Today is the due date for signatures to be handed in the recall of Los Angeles District Attorney George Gascon. Petitioners have said that they will be handing in more than the 566,857 signatures needed to get on the ballot. This will be almost certainly the fifth most signatures handed in for any recall in US history and the most for any non-state-wide office. If it gets to the ballot, it would be the third most required (after the two California Governor recalls, above the Wisconsin Gov/Lt. Gov). But the real question is: Will it be enough?
I'll update this once we get the actual numbers, but the number to look for is about 700,000. As explained below, a large number of signatures will be rejected -- Based on past experience, we should probably expect about 20% will be declared invalid. So the 700,000 gets the petitioners in that range of what they need. 675K, they are very likely to fail. 725K, they are looking very good.
Let's look at the recent recalls to see why we think 20% is the right number:
- In 2021, the Governor Gavin Newsom recall, which needed 1,495,709, saw a 19% failure rate across the state. Los Angeles County was almost a perfect mirror of the state, as 19.44% were rejected. Note that (as explained below), each signature was checked, as opposed to the sampling method that will take place in the Gascon recall (and the SF and Sonoma ones noted below).
- In San Francisco, we had four prominent recalls in 2022. San Francisco District Attorney Chesa Boudin saw 83,484 signatures handed in, and they needed 51,325. The sampling of 4174 signatures saw a 20.96% failure rate, more than enough to get on the ballot.
- The three San Francisco school board members saw signature failure rates of 20.8% for Alison Collins; 20.64% for Gabriela Lopez and 19.74% for Faauuga Moliga. Each needed 51,325 signatures and they all had over 71,000 signatures handed in. So all were easily adopted.
- Shasta County Supervisor Leonard Moty's recall effort saw a 25.79% rejection rate, though they only needed 4308 to get to the ballot.
- Sonoma County District Attorney Jill Ravitch recall in 2021 saw a 25.83% rejection rate (they needed 30,056 valids).
- (Update -- don't know why I left this out, probably because it didn't make the ballot) A recall effort against Los Angeles Councilmember Mike Bonin saw 39188 signatures handed in, but more than 13,000 were thrown out. They needed 27,317. They got 25,965. I don't see an official statement by the clerk though. So that is a rejection rate of about 34%.
- Going quite a bit further back, in 2003 Gray Davis gubernatorial recall, the failure rate was just under 18%.
So what's this about? As we’ve seen so often with recalls (and petitions in general), signatures will be tossed out. The reasons are usually mundane, which is what happens when you’re collecting signatures on the street and frequently paying people to collect. The problems range from the signers not being registered (based on the SF recalls, which has the info, the vast majority of failed signatures were due to not registered; second is wrong address) to signing more than once to living out of district or state or (in rarer instances) using fake names. While most recall attempts fail because petitioners do not hand in any signatures, over the last nine years (2012-2021) I count at least 146 recall attempts throughout the country in which petitioners handed in enough unverified signatures, but because of a high rejection rate resulted in the recall failing to get on the ballot. 40 of these rejections happened in California. So while not an everyday occurrence, signature failures are certainly something to consider.
In the past, I've seen articles suggesting that a general rule of thumb is a 10-15 percent failure rate, though I'm not sure where that is from (and the article links are now dead). We actually see a 10 percent number referenced in the law regarding local recalls.
Random Sampling Method:
Unlike at the state level, if there are more than 500 signatures submitted for the recall, the clerks use a random sampling verification technique (take sampling/determine the number of valid/extrapolate to the rest). Using this procedure, if valid signatures top 110 percent of the minimum number, the recall qualifies and moves to the ballot. (If it is between 90-110 percent, they verify each one. If it is below 90 percent, it fails). Note that the Sonoma County D.A. recall came in at 106%, requiring the manual recount.
Beyond the very recent recalls, we have some others worth considering.
The last two state level recalls against state Senators saw significantly worse results for petitioners, even though the recalls got to the ballot. In 2018, the failure rate for the State Senator Josh Newman recall topped 25 percent (73.66 percent were validated).
The 2008 recall effort against State Senator Jeff Denham, saw a 41.5 failure rate (58.58 percent were validated). This verification was under the random sampling technique (the law was changed in 2018 to remove state level recalls, at least partially to slow the Newman recall effort).
There is another recent recall effort that is also worth considering. In 2018, Santa Clara Judge Aaron Persky faced a recall, the first against a judge in California since 1932 (and the first in the US since 1982). The judicial district was much larger than a regular state senate one. Petitioners needed 58,634 signatures to get on the ballot. Using the random sampling rules, the registrar found that 3389 of 4727 signatures were valid. Since the petitioners handed in 94,539, this was more than enough to get on the ballot. But the failure rate was over 28 percent (the verification rate was 71.6 percent).
Do initiatives tell us anything on this front? Prop 22 was the most high profile initiative in the 2020 race, as petitioners spent more than $6.4 million to get on the ballot. Petitioners needed 623,212 and handed in 987,813 signatures. Once again using the random sampling method, 22.5 percent were found invalid (77.5 percent were verified). A similar result can be seen in a Proposition that required a higher signature total. Prop. 15, which spent nearly $6 million to get on the ballot, was a Constitutional Amendment, which therefore required 997,139. Once again, the random sampling method found 25.4 percent invalid (74.60 percent were verified).
In other states, we see a potentially different picture. The last state to hold a gubernatorial recall, Wisconsin had a very different recall law. As Richard Winger of Ballot Access News and Christian Schneider point out, Wisconsin allows the signatures of any eligible voter in the state, not just any registered voter (as in other states). This makes signature rejection much less likely.
The petitioners in the Scott Walker recall garnered an enormous cushion. They handed in 931,053 signatures and they only needed 540,000. At the end of the day, 900,938 were found valid (4,001 duplicates, 26,114 struck out by staff). This is a 3.2% failure rate. For Lieutenant Governor Rebecca Kleefisch (who appeared on a separate petition), the failure rate was 4%.
The Wisconsin verification also appeared to be different and I am unsure if the fact that Governor Scott Walker effectively conceded the success of the signature gathering portion eliminated the challenges.
In the Senate races in Wisconsin in 2018 we saw some similar numbers. For the Senator Galloway petitions, there were 21,022 signatures handed in, and the GAB struck out 1,658. Galloway further challenged 863. The board did not bother looking at the challenged signatures. So the failure rate on the GAB was 7.8%. If we add in the Galloway protests, it is close to 12%. The Senator Wanggaard petitions are a little more complex, as he challenged many more signatures -- if all of his signature challenges had been approved, it would be about a 20% failure rate. From the board's review it was less than 3%. For Senator Moulton, the board threw out 5.7% of the signatures. His challenges would have pushed it up to about 11%. For Senator Fitzgerald, the board tossed out 4.1%. With his challenges, it would be just under 12%.
In the 2011 Wisconsin nine Senate recall races, we saw a higher rejection rate, but again it was not clear what was happening. Two petitions posted 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.
What about the other state legislative recalls of the last decade? We have four to choose from.
In Colorado in 2013, two State Senators lost recall elections. Senate President John Morse (D) saw a 37.5 percent invalidation rate. Colorado historically seemed to have a high invalidation rate, so this wasn’t a surprise. But the second recall was. State Senator Angela Giron (D) saw a 6 percent invalidation rate with only 818 signatures tossed out of 13,466. Petitioners used a new program that allowed them to check the signatures, which explained their accuracy.
In 2011, there were two other recall efforts, with very different verification rates. Arizona Senate Majority Leader Russell Pearce recall saw a 42% failure rate. (Arizona also has a historically high failure rate on the local level). On the other hand, the 2011 Michigan State Rep Paul Scott recall had a 9.4% failure rate.
A last example to consider: The 2011 Miami-Dade Mayoral recall, which collected over 100,000 signatures, saw a 16% failure rate.
One other factor to consider is claims of deliberate signature fraud. There are a few examples of signing fraud. Japan just saw a crazy version of this, with the lead petitioner hiring people to simply copy the names out of the registration book. In general, widespread signature fraud appears to be both fairly rare and not that difficult to uncover.
That’s not to say that they won’t be made. The claim of deliberate fraud can certainly ignite a party’s base and help muddy the waters against a recall effort. Much as in the last election, the Republicans banged this drum repeatedly in the Scott Walker recall campaign, where they claimed that a huge proportion of the signatures would be fake – the so-called Mickey Mouse signature. However, elections officials claimed to have found only 5 fake names in the Walker petitions: Adolf Hitler, Mick E. Mous, Donald L. Duck, Fungky Van Den Elzen, and I Love Scott Walker Thanks.
The Davis campaign seems to have had a similar argument, though once again it was the Republican side that made the claim. One of the companies that said they led the petition gathering efforts claimed that Davis forces ran a "blocking" campaign by hiring all the good signature gathering companies. They also intimated that Democrats were planning on submitting deliberately false signatures, though that seems unlikely (and crazy -- more likely than not, if the verification process missed a fake signature, it would have been included in the count).
One last issue --. Los Angeles requires that anyone who wants to remove their signatures from the petition does it the day before the signatures are handed in. This is different than the state law, which gives 30 business days to fight fire with fire and collect signatures from signers of the petition asking to have their names removed. This law was adopted during the Newman recall and did not work for him, but it did help out a Newport Beach Councilman. Nevada also saw uses of a similar provision. As we can see, it is unlikely to make a difference here.