provisional.ballots.jpg

[Image courtesy of KSAZ]

Brian Newby’s latest post at ElectionDiary (“Provisional Ballots By The Numbers”) uses the idea of Pew’s Elections Performance Index as a jumping-off point into a very detailed look at how different methods of measurement capture different aspects of the elections process.

So, for example, he notes that Pew’s “score” for the State of Kansas dipped from 2008 to 2012 because of declining turnout, a function of a presidential election with no Senate race on the ballot. He also highlights the issue of data completeness – the quality and quantity of data reported by the state’s 105 and counties.

But the post gets even more interesting when Brian simply jumps into the provisional numbers themselves and lays out the numerous different ways and reasons such ballots do and don’t get counted.

First, the successful ballots:

In th[e 2012] election, we recommended 5,878 ballots to be counted. More to the point, that’s 5,878 envelopes to be opened. Only when opening them, do we know what is inside the envelope.

  1. Of the 5,878, the most common reason for a provisional ballot that can be counted is that the voter moved or changed his/her name and completed the required registration information (the back of the provisional ballot is a registration form). In this election, nearly half (2,624) of those recommended to count fit in this category. This number is usually much larger in a presidential election, simply because of the number of voters but also because many infrequent voters don’t think about voting until election day.
  2. 917 ballots were in a category that WE made provisional. The voter returned a ballot by mail and signed the ballot, but didn’t complete the address line as required by law. We worked with Secretary of State Kris Kobach’s office to request and obtain a 2012 Attorney General opinion that recommends these be counted. Previously, these would not have been counted. We’d advocated this position for years and are pleased that this minor technical error doesn’t invalidate the vote.
  3. The next number is a bit frustrating, really–771 voters received an advance ballot but voted at the polls instead. Now, sometimes, I think, voters worry that the mail return won’t be fast enough for the ballots to reach us by 7 p.m. election day (they can be dropped off, 24 hours a day), but often I believe these are voters who simply applied for an advance ballot so they had a sample (or even a souvenir, as I believe the case was in 2008).
  4. 478 voters went to the wrong polling place and cast a ballot that can at least be partially counted (for any races that apply to them, such as president). This number had dropped by 75 percent since our Joco-Polo voting location campaign and tools were rolled out in 2008.
  5. 412 ballots were “should have been perfect,” a kind way of saying, for whatever reason, the voters didn’t sign the pollbooks and vote. Sometimes, names hide, particularly in the cases of apostrophes and such in names.
  6. In 268 cases, we had provisional ballots issued for one of these categories and all was correct, except that the voter’s ID had not been marked as verified. We had these in a separate category because we didn’t know the ID had not been verified, only that it had not been recorded as such. Therefore, we recommended these be counted (we cite the legal reference to any recommendation in the right column of the canvass sheets).
  7. “FWAB” is the Federal Write-in Absentee Ballot. These essentially are military voters who can fax or email their ballots. These came in late on election day and have to be hand-counted because the ballots cannot be scanned. They could have been included in election night totals but the results would have been later in the evening. We had 214 of these.
  8. In 103 cases, a voter did not submit photo ID but did before the canvass. Most of these were ballots by mail.
  9. 68 times, we issued a replacement ballot in cases where the original was damaged or lost.
  10. We have a form for those voters whose name on their ID and registration don’t match, but they want to keep it that way. Usually, these are women who have different last names after life changes. In lieu of ID, they can complete an affidavit that explains the name discrepancy. 19 voters did this.
  11. In presidential elections, because voters who move can sometimes get caught between states’ voting requirements, there is a provision where voters can get a “presidential only” ballot. 4 voters fell in this category.

Additionally, on the “count side,” we have 174 voters who preferred to vote on paper. These weren’t provisional voters, but we have to verify these voters were eligible to vote (perfect voters at refer to voters where everything went the way it should–recognizing that everyone in Johnson County is perfect in their own way). We don’t bring these envelopes for approval by the Board of Canvassers to open, but we do wait until the canvass to process these ballots. We could add them in during the week leading to canvass but I prefer to have results updated once only after election night.

Then, the flip side:

On the not valid side, we had 2,136 ballots where we recommended the ballot envelopes not be opened.

  1. Most–1,330–were because the voter wasn’t registered.
  2. In 300 cases, the voter had an advance ballot by mail and did not sign and address the envelope or did not complete the registration form on the back of the provisional envelope at the polls.
  3. We check every signature on those voting by mail and 182 did not match their voter record.
  4. 76 voters did not provide a government-issued photo ID. In all cases in this election, these were ballots by mail.
  5. Sometimes voters use a different envelope when mailing back the ballot or put two ballots in one envelope. We can’t count these, and we had 71 of them.
  6. 25 voters, when casting a provisional ballot, used an invalid address as their registration address.
  7. Remember those voters who get a ballot by mail but vote at the polls? 15 voted both ways, likely because they either forgot (that happens) or they were worried we wouldn’t get the ballot in time. We count the first ballot cast.
  8. There is a Attorney General letter with a ruling that persons who act as Power of Attorney cannot vote on someone’s behalf. We had 14 of those cases.
  9. In 13 situations, a voter had some issue when voting and cast a second ballot provisionally.
  10. Again, remember the voters who requested paper but we have to verify that they were qualified? In 7 cases, they weren’t eligible to cast a vote and should have voted provisionally.
  11. 2 voters voted overseas ballots but didn’t meet the eligibility to do so.

The rejected provisional ballots represent about 0.7 percent of those cast. Provisionals, overall, are about 3 percent of the total, and that’s typical.

Although it helps, you don’t have to be a hard-core election geek to get several questions from these figures:

~ How do 1,300+ people go to the polls thinking they’re registered when they’re not? [This is one that always puzzles me – Confusion? Laziness? Misinformation?]

~ The 900+ voters who missed the address line on the envelope suggests a design issue (paging Drs. Chisnell and Quesenbery, stat!)

~ Would sample ballots reduce or eliminate the “advance ballot voters at the polls” problem – and if so are those benefits (plus voter information) worth the cost to produce and distribute them?

… and so on and so on and so on …

This is is just one set of numbers for one county in one state in one election year. Imagine what we could do if everyone was as thoughtful about this collection and analysis!

Thanks as always to Brian for pulling back the curtain on his operations for all of our benefit … with apologies to Stephen Sondheim,

To get the thing
That makes it worth
The journeying.
Into the w[ee]ds!

Posted in Uncategorized