Polling Critic #29 - California Inherits Florida Mantle

The parallel between Florida 2000 and California 2003 elections is closer than you think, but the outcome may be entirely different.

In polling, as we have seen in many Polling Critic columns, it is very important to ask the right questions the right way in order to get useful, meaningful and accurate understanding of public opinion. In elections, the questionnaire is called a ballot, but it is still about asking questions the right way. The basic candidate-preference question must be structured to determine the voter's preferences for various offices. While there are different ways of wording to determine these preferences, these differences generally do not matter -- until the California recall election. There it made all the difference. Based on the vote totals on October 9, 2003, with 100% of precincts reporting, it is extremely likely that more voters wanted Gray Davis to remain governor than wanted Arnold Schwarzenegger to become governor. Recall results reported by the California Secretary of State are the basis of this conclusion.

Part 1 of ballot

Vote Number of Votes Percent of votes

Yes 4,415,341
55.4% for recall

No 3,559,400 44.6% want Davis to remain governor

7,974,741

Part 2 of ballot

Percent of

Rank Candidate Party Number
of Votes total votes major Rep. votes

(1) Arnold Schwarzenegger Rep. 3,743,393 48.7 78.5

(2) Cruz M. Bustamente Dem. 2,432,446 31.6 0

(3) Tom McClintock Rep. 1,026,481 13.4 21.5

(4) All others together Mixed 484,254* 6.3 0

Totals: (1)+(2)+(3)+(4)
7,686,574

100.0
100.0

Any of the 3,559,000 voters whose "no" vote meant that they wanted Davis to remain as governor, and then went on to the next part of the ballot with a vote for Schwarzenegger as their second choice, meant to apply only if Davis did not prevail by getting over 50% of the first vote. Whether Davis would prevail was, of course, unknown to voters until after vote-counts were announced. But the desires and intentions of those voting first "no" and then "Schwarzenegger" in the two successive votes are perfectly clear. Their first choice for governor was Davis.

A judicial review of an election count for appropriately counting any ballot must rest on this principle: whenever it is clear what the intention of the voter was, then the count must reflect the intention. In Florida's 2000 election, dimpled chads and hanging chads played an essential role in recounts. Voter intentions often could be discerned to the satisfaction of both Democratic and Republican observers by no other means.

If the number who voted in this way (first "no", then "Schwarzenegger") was 183,994 or greater then, counting only the primary intention of their votes, Schwarzenegger would have fewer votes than Davis and any fair-minded observer would have to say that Davis was more favored than Schwarzenegger as governor of California. An analysis of the data in the Table shows that this is almost certainly the case. It should be noted that the "magic" number, 183,994, (a) depends on the final official count before any recount, (b) has changed little since October 9^th and (c) its exact value is immaterial.

Even if hypothetically the 2,916,700 voters (which were all of the Democrat and minor candidate votes) and the 288,167 voters (which were all voters in part 1, both "yes" and "no", who dropped out and did not vote in part 2), had been "no" voters, then there still must be 354,533 "no" voters that switched to two Republicans, Schwarzenegger or McClintock. Why? Well, 2,916,700+288,167+354,533 equals all of the votes that favored Davis, the "no's". Realistically, it is likely that roughly as many "yes" voters as "no" voters dropped out of voting in part 2 and some certainly distributed their votes among minor candidates and Bustamente, so that, as a practical matter, the number of "no" voters that switched to Republicans was substantially higher than 354,533, making it even more likely that Davis was favored over Schwarzenegger, than shown by the following analysis, based only on there being 354,533 "no" voters switching to one of the two top Republicans.

If fewer than the "magic"183,994 (51.9% of 354,533) of Schwarzenegger's votes came from "no" voter switches then all the remaining "no" voters, 48.1% or more, must have gone to McClintock, and it is precisely in this case that Schwarzenegger loses the election. The 78.5% to 21.5% ratio of votes the two leading Republicans received (See Table) would have had to shift to at least to 51.9% to 48.1% in order for Schwarzenegger not to lose the election. There is no reason to expect such a large shift, making virtually certain that more voters wanted Gray Davis as governor than Arnold Schwarzenegger. The two-part form of the recall ballot is to blame for this confusion, just as the confusion of the butterfly ballot contributed to the failure of the Florida election.

Furthermore, as in Florida 2000, this conclusion need not remain hypothetical. Although Al Gore had long conceded the election to George W Bush, a non-binding recount was later conducted by a fair and balanced consortium of media and university political scientists showing that if the Florida recount had been completed in all Florida counties, Gore would have been the winner. The right recount would have changed the outcome.

That such a recount will take place in California now seems almost certain. Though desirable, not every precinct need be re-counted. No matter which or how many precincts have been recounted in California, unlike in Florida, as soon as the "magic"183,994 ballots are found with a "no" vote in part 1 along with a vote for Schwarzenegger, then the case for a Davis win is made. But like Florida, the proof might occur too late to change the outcome, even though Davis remains governor for a while longer.

The analogy here is that in the Florida 2000 election, fearing the potential viciousness of the media, not one Democratic senator dared to force the election into the House of Representatives by the simple act of challenging the Electoral College vote-count, which was about to throw the election to Bush by one electoral vote.

In California, the media have overwhelmed responsible government officials and turned politics into entertainment. The media have persuaded the country that Arnold as governor represents the voice of the people and will not easily reverse their position. Legal challenges will face not only political but media opposition, both backed by big money and potential vicious treatment of individual reputations.

A small error in 1911, pointed out before the election in my Sept 30^th Polling Critic column #27, California Fiasco Setting Democracy Back a Century, produced a flawed election procedure, that might not affect many elections. It is staggering that nobody noticed the flaw in the intervening 92 years and that the very first attempt to unseat a governor found that the flaw had erupted and made a mockery of majority rule, the essence of democracy.

There have been globally tens of thousands of editorials, syndicated columns, and talk show pundits in the weeks following the election who have pontificated on the meaning and significance of Arnold becoming governor, including intriguing and complex variations on implications for other states, the nation, and the world.

But sadly, not one of the thousands of media commentators seems to have done the math showing Gray Davis almost certainly defeated Arnold Schwarzenegger.

After October 8^th I promised to come back and add a sentence or two to the Sept 30^th column, expecting to confirm my predictions made then. On September 30^th the polls did not show Arnold in the commanding lead that occurred after the first debate, the only debate that Arnold appeared in. His numbers soared and Davis' waned in the last week. As this column shows, the outcome was much different and much more significant than the outcome I analyzed in that column.