What's Next?: The California Governor's Recall and Its Polling Dilemmas

By Mark Baldassare

As the most populous state in the nation, California has more than its fair share of news events that capture national and even worldwide interest in the state's public opinion trends. In past years, for instance, I have been asked to monitor the public's attitudes in the aftermath of natural and manmade disasters such as the San Francisco earthquake, Los Angeles riots, the Orange County bankruptcy, and California's electricity blackouts. In the political arena, while California has become a predictably "blue state" in national elections, there is still intense media interest in the state's voters because of their penchant to use the initiative process for making sweeping changes and cutting-edge laws at the ballot box. In our past polling, ballot initiatives have led the Public Policy Institute of California (PPIC) to focus on such controversial issues as affirmative action programs, bilingual education programs, legislative term limits, public services for undocumented immigrants, same-sex marriages, school vouchers, stem-cell research, and tax limitations.

Recently, California pollsters were faced with arguably the most significant and unexpected event in the state's political history: a governor's recall from office. I offer the following recollections about the methodological issues faced by PPIC during the California recall process in 2003 and the polling decisions we made along the way, as well as current observations and recent data on the changing political landscape.

When the rumors of a recall signature drive began to surface after the November 2002 governor's election, PPIC's attention focused on a little-known provision in the 1911 state law that created the three tools of modern direct democracy in California. The initiative, the referendum, and the recall were enacted so that the state's voters could use petitions to create new legislation, repeal actions passed by the legislature, and recall judges and other state and local elected officials. Nationwide, eighteen states, including California, currently have recall provisions, but they are rarely used. Citizens of Oregon, among the most frequent users of the recall drive, have ejected six county officials, but none from statewide office. Prior to the 2003 California election, the only successful recall of a governor was the 1921 removal of Lynn Frazier in North Dakota.

Indeed, the recall was more a threat than a reality in California politics-it was used only three times in the first eighty-three years after its passage. The first modern attempts were in 1994 against two assemblymen and one state senator. Of the thirty-two efforts to recall a California governor, the removal of Governor Gray Davis was the first to have gone to a vote of the people.

This almost complete lack of precedent in actually removing a governor in any state through a petition drive and recall election presented a number of challenges to California pollsters. With no archive to consult for past data, there was no opportunity to look at how others had dealt with designing effective questions, trends over time in tracking surveys, or the relationship between disapproval and recall answers in past surveys. We used focus groups and extensive pretesting to fill in the information gaps.

On the other hand, there was entirely too much precedent for attempts at recall. The fact that every modern governor in California had been confronted with a recall drive raised critical questions about the implications of polling, and when and even if it should take place. Were we sounding an alarm bell for an event that was unlikely to happen? Would our polling raise the profile and validity of the petition effort underway? Was polling on the recall worth the cost, given competing events and interests for surveys?

The recall petition was certified for circulation on March 25, meaning that the recall proponents had until September 2 to collect the nearly nine hundred thousand signatures needed to qualify for the ballot. The early reports of actual success at signature-gathering were sketchy at best. We decided to pose the first recall question in a PPIC Statewide Survey in June 2003, when the signature-gatherers seemed to have a realistic possibility (though by no means a certainty) of qualifying a measure for the statewide ballot. At the time, 51 percent of voters told us they wanted to remove Governor Davis from office. Still, a host of important details about the recall had yet to be determined. Who, for instance, would be the candidates to take Davis's place if he were removed?

The recall ballot is actually a two-part measure-a yes or no vote determines whether or not the current officeholder should be removed, and a second vote decides who should replace the officeholder if the recall succeeds-so once we decided to go ahead with polling, there were actually two separate but closely related issues to be considered. Who would the replacement candidates be? And how much might support for the governor's recall change once voters were confronted with the potential replacements? To what degree would response to a recall question be simply a reflection of a highly unpopular governor rather than a true desire to make such a radical change through a never-used voting process? It was not until July 24 that Lieutenant Governor Cruz Bustamante's proclamation for a recall election set in motion the declaration of candidacy for individuals seeking to place their names on the second part of the ballot, and that process was not finalized until August 13, so there was plenty of time for opinions to change.

The problem of monitoring opinion on the removal itself was fairly straightforward: We made sure to include questions on the governor's performance and right/wrong direction and to look at the trends in responses and correlations through further data analysis. In our July poll, the job disapproval ratings for Governor Davis among registered voters was at 69 percent, and, similar to June, 50 percent said they would vote yes to remove him from office. The issue for us was whether such voting trends would persist in light of specific details on the recall and replacement candidates.