Both surveys are phone-based and neither provides a response rate, but that's where the similarities diverge.
What then explains the difference between these polls taken just two weeks apart?
For starters, SurveyUSA has a smaller sample size (n=600) and thus a larger margin of error than PPP (n=763). This generally means the results are less statistically precise for SurveyUSA.
Then there's the fact that SurveyUSA respondents are not necessarily likely Missouri voters, as is the case for PPP respondents.
And of course the elephant in the room is that SurveyUSA weights respondents' opinions to create a demographically-representative sample.
These differences make the surveys slightly incomparable. But on top of these factors are other inherent methodological differences that result in variation from accurate estimates. This variation was estimated by FiveThirtyEight.com's Nate Silver for all political polling firms nationally. Both SurveyUSA and PPP ranked comparatively low, but there was a three-quarter-point difference in their estimates, with SurveyUSA being slightly more accurate.
So which poll is more trustworthy? Silver would likely back SurveyUSA which has had more reliable predictions nationally, especially since the two polling firms' estimates aren't even within each other's margin of error. Given that the geographic sampling from PPP doesn't quite jibe with where the majority of voters live, we're inclined to back the SurveyUSA numbers as more likely a description of the state of Missouri voters' opinions.
One thing is certain, Nixon's popularity isn't what it once was. Both polls show sizable declines between Jan. and Nov., and that's something you can believe.
PPP Respondents by Missouri Area Code
314............21%
417............21%
573............17%
636............15%
660............6%
816............20%
Related Links:
SurveyUSA Approval Ratings for Nixon
SurveyUSA Methodology
SurveyUSA Regional Definitions
Posts
0 comments:
Post a Comment