prop 8: Nate Silver has my back

Last week, I wrote this:

The implication [of the AP story] is clear, and has been said outright, first-time non-white voters brought into the system by the Obama campaign provided the margin necessary to pass Prop. 8.

Except that if you look at the data from the AP exit poll [now a pdf], that isn’t clear at all.

Unless there are cross-tabulations from this poll that have not been made publicly available, I cannot see how the numbers support the certitude of the claim. . . .

Democrats overwhelmingly rejected Prop. 8, first-time voters overwhelmingly rejected Prop. 8, those who are in accordance with Obama’s positions overwhelmingly rejected Prop. 8, those who supported Obama in the primary overwhelmingly rejected Prop. 8, those who voted for Obama on Tuesday overwhelmingly rejected Prop. 8.

Five days later, Nate Silver of FiveThirtyEight wrote this:

Certainly, the No on 8 folks might have done a better job of outreach to California's black and Latino communities. But the notion that Prop 8 passed because of the Obama turnout surge is silly. Exit polls suggest that first-time voters -- the vast majority of whom were driven to turn out by Obama (he won 83 percent [!] of their votes) -- voted against Prop 8 by a 62-38 margin. More experienced voters voted for the measure 56-44, however, providing for its passage.

Now, it's true that if new voters had voted against Prop 8 at the same rates that they voted for Obama, the measure probably would have failed. But that does not mean that the new voters were harmful on balance -- they were helpful on balance. If California's electorate had been the same as it was in 2004, Prop 8 would have passed by a wider margin.

Furthermore, it would be premature to say that new Latino and black voters were responsible for Prop 8's passage. Latinos aged 18-29 (not strictly the same as 'new' voters, but the closest available proxy) voted against Prop 8 by a 59-41 margin. These figures are not available for young black voters, but it would surprise me if their votes weren't fairly close to the 50-50 mark.

There are few quantitative analysts I would trust more than Silver (and his presidential predictions were the best of the lot this cycle), so it is a real confidence builder for me to know that when he looks at the data he sees the same thing that I see—or, more accurately, he doesn’t see the same thing that I don’t see. There is nothing in the exit polling to support the narrative that the first-time African American voters brought to the polls by Barack Obama’s campaign provided Prop. 8 with its margin of victory—and, in fact, most evidence seems to point the other way.

Silver believes that the ballot measure owes its passage to older voters, noting that if no one over 65 had voted, Prop. 8 would have failed by “a point or two.” Silver suggests that as that demographic ages out of, um, life, bigoted efforts such as this one will eventually fail.

While I tend to agree overall—the younger you go, the more comfortable most seem with diversity—I think that Silver should take a look at the family factor. Those that are married and have children (31% of the sample) voted in favor of the measure 68% to 32%. All others voted against the gay marriage ban by a ten-point margin.

(I know what you’re thinking: “All others” includes most of the voting homosexual population. That’s probably true. Alas, there is no cross-tabulation for “married heterosexuals without children”—however, because the “all others” segment is so much larger than the “married with children” slice, even if you could subtract the gay vote, I suspect that this segment would still have rejected the proposition.)

The question becomes “Are beliefs about gay marriage static?” Will the young segments that voted against Prop. 8 continue to feel the same way, even as they age and/or have children? To ask it another way: Do those married with children tend to favor the ban more because they would tend to be older than those without kids, or did those that are younger reject Proposition 8 because they had yet to reproduce?

To my eye, the history of civil rights movements in the United States would favor Silver’s take on the numbers, but the numbers don’t confirm this, at least not with absolute certainty. I guess, as they—and the numbers—say, time will tell.


(cross-posted on Daily Kos, capitoilette, and The Seminal)

doubt, by the numbers

A snippet of “analysis,” courtesy of the AP (here via the LA Times), seems to have dictated what fast became the received wisdom on the “be careful what you fish for” conundrum of the Obama campaign’s drive for increased minority turnout vis-à-vis the very unfortunate passage of Proposition 8 (eliminating marriage rights for same-sex couples):

California's black and Latino voters, who turned out in droves for Barack Obama, also provided key support in favor of the state's same-sex marriage ban. Seven in 10 black voters backed a successful ballot measure to overturn the California Supreme Court's May decision allowing same-sex marriage, according to exit polls for The Associated Press.

More than half of Latino voters supported Proposition 8, while whites were split. Religious groups led the tightly organized campaign for the measure, and religious voters were decisive in getting it passed. Of the seven in 10 voters who described themselves as Christian, two-thirds backed the initiative. Married voters and voters with children strongly supported Proposition 8. Unmarried voters were heavily opposed.

The sound of this report, what with it being grounded in a poll and all, makes it hard to refute—and, indeed, it seems few have tried. The language in the top paragraph, or some slight variant, appears in most of the major reports I’ve read or heard in the two days since the November 4th referendum. The implication is clear, and has been said outright, first-time non-white voters brought into the system by the Obama campaign provided the margin necessary to pass Prop. 8.

Except that if you look at the data from the AP exit poll, that isn’t clear at all.

Unless there are cross-tabulations from this poll that have not been made publicly available, I cannot see how the numbers support the certitude of the claim. The above narrative is a possibility, but so are many other stories—and I feel that other conclusions are likely just as viable.

First off, while it is true that African American voters in California did vote overwhelmingly for the marriage ban—70% YES to 30% NO—Latinos were more closely divided: 53% YES to 47% NO. That’s significant, but not in the same league as the margin from African American voters. It also should be noted that African Americans accounted for 10% of those polled; Latinos, 18%.

More to my point, however, is the missing cross-tab. There are a good number of sub-samples available in the published results (more on some of those in moment), but “African American first-time voters that voted for Obama-Biden” or even just “African American first-time voters” are not among them.

It is quite possible that with that many cuts, the sample size is too small to yield results that pass statistical muster, but without the ability to run my own cross-tabs, I can’t tell you.

Here’s what I can tell you (based on what is posted), and it is some of these numbers that make me at least harbor doubts about the “new Black voters are conservative on social issues” storyline.

(The first number in brackets is the % of the total sample, the second number is the % that voted YES, and the third is the % that voted NO.)

Democrats (42) 36 - 64
Republicans (29) 82 - 18
Independents (28) 46 - 54

Is this the first year you’ve ever voted?
Yes (14) 38 - 62
No (86) 56 - 44

Union Household (25) 56 - 44
Non-union Household (75) 50 - 50

Who did you want to win in the nomination?
Dems for Clinton (15) 39 - 61
Dems for Obama (23) 31 - 69

Suburban voters (51) 59 - 41
Large city voters (45) 45 - 55

Do you think Obama’s positions are:
Too liberal (32) 74 - 25
Too conservative (7) (sample too small)
About right (56) 31 - 69

Voted for Obama (60) 32 - 68
Voted for McCain (38) 84 - 16

Democrats overwhelmingly rejected Prop. 8, first-time voters overwhelmingly rejected Prop. 8, those who are in accordance with Obama’s positions overwhelmingly rejected Prop. 8, those who supported Obama in the primary overwhelmingly rejected Prop. 8, those who voted for Obama on Tuesday overwhelmingly rejected Prop. 8.

Urban voters rejected the measure by a ten-point margin, while suburban voters supported it by eight. I don’t know what the racial breakdown is of California’s suburbs, but I would assume that large cities would have the larger African American populations.

The only category that could have been influenced by the “Obama effect” (for lack of a better name) is the union vote. It has been widely reported how active organized labor was in getting out the vote for Barack Obama, and, as you see above, union households favored the marriage ban by 12 points. However, and this is a big however, this number might also be deceptive because the “households” category, almost by definition, includes a lot of families. Married with children (31% of the sample) voted YES 68% of the time (all others—69% of the sample—rejected Prop. 8 by ten points).

While none of what I have just detailed rules out the hypothesis that first-time African American voters brought to the polls by the Obama candidacy proved the difference in the passage of Proposition 8, I think there is enough here to call that narrative into question. For all we know, most of the 70% of the African American population that voted YES on 8 would have come out and voted even if Obama wasn’t on the ballot. Conversely, it might be the case that the 30% of African Americans that rejected the measure are the ones voting for the first time. Might be—I just can’t tell.

And if I can’t tell, I am figuring that most of the establishment press parroting the AP’s narrative probably can’t tell either.

No doubt there is much to be done to combat the homophobic bias evidenced in this tally, and in similar outcomes in other states, but the discussion about what is to be done could be influenced by perceptions of which groups bear responsibility for the final outcome. Indeed, the way Obama governs could be shaped by the larger story about what kinds of voters provided the president-elect’s margin of victory. Without the ability to further analyze the exit poll data, we should not accept the center-right narrative or claims of any particular Obama effect.


(cross-posted on capitoilette and The Seminal)