Blog Post < Previous | Next >
Guy
Updated California Props Probabilities of Passing
The PPIC conducted two polls regarding the California Propositions on the November ballot. The first one was in September (August 29 - September 9), the second one in October (October 7 - 15).
I reviewed the PPIC September polls in my earlier article. This follow-up article reviews the October ones.
As a reminder the table below describes what the California Propositions are about.
Upfront caveat
The math approach I use to figure out the Props probabilities of passing is too deterministic. Thus, one should take
the probabilities I generate with a grain of salt. On the other hand,
these probabilities may be more informative than the customary polls
disclosures. The latter states that if the % Yes and % Nos are less than 6%
apart; it is too close to call because it is within the error margin
(based on a sample of 1,000 and an error margin of + or - 3%). So more often than not, pollsters' disclosures tell you very little.
For more on the subject, please refer to my earlier article.
Math reference
In order to figure out the Props probabilities, I use the same math framework as in my earlier article. I will use the latter as a reference for the reader who wants more info on the math.
Figuring out 95% Confidence Intervals
Figuring out probabilities of passing
As shown below, the math approach renders probabilities very sensitive to poll levels. Thus, while a Prop that polls at 50% does have a 50% probability of passing, a Prop that polls at 49% is already 2/3d of a standard deviation below the 50% threshold. This corresponds to only a 25% probability of passing. We will illustrate this situation when we focus on Prop 5.
Comparing the September vs the October polls and probabilities
When we adjust the Yes % by taking out the Undecided % so the Yes % and the Nos % add up to 100%, the adjusted Yes % for several of the Props dropped between September and October. As shown below, the black columns for October are often a bit shorter than the green ones for September.
The drop in the adjusted Yes % resulted in a drop in the Props probabilities of passing. For Props 6, 32, 33 the drop in probabilities was massive. As of this October poll time, these three Props appear very unlikely to pass.
As a reminder of what Props 6, 32, and 33 are see the table below.
An exception to the overall drop in polls and probabilities, support for Prop 36 (tougher on crime than Prop 47) remains strong. It even appeared to have strengthened a bit between September and October. This one is pretty sure to pass.
A closer look at Prop 5
Let's have a closer look at Prop 5. That is the one that would lower the passing threshold from 2/3d to 55% for local bonds. According to the October poll, its most likely outcome is to garner 49% of the votes. And, it is associated with a 95% Confidence Interval ranging from 46% to 51.9%.
When we convert the above into probabilities of passing, the 49% is close to 0.7 standard deviations below the passing threshold of 50%. And, it is associated with a 24.8% surface area under the overall Normal distribution curve (shown below). This 24.8% surface area translates into a probability of passing. Notice that it is much below the 37.1% probability of passing in the September polling.
As a reminder my math framework is too deterministic. It does capture the uncertainty associated with the quantitative small sample error. But, it does not capture the qualitative flaws in the structure of the polls' weighting of various voters categories that may differ substantially from the actual voters' mix by November 5th.
In view of the above, if there is a 25% chance of rain, you probably should take your umbrella. I kind of think of the Prop 5 probabilities in a similar way. Its outcome is still a bit on a razor's edge.
The situation is quite different than for Props 6, 32, and 33. I am quite confident that these three won't make it. In this case, if any of them come to pass I would view that as a genuine upset.
Consideration
Several of the California Propositions' outcomes will have a material influence on our welfare and lifestyle. That's something to keep in mind. State and local laws and regulations have often as much if not more impact on our lives than what is our Federal Government's party-composition.
THE END
Gaetan Lion is an independent researcher with expertise in economics, econometrics, investments, risk management, statistics, and quantitative analysis. His areas of interest include demographics, economics, investments, Climate Change, health care, sports, and politics.