Monday, January 16, 2012

Signature Failure Rates -- Past statistics of invalid signatures in recall campaigns

With the Walker recall petitions due tomorrow, the petitioners will need 540,208 valid signatures to get on the ballot. According to reports, they will be handing in hundreds of thousands of signatures above that limit, which leads to a basic question -- how much of a cushion do the recall proponents need to qualify for the ballot? Let's examine some other examples to see what we know about signature failure rates.

Let's first start with the basic numbers.The 540,208 is derived from the 25% of the total votes cast in the 2010 gubernatorial election, which was 2,160,832. Wisconsin has 5.68 million people. As a point of comparison, Gray Davis' recall needed 897,158 signatures in a state of close to 36 million people. In California, the recall proponents turned in almost double the amount needed --1,660,245 signatures, of which 1,363411 were deemed valid. Doing some quick math on that one, the failure rate was a shade under 18%.

The other best point of comparison is the Wisconsin Senate recalls this past summer, but unfortunately, the data appears to be incomplete. Since there are nine, we won't discuss them all. Here's the GAB page and the  Ballotpedia link that lists the numbers handed in (which is an estimate -- they all didn't end up on a nice round number), and the number validated. The numbers don't fully add up -- not only the fact that some candidates have more than every single signature approved, but contemporary reports site divergent amounts of signatures handed in. For example the Dan Kapanke recall cites a huge error rate (30,000 handed in, only 21,776 valid -- a 27.5% failure rate). However, contemporary reports cite much different numbers (22,000 handed in) as well as for the other efforts. Did the GAB not check the remaining signatures once they approved the recall (in what may have been a two stage signature hand-in process?) I don't know, and it is strange that there is such a discrepancy. However, just to list it, there are two petitions with an alleged 25-27.5% failure rate, there are two (both Democrats) with a 17-18% failure rate, there are three with a 6-8%, and two with effectively zero.

These numbers should prove some comfort for the Democrats. Even if we take the likely incorrect figure of a 27.5 error rate for the Kapanke recall, the Walker petitioners would need a little less than 689,000 signatures to qualify. Since they had 500,000 half way through (though those signatures were the low hanging fruit), it could be that, as commentators and the Governor himself suggest, this will fly through without a problem.

Of course, there are other states to look at. This article lists a 10-15% signature error rate in a Michigan recall. And there are also examples of recalls with many more signatures being invalidated. In this Phoenix city council recall, over 5,000 signatures were handed in. The recall only needed 2,329. It didn't make it. The error rate was close to, if not over, 50%. 

Another recent matter is this on-going Las Vegas recall against City Councilman Steven Ross. The only contender on the ballot got on by two signatures (it was originally thought he didn't make it). They actually don't seem to analyze each signature, but take a sample and extrapolate (which would seem to be the basis of a lawsuit, but Ross has eschewed this option). California also uses a random sampling rule, though if the number falls between 95 and 110%, it gets the through search.

I'd love to give a rate on the recall that brought Walker into the Milwaukee County Executive's office, against Tom Ament in 2002, but hard numbers don't appear to be online. They may never have been checked by the Election Commission as Ament resigned before the recall took place and Walker in a special election to fill the seat.

There's also the possibility of a "blocking" campaign or deliberately false signatures.  This site, run by the people who claim to have gotten the Davis recall on the ballot, note that the Davis blocking campaign was extensive. Still, the blocking campaign failed. I don't think deliberately false signatures will have much of an effect -- the parties are keeping a running tally of whether the signatures look good. They will be able to weed out the obviously fake ones, especially after the Mickey Mouse controversy.

There's one other point that is worth considering. There may be more of an effort to examine these signatures than in other recalls. There also may be different numbers for Walker and for his LG (which is a separate recall petition). There's a lot of money in the coffers of both sides, and a lot of attention is being paid to this recall. This is not to say that other recall signature fights have been ignored. I'm sure they've been tough. But this could be more serious than any in the past. The Gray Davis recall may not have had the same level of scrutiny, simply because the signature amount was so overwhelming as to limit the value of a real fight by Davis. It could be the same thing happens here, but if the amount of signatures handed in is close (I would say sub-800,000) we could see a real fight on our hands.


  1. The average joe signing recall petitions is fairly innacurate at filling in the forms. Middle initials that do not match with signatures, illegible names and addressess, etc. Ignoring fraud, my inclination is for the error rate to easily exceed 25%

  2. I helped collect quite a few signatures here in Wisconsin this winter. Twenty to 25 per cent sounds about right as a failure rate. There may be other areas of life that wouldn't pass close scrutiny either.


Note: Only a member of this blog may post a comment.