Includes Interviews with Cell Phone Respondents: For this poll of Washington state, SurveyUSA completed interviews with 492 suspected cellphone respondents and completed interviews with a separate, larger group of landline respondents. Of the 492 suspected cell respondents, 6 (1%) were landlines, not cellphones; 32 (7%) were cellphone respondents under age 18, who were excluded; 218 (44%) were cellphone respondents who also had a home telephone, and who were counted toward, and included among, the sample of landline respondents; leaving 236 (48%) suspected cellphone respondents who did not have a home telephone and who qualified as CPO (Cell Phone Only). Of the 236 adult CPO respondents, 177 (75%) were registered to vote in the state of Washington, and of the registered voters, 120 (51% of CPO adults; 68% of CPO registered voters), were determined by SurveyUSA to be likely to vote in the 08/17/10 primary. Context: A "landline" phone is a traditional home telephone. A "landline respondent" is someone reached on their home telephone. A "cellphone" respondent can be either a cellphone respondent who also has a home "landline" phone, or a cellphone respondent who does not have a "home" landline phone. A cellphone respondent without a home landline phone is here labeled a CPO (Cell Phone Only) respondent. A CPO respondent is of unique interest to survey research professionals because: a) the population of CPO adults is growing; b) there are restrictions on how CPO respondents can be contacted, and c) it is unclear to researchers if and to what extent CPO respondents differ from landline respondents. In this survey, 62% of CPO respondents were under age 35, compared to 17% of landline respondents who were under age 35.

But How Many CPO Respondents to Include? Separate entirely from the challenge of reaching CPO respondents at a time when they can be interviewed without distraction, is the challenge of knowing how many CPO respondents to include in a given group of respondents. 25% of Americans today are CPO, according to the best government estimates. But that percentage is not flat across states: Is WA 20% CPO? 25%? 30%? There is no current state-level data. For this report, SurveyUSA blends CPO respondents into this sample as 25% of total WA adults. This results in CPO respondents being 23% of registered voters and 23% of likely voters. Had SurveyUSA blended CPOs differently, here's what would have happened:

* At 30% CPO, Murray would lead Rossi narrowly, 37% to 32%, within the survey's theoretical margin of sampling error.
* At 20% CPO, Murray would lead Rossi narrowly, 37% to 33%, within the survey's theoretical margin of sampling error.
* At zero percent CPO, Murray would lead Rossi narrowly, 39% to 34%, within the survey's theoretical margin of sampling error.

To ensure that SurveyUSA could run a number of hypothetical scenarios combining landline and cellphone respondents in different ratios, landline respondents were oversampled.

Field Period: Interviews for this survey were conducted 06/25/10 through 06/28/10. Cellphone numbers were dialed one at a time, cellphone respondents were interviewed by call center employees. Landline respondents heard the recorded voice of a SurveyUSA professional announcer. Questions about the methodology of this and any SurveyUSA poll can be addressed to editor@surveyusa.com.