Stanford University Home

Stanford News Archive

Stanford Report, September 24, 2003

Schwarzenegger leads the pack in online recall surveys


Two recent Stanford online surveys of California voters show GOP candidate Arnold Schwarzenegger beating Democratic Lt. Gov. Cruz Bustamante in the Oct. 7 election that could replace Gov. Gray Davis.

Unlike conventional telephone polls, which have reported Bustamante narrowly leading the recall race, Californians who participated in the online surveys voted 40 percent for Schwarzenegger during two polls held between Aug. 29 and Sept. 8, and between Sept. 11 and 21.

Although Bustamante continues to trail Schwarzenegger, support for his candidacy increased from 28 to 32 percent between the dates of the two surveys. One significant shift, however, came from registered Hispanic voters whose support for Bustamante jumped from 40 to 63 percent during the two-week interval. At the same time, Hispanic support for Schwarzenegger dropped from 37 to 13 percent.

"This is a spectacular change," said Shanto Iyengar, chair of the Department of Communication. Despite this development, he said, movie star Schwarzenegger's name recognition remains a powerful polling card. "What this shows is that undecided voters gravitate to the most visible candidate, the candidate with name recognition," he said.

Meanwhile, the two polls showed little change in voter support for the recall itself. Among registered voters, 62 percent indicated "yes" for the recall and 38 percent said "no" in the first survey. In the second poll, 61 percent of participants voted "yes" and 39 percent voted "no." Poll results have a 4.3 percent margin of error.

Iyengar and political science Professors David Brady and Morris Fiorina sponsored the poll to learn more about the logic behind voters' decision-making leading up to the special election. Brady and Morris are also fellows at the Hoover Institution.

Knowledge Networks, a Menlo Park-based research firm, surveyed a representative sample of 852 adult Californians, 528 of whom are likely voters, in the first poll. Preliminary results from the second poll of 519 respondents released Sept. 22 reflect interviews with 391 registered voters only. According to Iyengar, the smaller number of participants in the second poll was due to budget restrictions but is still considered scientifically representative. A third poll is scheduled to be held just before the Oct. 7 ballot, he said.

Unlike conventional polls, which restrict respondents' choices to the top candidates and give them the option of selecting 'undecided' as a response, the online surveys gave participants the actual ballot that included the question about recalling Davis and listed the 135 replacement candidates. Fiorina said this scenario is closer to the actual situation voters will face.

"What this shows is that when voters have to choose, they're voting for Schwarzenegger," he said. "I won't say he'll win -- the situation is fluid -- but Schwarzenegger has more upside potential. The undecided vote won't break 50-50."

Mike Dennis, a vice president at Knowledge Networks, agreed that the two results reaffirm the finding that Schwarzenegger stands out on a ballot crowded with many unfamiliar names. "Polls conducted on the telephone might be underestimating the depth of Schwarzenegger's support among voters," he said.

According to results from the first poll, support for the recall was strongly associated with political party affiliation -- 91 percent of Republicans and 42 percent of Democrats "voted" to recall Davis. Only 3 percent of Republicans chose Bustamante but 19 percent of Democrats selected Schwarzenegger.

On Sept. 22, preliminary results showed that support for the recall had slipped to 86 percent among Republicans, and to 38 percent among Democrats. This change, however, was offset by increased support among independents for the recall -- a jump from 57 to 67 percent between the two polls.

Fiorina speculated that shift may come from independent voters opposed to Davis' recent decision to give illegal immigrants the right to apply for driver's licenses. "I think there's a backlash," he said.

Results from the first survey revealed that no other candidate received more than 10 percent of the "votes." Republican Sen. Tom McClintock received 8 percent; Peter Ueberroth, who is no longer in the race, garnered 7 percent; and independent Arianna Huffington received just 1 percent. Early results from the Sept. 22 poll did not include this breakdown.

Survey methodology

According to Dennis, the polls were conducted with a random sample of adults in California who are members of a national "web-enabled panel" created and maintained by Knowledge Networks that is designed to be representative of the U.S. population. Initially, participants were chosen scientifically by a random selection of household telephone numbers. Individuals in selected households were then asked by telephone to participate in the research panel. Those who agreed to participate were sent an Internet appliance and received an Internet service connection provided by Knowledge Networks. Some people who already had computers and Internet service were also permitted to participate using their own equipment, Dennis said. Panelists received unique log-in information for accessing surveys online, and then were sent e-mails three to four times a month inviting them to participate in research. For the recall surveys, Knowledge Networks randomly selected adult panelists in California.