Statistically, the Gascón Recall Will Likely Fail to Qualify

July 19, 2022

Dear Editor,

Residents, business owners, employees, and City officials who are concerned about crime in Santa Monica should not count on the George Gascón Recall to reduce it.

Because based on standard statistical analysis, the recently released signature verification numbers say the Gascón recall will very likely fail to qualify for the ballot.

As background, at this stage of the recall process, the LA County Registrar-Recorder examines a large random sample of 5 percent of the total signatures collected, in this case examining 35,793 signatures.

According to the Registrar’s office, that examination revealed 27,983 “valid” signatures (i.e., people registered to vote in LA County, non-duplicate signatures, etc.). The result from tabulating this sample was high enough to merit a full and final examination of all 715,833 signatures submitted, which must be completed by August 17.

But was the sample result high enough to forecast that the full, final examination will reveal enough valid signatures for the recall to qualify for the ballot? In a word: No.

With 715,833 signatures submitted, a total of 566,857 signatures need to be deemed “valid” in order for the recall to qualify for the ballot, a percentage of slightly over 79 percent (79.19 percent, for the detail obsessed).

The large sample of 35,793 signatures yielded just over 78 percent (78.18 percent) deemed valid, and the familiar “plus or minus margin of error” in statistical calculations indicates it is very unlikely the 79 percent-plus threshold will be attained by examining all signatures.

Based on the results of this large sample, and using the typical statistical “95 percent confidence level,” the percentage of valid signatures in the full, final count will most likely range between 77.8 percent and 78.6 percent (i.e., “plus or minus” a mere four tenths of one percent), short of the 79 percent-plus needed.

Moreover, if the statistical “confidence level” is expanded to 99 percent, the range of percentages reaches as high as 78.7 percent, still short of the 79 percent-plus needed.

Put another way, only 1 in 100 random samples as large as the one examined would produce the result the Registrar obtained if, in fact, the total signatures submitted were sufficient for the measure to qualify.

Bottom line: If the Registrar’s 5 percent sample was representative, standard statistics indicate the final result will most likely fail to qualify by several thousand valid signatures.

So . . . does all this fancy number-talk mean the recall will definitely fail to qualify for the ballot? Of course not.

The final results will be released by August 17, so we won’t know how accurate the Registrar’s sample was until then. (For technical reasons I won’t bore anyone with, there are a number of ways statistical samples sometimes turn out to be unrepresentative.)

The only way to be 100 percent confident of the number of valid signatures is for the Registrar-Recorder to examine them all, a process now underway.

And as a wry pundit once observed, if statistical projections were perfect, Brexit wouldn’t have happened and Hilary Clinton would have defeated Donald Trump.

Finally, let me add that I am not now and never have been a professional statistician. After studying basic and advanced statistics in undergraduate and graduate school, I applied that learning to a variety of projects throughout my business career and during a short stint in political polling.

And from those experiences, I can confirm that the wry pundit’s cynical observation about statistical projections is fully warranted.

Despite the caveats, however, the recall’s ballot qualification chances can be translated into clear number-talk: don’t count on it.


Peter DiChellis - Unaffiliated moderate
Santa Monica - 90403

PS -- I tried to present, as best I could, an objective and politically neutral analysis of publicly reported signature verification data. Having done that, I will disclose for the record that I strongly favor recalling DA Gascón.

Back to Lookout News Santa Monica Lookout is owned by Copyright 1999-2022 . All Rights Reserved. EMAIL Disclosures