Question:
There will also be an election for U.S. Senate in November. Democrat Dianne Feinstein is running against Republican Elizabeth Emken. If the election for U.S. Senate were being held today, for whom would you vote – Feinstein or Emken? (ASKED OF ALL LIKELY VOTERS)

About this Poll
This survey was conducted jointly by The Institute of Governmental Studies at UC Berkeley and The Field Poll September 6-18, 2012. The findings are based on interviews conducted with 1,183 California registered voters, including 902 voters considered likely to vote in the November 2012 general election. The survey was conducted by telephone using live interviewers in six languages and dialects – English, Spanish, Cantonese, Mandarin, Korean, and Vietnamese.

The overall registered voter sample was supplemented additional interviews among Chinese-American, Korean-American and Vietnamese-American voters to enable the results from these voter subgroups to be compared to those of other racial/ethnic voter populations. Funding for the multi-ethnic samples was provided by the New America Media, through a grant from the Blue Shield of California Foundation.

Up to six attempts were made to reach and interview each randomly selected voter on different days and times of day during the interviewing period. Interviews were completed on either a voter’s landline phone or a cell phone. In this survey 909 interviews were conducted on a landline phone and 274 were completed through a cell phone contact. After completion of interviewing, the overall sample was weighted to align it to the proper statewide distribution of voters by race/ethnicity and other demographic characteristics of the California registered voter population.

In order to cover a broad range of issues and still minimize voter fatigue, some of the questions in the survey were asked of a random sample of 571 registered voters. Sampling error estimates applicable to any probability-based survey depend upon its sample size. According to statistical theory, 95% of the time results from the overall likely voter sample are subject to a maximum sampling error of +/- 3.4 percentage points, while findings from the random subsample have a maximum sampling error of +/- 4.3 percentage points. The maximum sampling error is based on percentages in the middle of the sampling distribution (percentages around 50%). Percentages at either end of the distribution have a smaller margin of error. Sampling error will be larger for analyses based on subgroups of the overall sample.