If recent polling is to be believed, the last six weeks have seen two seismic shifts in the San Diego electorate. According to SurveyUSA/U-T SanDiego/KGTV polls, the city lurched 15 points to the right between 12/6/13 and 1/10/14, giving Kevin Faulconer a 16 point lead. Public Policy Polling (PPIC) data released this week says the city snapped back and then some, with Alvarez now holding a one point lead.
How did we all sleep through the political earthquakes? Probably because they didn’t happen.
Polling is as much art as it is science. Who to ask and how to ask are at least as important as what you ask. The assumptions required to build and analyze a sample can cause bias in the best intentioned polls. That’s before the variance in who happens to pick up the phone that night.
Analyzing the change the SurveyUSA polls are easier to understand than comparing the different pollsters, because most of the assumptions will have stayed the same. Start with the margin of error: it was +/- 4.4 in December, and +/- 4.3 in January. In sum, 7.7 points, or half the change, could be due to statistical error in the polls. Still, seven points would be a big shift for a seemingly quiet holiday month, so let’s look at some of the other changes.
The most significant difference was an eight point increase in respondents making more than $80,000/year, a group Faulconer wins by 14 points. Also likely buffing Faulconer’s numbers were a slightly older population and fewer “Hispanic” voters (their term). None of that means the December poll was better; indeed, even Alvarez supporters thought the share of the Hispanic vote might have been too high. The differences merely begin to explain how poll results might change more than the actual electorate.
The PPIC press release provides different data than the SurveyUSA Web site, but we can compare questions that were similar. PPIC sampled more female voters, more Democrats and fewer Republicans, all of which would likely help Alvarez. They had a similar number of Hispanic respondents but notably more white and fewer Asian voters, which is harder to interpret. PPIC split their age groups differently, but had more voters over 65, which should help Faulconer.
Faulconer supporters will point out that the Democratic Party commissioned the PPIC poll, while Alvarez backers will question the independence of any U-T poll given Doug Manchester’s involvement in local politics. Those are reasonable concerns, but SurveyUSA and PPIC both need a reputation for accuracy to stay in business, so it’s unlikely their results are skewed by anything other than different underlying assumptions.
A rough average of the polls gives Faulconer a lead of 5 percent, at or just outside the margin of error of the polls. I’m guessing that’s where we are, with Faulconer picking up a few disaffected Fletcher voters over the holidays, but not 15 percent of the electorate, and not enough to put the race away. That leaves Alvarez the same challenge he had in December: solidify Democratic support and turn it out. The January SurveyUSA poll had Alvarez getting only 65 percent of the Democratic vote. That’s the number he’ll have to fix to prove PPIC right.
