I've alluded a couple of times to the fact that Barack Obama's problems in Appalachia, which had seemed to be so pervasive before, suddenly seem to have righted themselves. While he is not running as strong as Hillary Clinton probably would in the region, his numbers now look fairly normal for a non-Southern Democrat.
Two purely Appalachian states have had polling released since the end of the primary campaign. Those are Arkansas, where Rasmussen showed Obama with a 15-point improvement, and Kentucky, where SurveyUSA showed him cutting his deficit from 24 points to 12.
We can also look at the before and after versions of Quinnipiac's Swing State polling, concentrating on Obama's numbers in Southern Ohio and in Southwest and Central Pennsylvania, which are usually classified as part of the Appalachia. Obama's numbers have improved significantly in those regions too:Region May June Change
Southeast OH -20 -15 +5
Southwest OH -14 +3 +17
Southwest PA -13 +2 +15
Central PA -14 TIE +14
Arkansas -24 -9 +15
Kentucky -24 -12 +12
=============================================
AVERAGE -18.2 -5.2 +13.0
Beyond Ohio and Pennsylvania, this could have implications in West Virginia, where Rasmussen had him trailing John McCain by 8 points just before the primaries concluded. If he gets the same 10+ point bounce in West Virginia that he has gotten elsewhere in the region, the state suddenly looks extremely competitive -- which is why our model is inferring that West Virginia should be a pale shade of pink rather than a ruby red.
Wednesday, June 18, 2008
Does Obama still have an Appalachia Problem?
Monday, June 16, 2008
A Refinement to the Adjustment, Part II
The principal criticism of the Trend Adjustment that I introduced on Saturday is that it assumed that the trend was uniform across all states. Even if we can demonstrate that Barack Obama has gained, say, 3 points in his polling on average, and even if that average was taken across a fairly robust group of state and national polls, it might not hold that the bounce would be felt the same in Utah as it might be in Massachusetts.
I agree entirely with this criticism in theory. I would also argue that it is probably better to assume a uniform trend than no trend at all. The polling has become dense enough (particuarly if we include national polls) that we're getting a pretty fair mix of state and national polls in any given week. It is unlikely that Obama could improve his position in say 10 out of 12 state polls, and 5 out of 6 national polls, without his also being likely to have improved his position in other states that weren't polled during this period.
Nevertheless, it would clearly be best if we could have our cake and eat it too: adjust for the most recent trends (in a somewhat cautious way) without having to take some of the state-by-state specificity out of our model. I think I've developed a reasonable way to accomplsih that.
The basic way that we developed the trend estimator was to express each polling result as a combination of two dummy variables, one representing the state/pollster combination (e.g. "Quinnipac-Florida" or "Zogby-Delaware") and the other the week in which the poll was conducted. Each poll in our database can thus can be expressed in the form of a regression equation:
...'Margin' represents the polling result (Obama's total less McCain's), whereas the squiggly little 'e' you see is a term denoting the residual error/uncertainty. Technically speaking, there are coefficient terms on the two dummy variables, though over the long run, these coefficients will by definition equal one. Likewise, the error term will definitionally equal zero over the long run. However, just because the coefficients equal one on average does not mean that they do so in every single case. Another way to express our regression would be to embed the uncertainty term in the time-trend dummy, as follows:
In this equation, m represents a multiplier on the weekly trend variable. It is trivial to solve for m.
In a state which is more impacted by a time-defendant trend, m will be greater than one. In a state that is less impacted by the trend, it will be less than one.
Once we have a derived an m for each poll in our database, we can then regress it against a series of demographic variables in the state where the poll was conducted to see whether there is any pattern to the residuals. Since our particular concern is with recent trends, we weight recent polls much more heavily when conducting this analysis. (A couple of technical notes: we discard any cases in which the pollster has polled the state just once, as m will always be one in these cases. Also, we discard cases where the weekly dummy is a very small number -- anything less than one, in fact -- as this can produce very large, highly erratic values of m).
The demographic regression that I perform on m includes relatively few variables. This is because there aren't all that many useful data points to work with -- we need very recent polls, and for those polls to have been conducted in a state that the pollster surveyed previously -- so there is more risk of overfitting the model. The particular variables we include are a state's partisan ID index, its Kerry vote share in 2004, its black population, its Hispanic population, its average per capita income, its percentage of senior citizens, and its percentage of evangelicals. With the exception of the Kerry and 'partisan' variables, which are too fundamental to the model to be excluded, these variables have the virtue of not being strongly intercorrelated with one another.
As it turns out, there are some patterns in where Obama's bounce is showing up. It is coming in states where Democrats have a strong party identification advantage (no surprise), and seems to be especially strong in states where many voters are registered as Democrats, but where John Kerry did not perform well in 2004. This particularly describes states like West Virginia and Arkansas, where Obama's numbers have improved significantly, and where (assuredly not coincidentally) Hillary Clinton also performed well. The other observable trend is that Obama's bounce has been larger in states where there are not a lot of African-American voters, simply because there are few marginal gains for him to make among that demographic. It will probably always be the case in this election that states with lots of African-American voters will be less responsive to trends in the polling numbers.
This demographic regression allows us to estimate a unique value of m for each state. I cap the values of m at 0.0 and 2.0, respectively. The average value of m will not necessarily be 1.0, as it could be the case that particular kinds of states are especially predisposed to a bounce, and those states have also been polled more frequently (in fact, this does appear to have been the case to a small degree over the past couple weeks). The present m values for some representative states are as follows:Kentucky 1.98
In adjusting our polling numbers, we take the trend from our LOESS estimator and multiply it by m. For example, say that our LOESS curve estimates that Barack Obama is polling 3 points stronger now on average than he was three weeks ago. If we take a 3-week old poll from Kentucky, we will adjust it upward (toward Obama) by (3 x 1.98) = 5.94 points. In California, we will adjust it by (3 x 0.97) = 2.91 points. And in Arizona, we would adjust it by only 0.87 points.
Arkansas 1.93
Massachusetts 1.76
Oklahoma 1.66
New York 1.37
Michigan 1.05
North Carolina 1.01
California 0.97
Pennsylvania 0.93
Florida 0.71
Nevada 0.70
Ohio 0.54
Arizona 0.29
Utah 0.00
Taking into account the sensitivity of individual states to time trends produces a slightly less impressive result for Obama than we had been figuring on over the weekend, as his bounce seems to be most profound in states where he was already well ahead (like Massachusetts), or where he is probably too far behind to catch up (like Oklahoma). Still, we have seen at least some bounce for Obama across a large and relatively diverse array of states, and can expect to see that trend manifested in other states where new polls will come out unless his bounce begins to recede nationally.
Saturday, June 14, 2008
McCain's Atheist Problem?
According to Gallup, John McCain trails Barack Obama by 25 points among voters for whom religion is not "an important part of [their] daily life". McCain leads by 5 points among those who answer that question in the affirmative.
These sorts of numbers are generally described as a problem for the Democratic candidate. However, as Ruy Teixeira pointed out four years ago, if you had to pick a sign of this divide to be on, it might be on the side of the secular. That is because by almost all indicators, religious participation in the United States is decreasing. According to a Pew poll, 45 percent of Americans now completely agree with the statement that "prayer is an important part of my daily life", down from a peak of 55 percent in 1999. (There does appear to have a bit of a "God Bounce"/mini-revival in the mid-late 1990s -- not so much in the number of religious Americans, but in the activity and enthusiasm of those that do practice).
Moreover, the younger generation is less religious than the older generation. 19 percent of those born after 1977 say they are atheist or agnostic, as compared with 11 percent of Boomers (born 1946-1964), and 5 perecnt of pre-Boomers (born before 1946).
Barack Obama, of course, does need to at least hold his own among actively religious voters, who constitute 65 percent of the electorate according to Gallup. He is able to do so thanks to substantial support from African-American and Latino voters, while trailing McCain by 25 points among actively religious, non-Hispanic whites. Nevertheless, if these generational trends hold, then each year a coalition based on actively religious voters will become marginally less successful.
Monday, June 2, 2008
South Dakota Projection: Obama by 5
I thought that South Dakota was going to be a fairly simple little state to project. But it's actually rather idiosyncratic, in ways that tend to perplex the model.
What Barack has going for him: For the most part, South Dakota is as white as a fresh snowfall over Sioux Falls. If we treat it as a congressional district (which is exactly what South Dakota is), it ranks 426th out of 435 in terms of the percentage of African-Americans. But it's also a certain kind of white -- the whites that we call "WASP"s and which represent people of German, English and Scandinavian descent. Barack Obama has tended to do well with those kinds of white people, who are also fairly prevalent in (for example) Oregon and Wisconsin.
What distinguishes South Dakota, however, is its extremely large Native American population. Nearly 10 percent of its population is in whole or in part Native American. Although Native Americans have fairly low turnout rates -- they tend to vote Democratic when they do turn out. So we can probably expect a similar share in tomorrow's electorate.
Which candidate tends to do well with Native Americans? So far as I can tell, there is no polling data on this issue. Frustratingly, in the other states with sizable Native American populations like New Mexico and Oklahoma, Edison-Mitofsky just lumped them into the "other" cateogry in its exit polling and did not break out their data.
But the regression model is fairly well convinced that this is a good group for Barack Obama. The case of New Mexico in particular might be instructive. New Mexico is a relatively poor and heavily Hispanic state, but Obama, somewhat unexpectedly, nearly tied Hillary Clinton there. He also performed much better in AZ-1, which has a huge Navajo population, than you'd anticipate from the underlying demographics. And he did exceptionally well in Alaska's caucus, another substantially Native American state.
There are, naturally, some counterexamples; Barack Obama didn't perform well in Oklahoma for instance (although he barely fielded an organization there). But the relationship is fairly substantially statistically significant, and it dovetails with anecdotal evidence that suggests that Obama's Native American outreach has been unprecedented in its breadth. Native Americans could save Obama in South Dakota -- either bringing him a victory that he might not have earned otherwise, or keeping the margin close in the event of a loss.
As an aside, Native Americans are also a reason why South Dakota might be difficult to poll. I have no idea about the mechanics of reaching someone by telephone on an Indian Reservation (reservations occupy perhaps 20 percent of South Dakota's territory). And if you're not used to polling the state, you might not know enough to recognize their absence.
What Hillary has going for her: Although South Dakota might quite be Clinton's kind of white, it still is very white. It's a fairly old state, and rather uniformly middle class. Unlike some other Western states, where the Democratic half of the electorate can actually be quite liberal, South Dakota's Democrats gravitate toward the political center. And it has a closed primary, which our model is (finally) getting around to recognize as a slight advantage for Clinton.
But Hillary probably would not be especially likely to win South Dakota if she hadn't spent quite a bit of time there. The New York Times records her as having made 10 campaign appearances in the state, an exceptionally high ratio relative to the number of delegates available. Obama (quite wisely I think) returned to South Dakota over the weekend, and so this is not a case of Obama blowing off a state like he did in West Virginia and Kentucky. Still, the advantage in time spent on the ground is worth several points to Hillary.
Although South Dakotans aren't very much like Kentuckians, there may also be some truth to the notion that these are the types of voters that Clinton is doing better with than she had been earlier in the nomination process. I played around with a whole bunch of different interaction variables related to the timing of different primaries, leading to some versions of the model that showed Clinton a few points ahead in South Dakota. However, this led to a messy model with all sorts of multicollinearity issues, so what I eventually did was to compromise by removing the interaction variables, but giving more weight to recent primaries in determining the regression coefficients.
That brought Clinton a couple of points closer to Obama, and if I'm wrong about something like Obama's performance among Native Americans, she could very easily win South Dakota (it should probably be thought of as a "toss-up"). Even a win of some magnitude (high single digits or very low double digits) would not completely stock me; I'm not saying that this is the most likely scenario, just that it's a difficult enough state to pin down that we shouldn't rule it out.
Still, this is not a state with an especial amount of affection for Clinton -- witness, for instance, her exceptionally poor fundraising numbers. Something like the ARG scenario seems completely batshit crazy to me.
Prediction: Our model's official prediction is Obama 52.5 percent, Clinton 47.5 percent, for a margin of 5 points exactly. We're also projecting distinctly heavy turnout of 130,915 voters out of South Dakota's roughly 200,000 registered Democrats, for voting tallies of Obama 68,701, Clinton 62,213, a net gain of about 6500 votes for Mr. Obama.
Delegate wise, with just one congressional district, South Dakota is completely boring. The final split will almost certainly wind up being 8-7 for one or the other candidate. A candidate would need to win by 22.2 points to get a 9th delegate, and would earn a 10th delegate if they won by 25 points. So our delegate projection is Obama 8, Clinton 7.
Wednesday, May 28, 2008
A Senior Surge?
Perhaps I'm a little sore that Hillary Clinton cited Karl Rove's electoral map analysis in her letter to superdelegates and not ours here at 538, but there's one passage in that letter that's particularly misleading.
The increase in participation in the primaries has been driven by core groups favoring Hillary, led by women, Latinos and older voters.There is no doubt that the share of Latino voters increased dramatically in the primaries, nor that the share of women voters increased somewhat. But older voters?
Overall, more than 22 million Democratic primary voters were over the age of 45 this year, as compared to less than 10 million who voted in the 2004 Democratic primaries.
At the end of this article is a comparison of the composition of the Democratic electorate in the 23 states in which exit polling data was available in both 2004 and 2008. The key findings are as follows:
* The share of the electorate aged 65 and older decreased in 21 states, increased in one state (Wisconsin), and was unchanged in one state (New Hampshire).
* The share of the electorate aged 45 and older, likewise, decreased in 21 states, increased in one state (Delaware), and was unchanged in one state (New Hampshire).
* The share of the electorate aged 18-29 increased in all 23 states.
* Weighted by the turnout in each state, voters aged 65+ made up 18.0 percent of the electorate in 2008 as compared with 23.3 percent of the electorate in 2004; a 22 percent decrease.
* Weighted by the turnout in each state, voters aged 45+ made up 60.9 percent of the electorate in 2008 as compared with 67.9 percent of the electorate in 2004; a 10 percent decrease.
* Weighted by the turnout in each state, voters aged 18-29 made up 14.5 percent of the electorate in 2008 as compared with 9.4 percent of the electorate in 2004; a 53 percent increase.
Did the number of older voters increase in absolute terms? Of course -- since something like three times as many Democrats cast ballots in the primaries this year. The turnout of midgets of mixed French Creole/Albanian ancestry also increased in absolute terms. But the average age of a Democratic voter decreased from about 52 in 2004 to 49 in 2008.
I don't know who runs the Clinton communications shop these days, but there is a certain amount of bottom-feeding in their argumentation that tends to impeach their credibility on other issues. Why not make the argument about women and Latinos -- which ain't a bad argument at all -- and leave it at that?
Friday, May 16, 2008
But Dukakis won West Virginia
Just because the Clintons say something doesn’t mean it’s untrue. Hillary’s claim that she would do better against John McCain in swing states such as West Virginia — no Democrat has captured the White House without winning there since 1916 — is quite plausible. Obama is in danger of being cast as the Michael Dukakis of the 21st century (fairly or not). Polls show that in West Virginia, Obama wins only 53 percent of Democratic primary voters in a matchup against McCain. When only half of the party base is willing to vote for the nominee against a Republican, that nominee and that party have real problems.Goldberg has something of a point. West Virginia very much is an electoral advantage to Clinton. Although the general election matchups in the state have not been polled in a long time, we have Clinton winning the state against McCain whereas Obama is way behind. It along with Arkansas are the two states where Clinton has the largest inherent advantage against Barack Obama.
The irony is that Dukakis -- the nerdy, Willie Horton-releasing technocrat from Massachusetts -- actually carried West Virginia by 5 points. And lost the election by 315 electoral votes.
Arguably the fundamental change in American partisan politics over the past 30 years is that the Democrats have gone from being the party of the working class to the party of the bourgeoisie. Jimmy Carter's coalition was built on folks who did not attend college -- but Ford beat him by 10 points among college graduates, and Reagan beat him by 16 four years later. The same pattens largely held for Michael Dukakis.
Clinton -- Bill Clinton -- captured the best of both worlds. He performed well among those voters who hadn't attended college as well as among those with postgraduate educations. But by the time we got through the rabbit hole to John Kerry in 2004, the coalitions had flipped. Kerry was still carrying the postgraduate crowd, but had lost the "no college" voters.
These effects are particularly noticeable in West Virginia, which is among the least-educated states in the country. Indeed, Thomas Frank's book should really be entitled "What's the Matter with West Virginia?"
Hillary Clinton's issue is that she rolls the clock a little too far back - not to the Clinton years but to the Carter years. She wins back the low-education voters that Kerry had lost -- and hence, she wins back West Virginia. But, as of about a month ago, she was actualy losing college graduates to McCain. Hence, well-educated states like Wisconsin, Iowa, Minnesota, Virginia, Colorado, and Oregon give her challenges. McCain has even been within a couple of points of her in some polls of Connecticut. Each Democrat would face challenges in building a winning coalition, and it is not intrinsically obvious which one is greater.
p.s. As long as we're talking about the National Review and Michael Dukakis, I've been looking for an excuse to present the following:
Thursday, May 15, 2008
Among working class whites, Clinton kicked Edwards' ass too
It seems to be taken for granted that John Edwards had some stranglehold on the working class white vote. But this was not really the case. The working class vote was not a particular strength of John Edwards in either relative or absolute terms.
Following is a summary of some key exit polling metrics from among the five states in which John Edwards competed (using the term 'competed' liberally in the case of Florida).
A couple of things to call your attention to.
Firstly, John Edwards actually performed slightly better among voters making more than $50,000 per year than among those making less.
Secondly, while we only have this data available in three states, there was no real difference in the education levels of Edwards supporters.
Thirdly, in every state where we have data available, Barack Obama performed significantly better than Edwards among voters making less than $50,000 per year, and among voters who did not attend college. This held true in racially mixed states like Florida, as well as extremely white states like Iowa and New Hampshire.
Fourthly, in every state where we have data available, Hillary Clinton performed significantly better than Edwards among voters making less than $50,000 per year, and among voters who did not attend college.
*-*
To the extent that Edwards had a demographic base, it was not determined by class lines but rather by geography: Edwards performed better in rural areas than he did in the suburbs, and better in suburbs than he did in the cities. But really, John Edwards didn't quite have a base: there was no commonly-identified demographic group amongst which he had a plurality, yet alone a majority of the Democratic vote.
This is not to disparage John Edwards. He had the misfortune of running against two All-Star level candidates who would probably have had the stage to themselves in any other year. To the extent he had a strength, it was among rural voters, and that could be helpful to Barack Obama in Kentucky, which among other t
67 comments
Post a Comment