Has the Govinator gotten his mojo back? That’s what a new poll released this week seems to suggest.
Three of Gov. Arnold Schwarzenegger’s four ballot initiatives appeared to surge in the latest statewide poll of likely voters for next week’s balloting in California. The encouraging results for the governor came just a week after another poll seemed to doom all three measures.
The two polls differ widely not only in their results but also in their methodologies. The poll that seemed to doom Schwarzenegger’s initiatives was conducted by telephone in a traditional manner. Its sponsor was the Public Policy Institute of California (PPIC). The more encouraging results were drawn from an online, or Web-based, survey sponsored by a consortium led by the prestigious Hoover Institution at Stanford University.
Hoover Senior Fellows David Brady and Morris Fiorina, two of the brightest minds in contemporary American political science, have teamed with Knowledge Networks, the progenitors of perhaps the best Web-enabled panel of respondents available today. Unlike other Internet panels that rely on volunteer samples of existing Web users, Knowledge Networks first selected a household sample and then placed Internet appliances in the chosen households, largely eliminating the potential for error inherent in other Web surveys that exclude a sizable slice of the population that’s not on the Net.
But the most interesting divide between the two studies is not the samples. No, it’s a matter of how the questions are presented to voters that’s intriguing. The PPIC poll, because it was conducted by telephone, necessarily had to read voters a paraphrased version of the four ballot initiatives. By comparison, the Hoover poll was able to show respondents the measures exactly as voters will see them on their Nov. 8 ballots. Hearing the ballot summaries and reading them is an entirely different experience. And the difference would seem to favor the Web version.
Take Proposition 76, the measure that mandates changes in the state budget process. Its official summary, approved by California’s attorney general, includes 89 words. But these are broken down into eight bulleted paragraphs to facilitate comprehension by the reader. The longest of these paragraphs is 17 words; half of the bullets contain fewer than 10 words. The result is that the measure is eminently readable on the printed page or computer screen. By comparison, PPIC’s 100-plus-word phone-survey question regarding Prop 76 is hard to read (even on paper) and even harder to listen to. It’s just a blob of ink and sound. There’s no way to bullet blobs of words.
Hoover’s Internet surveys are superior to the telephone polls in other ways, too. Web-based surveys, like a real ballot, allow the voter to take a moment to read, think and then re-read any part of the printed summary that raises concerns. The audible survey doesn’t facilitate that sort of reexamination. Sure, a respondent can ask an interviewer to repeat the question to try and clear up a small concern, but because the entire question must be re-read, the original phrase of concern may again be heard awkwardly, not resolving the respondent’s concerns.
Another advantage for Hoover’s methodology is that it doesn’t provide for refusals or undecided voters. As such, it’s closer to a real ballot. Brady told me this week that he believes this twist may account for some of the variance between his survey and PPIC’s. If he’s right, that means most undecided voters on the California initiatives will vote yes when the time comes. I’m not entirely convinced that his expectation is necessarily valid, but it bears watching as other polls are conducted and the votes are counted.
Hoover’s polling may reflect more than methodological differences from the earlier survey. It may also show real longitudinal progress by Schwarzenegger’s campaign. The Hoover poll finished surveying 12 days after PPIC started its interviewing. Those 12 days of campaigning by the governor may have pushed several of his measures to victory.
Hill is director of Hill Research Consultants, a Texas-based firm that has polled for GOP candidates and causes since 1988.