All the News on Jews’ Views: Comparing 3 Polls
There’s good news and bad news for President Obama in a new survey of American Jewish opinion released Thursday by the Workmen’s Circle. First, the bad news: Jewish voters favor Obama over Mitt Romney by about two to one — 59% to 27%, with 14% undecided. If undecideds follow the same 2-to-1 split, the result will be 68% to 32%. This points to a 10% drop from November 2008, when Obama got 78% of the Jewish vote, according to national exit polls at the time. The good news is that it’s not November yet, and if you compare June 2012 to June 2008, Obama is doing considerably better now than he was then. At this point in 2008 Jews were backing Obama by only 62% to rival John McCain’s 31%, according to Gallup’s tracking poll. Obama dropped further in July 2008, to 61-34, before beginning a steady rise in August. In fact, a surge might already be discernible this year, if we compare the Workmen’s Circle survey with a similar survey released two months ago, April 3, conducted by the Public Religion Research Institute for the Nathan Cummings Foundation.
Will the president repeat his 2008 late-summer uptick? Hard to say. Romney isn’t likely to give him the sort of gift McCain offered when he chose the spectacularly unqualified Sarah Palin as his running-mate. On the other hand, everything else in the Workmen’s Circle poll, which was conducted by Professors Steven M. Cohen and Samuel Abrams, points to a Jewish public that remains solidly liberal. Given the starkly conservative cast of the Republican campaign so far, it seems unlikely that Romney could muster more enthusiasm among Jewish voters than the more moderate McCain did in 2008. It could be that distress over Obama’s Israel policies will lower his Jewish support, but both surveys show Israel playing very little role in Jewish voters’ thinking. In fact, Cohen’s statistical analysis of respondents’ preferences and demographic characteristics indicates that people who have strong opinions about Israel tend to show a host of other tendencies that factor as strongly if not more so into their decisions.
In some ways the Workmen’s Circle survey confirms the trends that turned up in the Cummings Foundation survey in April; in other ways the WC sample is more conservative (I’m not sure why, and I won’t speculate right now). In certain ways, both polls — and a third one, the American Jewish Committee annual survey, released April 30 — look remarkably similar. Remarkable, that is, considering that they use different methodologies, draw on different population samples and reflect a variety of sponsors’ ideologies from the upscale liberal Cummings Foundation to the grittier left-liberal Workmen’s Circle to the devoutly centrist AJC.
AJC didn’t ask a lot about social policy, so we’ll focus first on WC and Cummings. Both found overwhelming support for legal abortion in all or most cases (close to 90%). Both found strong support for legal same-sex marriage, though not identically so: Cummings found 81% support, WC 68%. On the question of whether or not the government should increase environmental protection, even at the cost of fewer jobs, Cummings found a strong majority in favor, 69% to 30%, while the WC found only plurality support, 41% to 29% with 31% “not sure.”
None of this is terribly surprising. What is unexpected is the strong support in both surveys for what used to be known as lunch-bucket liberalism—egalitarian economics, higher taxes, regulation of the financial industry and trade union rights. I go through some of those numbers in my latest column in this week’s Forward Forum, elsewhere on this site. I recommend following the links and looking at the surveys themselves to get the full picture. It’s an eye-opener. I also explain there how I got this point exactly wrong when I first wrote about the Cummings survey two months ago, by the way.
The bottom line: The first time around I was reporting the numbers reported by the survey organization rather than working with the raw findings. When I went back this week and dug into the actual survey results, in order to compare them with the WC survey and try to understand differences between the two, I found a bunch of surprises. I also found that the two surveys are much more similar than you’d realize if you simply read the sponsors’ reports. (Full disclosure: I got to see the full WC survey before it was released because I’m doing some consulting work for the WC, though not on the survey project.)
Where the two surveys bore the most remarkable resemblance to each other and to the AJC survey was on broad questions of how American Jews identify themselves politically and ideologically. WC and AJC found, respectively, 44% or 46% liberal or moderate-leaning-liberal, 18% or 19% conservative or moderate-leaning-conservative and 34% or 35% dead center. (Cummings didn’t ask respondents to place themselves ideologically.) On party identification, both WC and AJC found their respondents identifying, respectively, 55% or 52% Democrat or independent-leaning-Democrat, 22% or 19% Republican or independent-leaning-Republican and 17% or 26% firmly independent.
Cummings doesn’t report independents: instead it shows respondents as 50% Democrat, 20% “leaning Democrat,” 13% Republican and 16% “leaning Republican.” This doesn’t sound anything like the other two, until you dig into the raw numbers. AJC asked its 26% self-described “independents” which way they would lean if forced to choose, and the answers were 2-to-1 Democrat over Republican. If you do the math, you come out with 69% Democrat or leaning Democrat and 28% Republican or leaning Republican, which is almost exactly what Cummings found. Apply the same formula to the WC findings—that is, split the independents 2-to-1 Democrat over Republican, following the AJC’s responses—and you get almost exactly the same result.
By the way, the WC’s executive report of its findings offers numbers that look very different. I won’t go through them all here—you can find them by following this link. If you look at their raw numbers, though, it turns out that they got the exact same results as the other two—they just described them differently. WC respondents were asked to place themselves on a 1-to-7 scale for both party and ideology. The breakdown was then reported with a broad middle range, the 3s, 4s and 5s, grouped together as “moderate” or “independent.” Only the two highest and lowest groups (1s plus 2s, 6s plus 7s) are reported as Dems or GOP, liberals or conservatives. There’s nothing wrong with this method; it makes Jews look far less partisan and less ideologically identified than what appears in the other two polls (and countless past polls). However, if you take the numbers just to the left and right of center (the 3s and 5s) and assign them to the edges rather than to the center—call them “leaners,” that is—you get pretty much exactly the same result as AJC, Cummings and countless other surveys in past years have shown.
All of which goes to show the danger in deciding to figure out a better method of reporting statistics: It may be intellectually more satisfying, but it makes the whole thing look suspicious to the public. The bottom line is that the numbers don’t usually lie, but how you interpret them can render them very confusing, if not downright misleading.