Here Are The Results of SurveyUSA Election Poll #10559

Geography Surveyed: State of Washington
Data Collected: 10/27/2006 - 10/29/2006
Release Date: 10/30/2006 2:50 PM ET
Sponsor: KING-TV Seattle

Cantwell Re-elected, WA Senate Seat Stays Democrat: In an election today for U.S. Senator from the state of Washington, Democratic incumbent Maria Cantwell defeats Republican challenger Mike McGavick, according to a SurveyUSA poll of 774 likely state of WA voters conducted exclusively for KING-TV Seattle. Today, it's Cantwell 51%, McGavick 43%. Compared to an identical SurveyUSA poll released 2 weeks ago, on 10/16/06, Cantwell is up 3 points, McGavick is down 2 points. Cantwell had led by 8, now leads by 13. Among women voters, Cantwell has gone from a 16-point lead to a 30-point lead, a 14-point improvement. Among Independents, Cantwell had led by 1, now leads by 15, a 14-point improvement. The last day to vote is 11/7/06.

WA Initiative 920 Now Trails No Matter How You Slice It: Today, 10/30/06, Initative 920 fails by 17 points when voters are not read the full ballot language and fails by 11 points when voters are read the full ballot language, according to a SurveyUSA poll conducted exclusively for KING-TV Seattle. Respondents were divided into 2 groups; each group heard the ballot question read differently. When voters are asked about 920, but are not read the full text, 37% say they are "Not Certain" how they will vote. 2 weeks ago, 68% were Not Certain how they would vote. Of those who are certain how they will vote on 920, the measure fails, 40% to 23%. When the actual ballot language is read to voters, 40% today say they will vote "Yes," 51% say they will vote "No." Compared to an identical SurveyUSA poll released 10/16/06, "Yes" is down 5 points, "No" is up 11 points. 920 had been favored by 5 points, now trails by 11 points, a 16-point swing, and the difference between Pass and Fail.

WA Initiative 933 Losing support, Still Too-Close-To-Call: Today, 10/30/06, Initative 933 trails by 24 points when voters are not the full ballot text and fails by 4 points when voters are read the full ballot text, according to a SurveyUSA poll conducted exclusively for KING-TV Seattle. Respondents were divided into 2 groups; each group heard the ballot question read differently. When voters are asked about 933, but are not read the full text, 33% say they are "Not Certain" how they will vote. 2 weeks ago, 53% were Not Certain how they would vote. Of those who are certain how they will vote on 933, the measure fails, 46% to 22%. When the actual ballot language is read to voters, 41% today say they will vote "Yes," 45% say they will vote "No." Compared to an identical SurveyUSA poll released 10/16/06, "Yes" is down 2 points, "No" is up 5 points. 933 had been favored by 3 points, now trails by 4 points, a 7-point swing to the opposition, and possibly, but not certainly, enough to make the difference between passage and defeat.

WA Supreme Court Contest Still Too-Close-To-Call With 33% Undecided: In an Election for the Supreme Court of the state of WA today, 10/30/06, Stephen Johnson and Susan Owens are effectively tied, according to a SurveyUSA poll conducted exclusively for KING-TV Seattle. Owens gets 36% today, up 5 points in the past 2 weeks, Johnson gets 31%, up 2 points. With 33% of Likely Voters undecided today, any outcome is possible.

Purpose of This Parallel Test: In a study presented by SurveyUSA to the American Association for Public Opinion Research, SurveyUSA found that all pollsters have a difficult time forecasting the outcome of a ballot measure. When 446 statewide polls from 104 separate polling organizations were analyzed, the average error on a statewide contest (such as a contest for Governor or U.S. Senator) was 4.6 percentage points, for all pollsters. But: the average pollster error on a ballot measure was twice that, 9.2 percentage points, for all pollsters. Telephone pollsters struggle with ballot measures, since words are read into the ear of a poll respondent, but voters see words printed on a printed or electronic ballot. Since SurveyUSA completed the study, SurveyUSA has experimented with different ways of presenting ballot measures to poll respondents. Some ballot measures are so long-winded, and/or are written so obtusely, that reading them to poll respondents over the telephone is almost impossible. Yet, summarizing the ballot measure, no matter how careful the pollster is, may introduce the possibility of bias, since the pollster may leave out concepts or words that are key to how the respondent makes up his/her mind. In the parallel test presented here, SurveyUSA compares the effect of reading to poll respondents no summary whatsoever, with the effect of reading to respondents the full ballot text. The two approaches, at this point in the campaign, 8 days to Election Day, produce results that have converged compared with the previous poll 2 weeks ago, but still differ enough that the outcomes cannot be projected with full confidence. Important reminder: in Washington State, unlike many states, many voters vote before Election Day. These "vote by mail" voters have the ballot in their hands for potentially days or weeks before marking it. This may give Washington state voters an advantage over voters in other states, who see the actual ballot text only when they walk into the voting booth, and who may feel obligated to skim the language quickly because of anxiety over the long line of voters behind them.

Filtering: 1,200 adults from the state of Washington were interviewed 10/27/06 through 10/29/06. Of them, 997 were registered to vote. Of them, 774 were identified as Likely Voters for the U.S. Senate and WA Supreme Court contests. For ballot questions 920 and 933, Likely Voters were divided at random into two groups. One group was read abbreviated text. A second group was read the full ballot text. For 920, the following 16 words were added by SurveyUSA after the full ballot text was read: "A Yes vote would repeal the tax. A No vote would leave the tax in place."