Here we have crosstabs and more details about that Chron poll from the weekend. The first thing that leaps out at me is that they also polled the Controller’s race:
“If the election for city controller was held today and the candidates were Ronald Green, Pam Holm, and MJ Khan, for whom would you vote?”
Candidate Frequency Percent ===================================== Ronald Green 103 17.2 Pam Holm 90 15.0 MJ Khan 62 10.4 Someone else 8 1.3 Not sure 338 56.2
As there are only three candidates on the ballot for this race, those eight people who replied “Someone else” are in for a disappointment. Otherwise, I’d say this roughly conforms to my perception of that race.
The crosstabs themselves were about what you’d expect as well – the partisan split (Dem 43.5, GOP 35.5, Independent 21.0) and breakdown by race (White 51.0, Hispanic 14.7, African-American 26.5) seem reasonable. Getting down into the subgroups was very interesting. Brown led among men, with Parker second. He also led among women over Parker by a tiny bit. Parker led among Democratic voters, but Brown had a big lead among Republicans, with Locke trailing all three of his opponents; I’m going to guess the timing of his release in which he touts the support of numerous GOP bigwigs isn’t coincidental. Locke did lead among African-American voters, but not by much over Brown, and Brown had a large lead among Hispanic voters; despite sweeping the endorsements among Latino groups, Locke was in third among this group.
You shouldn’t read too much into any of this, as the subsamples are pretty small, but they’re fun to look at nonetheless. I still don’t have a satisfactory answer to my question about how this sample was done. They did use voters who self-identified as “very likely” (84%) or “likely” (16%) to vote, but frankly unless they pre-screened the pool to only quiz those who had some recent history of voting in city elections, I wouldn’t put too much stock in it. If all they did was ring up registered voters and use a “how likely are you to vote this fall” question as the screen, then unless they called about 2000 people to begin with, I think they’re way oversampling unlikely voters. In the end, we’ll just have to see what the scoreboard says. Campos and Greg have more.
I think they called and asked the respondents to self-identify. I went back and found the Chron article on the 2008 Zogby poll during early voting and the methodology for that poll was exactly that – asked the voters if they were likely to vote and not based on public data on voting records.
Pingback: Chron report Controller numbers from its poll – Off the Kuff
Pingback: Green’s tax liens – Off the Kuff
Pingback: “I don’t know” wins big! – Off the Kuff