of any race. There was simply an error made after the fact in creating an individual voter's record, not an error in the returns.

The 434 divisions the Schmidt cited averaged less than two overvotes apiece. Could easily be scanning errors.

What now?

There are some steps the city commissioners can take to tighten up procedures, including one recommended in the report. Schmidt should work with his fellow commissioners to make sure they happen.

Schmidt also says this stuff should be investigated further, and that's a good idea. While I don't think this report proves there's a serious problem in Philadelphia, the data doesn't rule it out either.

Doing a serious investigation would involve a lot of work that Schmidt doesn't have the staff for. It would be great if a non-profit or academic institution could get a grant and do this right.

They could take a larger, representative sample of city voting divisions in this election, look at all the records in those polling places and interview voters to either clear up discrepancies or uncover misconduct.

That might tell us something.