Report #11 on the Six Californias Signature Verification Process

Good news! Tuesday’s random sample update includes raw signature counts from Amador and Trinity counties! Amador reports 1,750 raw signatures, and Trinity reports 779 raw signatures. That would have brought the total raw signature count to 1,137,875, except that Kern County, which finished its sampling, reduced its raw count from 26,444 to 26,422, and Humboldt County, which hasn’t finished its sampling, reduced its raw count from 7,280 to 7,230. That brings the total raw signature count to 1,137,803. (I was going to say “final total raw signature count”, except as we have seen a county might change its report of raw signatures when it completes its sampling.)

In addition to Kern (validity rate of 74.9%), three other counties have also completed their random sampling. They are San Bernardino (61.4%), San Mateo (68.0%), and Siskiyou (71.2%). That brings the overall validity rate to 66.9%, down from the 67.5% in my previous report, and gives an overall estimate of 761,190 raw signatures, well below the 767,235 needed to qualify for a full count.

There are still 21 counties that need to finish their sampling (including Amador and Trinity). With both San Bernardino (it had been ranked second) and Kern (it had been ranked seventh) having reported, the top ten are now Los Angeles (311,924 raw signatures), Riverside (74,478), Orange (52,217), Alameda (51,366), Fresno (38,382), San Luis Obispo (12,906), El Dorado (11,649), Humboldt (7,230), Tehama (4,855), and Toulomne (4,732).

Updated September 3rd: In my original posting I listed Kern in the top ten unreported counties even though it had finished. I have corrected that above by removing Kern and adding Toulomne.

–Steve Chessin

President, Californians for Electoral Reform (CfER)

The opinions expressed here are my own and not necessarily those of CfER.

One thought on “Report #11 on the Six Californias Signature Verification Process”

  1. The estimates continue to straddle success and failure.

    The estimated simple validity rate (ignoring duplicates) is 71.9%.  This would project to 818,585, which would barely qualify the measure.

    The estimated validity rate with duplicates is 67.0% which projects to 762,044, which is slightly below the trigger for a full count.  This indicates a 5.0% (4.97%) duplicate rate.

    Sample (or full) counts have now been been processed in counties representing 53.1% of the vote.  This would ordinarily mean that we are zeroing in on a final result.  But over half of the remaining raw signatures come from Los Angeles.  There is no reason that we should expect Los Angeles County to be typical in any sense, or even like neighbors such as Kern, San Bernardino, and Ventura which have completed their counts.

    Los Angeles has horrible voter turnout.  But does that mean that it is hard to collect signatures, or hard to collect valid signatures?  The SOS had to wait until Los Angeles returned its raw count before the sample count was triggered.  I suspect that its sample of 9000+ is large enough and its election staff large enough, that it can dedicate individuals to checking signatures over the maximum period of time allowed.  

    The smaller counties that are still out had high signature rates relative to their population, and possibly higher support for Jefferson.  They probably have one person checking when they can break away from other duties (or perhaps paying overtime to work when they aren’t open to the public).  Corporate-sponsored initiatives may not bother with smaller counties, because collection of signatures is less efficient there.

    If there were a 5.0% duplicate rate in a county, if the number of signatures is between 12,500 and 22,222; it would be expected that less than one duplicate would be detected in the sample.  If one were detected, an estimate of greater than 5% would be made.  If two were detected, an estimate of over 10% would be made.   And if none were detected, they would estimate no duplicates.

    What if the sampling methodology was based on statutory initiatives, and when California had fewer voters?  There might be fewer counties in this zone of statistically-suspect estimates.  A few bad estimates is not going to deny a full count.

Comments are closed.