California Ballot Polls Miss Mark

For California voters, the numbers released by polling firms in the final days before an election often do not match the final count on the ballot box. A review of recent tax measures reveals a consistent gap between predicted and actual results. In several high-profile votes, the data failed to capture the true level of support for specific fiscal policies.

The most dramatic recent discrepancy occurred with Proposition 218 in 1996. This measure, sponsored by the Howard Jarvis Taxpayers Association, aimed to change local tax rules. Final Field Polls released before the November 5, 1996, vote showed the measure failing badly. The survey reported that only 36% of likely voters supported the proposal. The actual outcome was the opposite. Proposition 218 passed with 56.55% of the vote, a swing of nearly 21 points from the last major poll.

A similar pattern emerged with Proposition 5, a recent legislative amendment. This measure sought to lower the vote threshold for some local bonds from two-thirds to 55%. The Public Policy Institute of California conducted a poll just before the election. That survey found likely voters split evenly, with 49% supporting the change and 50% opposing it. The final results showed a clear defeat for the proposal, which lost by a margin of over 55% to 45%. This represented a polling error of approximately 5 percentage points.

The history of these measurement errors goes back to Proposition 13 in 1978. Early polls in that cycle showed deep division, with support around 25% and opposition at 27%. By spring, support had risen significantly. Even the last Field Poll before the June 6 primary understated the final result. It projected passage by a margin of 57% to 60%. The actual vote was 64.8% in favor.

These examples suggest a specific trend in California politics. Discrepancies between polling data and final results tend to undercount interest in taxpayer issues. Jon Coupal, president of the Howard Jarvis Taxpayers Association, argues that this gap may reflect how voters interact with surveys. He suggests that many individuals who hold fiscal conservative views either do not respond to polls or prefer to keep those views private. This behavior can skew the data provided by polling firms.

The reliability of polls has become a point of contention in the current election cycle. The perception that polls are used to influence outcomes rather than measure sentiment has grown. This skepticism was fueled recently by the Wedgewood Polls scandal. That firm admitted its data was entirely fabricated. Despite lacking formal credentials or real data, Wedgewood was added to major aggregators like The New York Times, FiftyPlusOne, and VoteHub. The inclusion of bad data in these aggregators skews the overall picture presented to the public.

Respected pollsters with proven track records also face challenges. In 2024, a Des Moines Register poll by J. Ann Selzer showed Kamala Harris leading Donald Trump in Iowa by three points. Trump won the state by 13.2 points. The error led to Selzer's retirement and damaged the reputation of the polling organization. These incidents contribute to a climate where voters question the utility of pre-election data.

The Howard Jarvis Taxpayers Association notes that polls remain indispensable for testing messages and assessing arguments. However, the historical data on ballot measures indicates that the final count often differs from the last published survey. The gap between the predicted and actual results for tax reform measures suggests that polling may not fully capture the strength of certain voter groups.