Dubious Polling in California

In Defying the Odds, we discuss state and congressional elections as well as the presidential race. 

Paul Mitchell at Capitol Weekly:

To make a phone poll workable, pollsters are limiting the poll to those candidates they think are most viable.

And this can result in a flawed result, such as when SurveyUSA only included two Republican candidates out of 11 running, and found (incorrectly, mind you) that the leading candidate for the second spot in the U.S. Senate race was Patrick Little, a candidate who says he’s dedicating his life to exposing the Jews who are running this country.

Making matters worse, SurveyUSA claimed in their findings “for the first time, there is clarity on [Senator Dianne Feinstein’s] likely November opponent” without even making mention of the fact that they didn’t give Republicans the opportunity to select any of the other nine Republican candidates.

The increased use of online surveys using emails from the voter file can present a ballot that more closely resembles what a voter will actually receive, but not all public polls are using this methodology.

Aside from these known issues with public polls, there is a new one which has been creeping into the work of pollsters, particularly in this coming gubernatorial election cycle. This has to do with the makeup of the expected electorate – what pollsters call the “weighting” of the survey.