...which is heavily unionized and establishment. But in lily white Reno and Washoe counties, hillary got smoked across the board.
January 27, 2008
Demography and the Democratic Race
There is a simple explanation for Obama's victory last night: he won African American voters. They constituted 53% of the vote, and 80% of them went for Obama.
This is an incredible result. Of this there is no doubt. But it begs a question - can Obama win white voters?
Since he lost Nevada, pundits have been suggesting that he cannot. But I think the picture is much more complicated than it first appears.
First of all, we have to acknowledge that Clinton is currently beating Obama in the polls among white voters nationwide. The latest LA Times/Bloomberg poll tells the story pretty succinctly. Clinton leads Obama among white voters by 19 points. Compare this to Clinton's 9 point lead among all self-identified Democratic primary voters - and it should be clear Clinton is doing better with white Democrats than with the Democratic electorate at large.
So, the more precise question is: can these numbers change, or has Obama "maxed out" among white voters?
Let's take a look at recent results to see if we can get an answer.
Obama won Iowa by a decisive margin. And, as Iowa is 93% white, it should come as no surprise that he won white voters by 6 points. Clinton won white women in New Hampshire; Obama won white men; and, as more women voted than men, Clinton won the state. But both of these states indicate that Obama can win sufficient numbers of white voters.
But wait a minute - what about Nevada? Clinton beat Obama by 18 points among whites. She even won white men by 6 points. Doesn't that indicate that the dynamic of the race has changed, that Obama has perhaps been marginalized as the "black candidate?"
Well...not so fast.
Clinton won white voters statewide. Of this there is no doubt. But we might ask which white voters did she win? The Nevada entrance poll clearly indicates that she won white voters in and around Las Vegas - but that she and Obama at least split them outside Vegas:
These results are really intriguing. Consider that, according to the Census Bureau's 2006 population estimate, Clark County (where Vegas is) is home to about 93% of the state's African American population. So, if Obama won Reno and split the rural counties, he must have been strong among white voters outside Vegas. There are simply not enough African American voters outside Clark County for Obama to do so well having lost white voters.
Let's expand on this. Unfortunately, the Nevada Democratic Party does not release raw vote totals. But we can look at delegates, which are a sufficiently unbiased estimate of how the votes went in each county. The following table reviews counties in Nevada according to: the percentage difference in delegates between Obama and Clinton, the percentage of whites, and the estimated population from 2006.
Obviously we cannot know the exact votes per county. These are delegate totals, not raw vote totals. Above all, we do not have breakdowns of vote by race per county. Nevertheless, this should make it clear that the entrance poll was certainly on to something. Obama won Reno (Washoe County), which is largely white. He won Carson City, which is largely white. And he did very well in the rural counties, which are also white.
It should be clear from this that it is insufficient to say that Clinton won "the white vote" in Nevada. It is better to say that she won a certain type (or types) of white voter. But what type? Why did white voters in Vegas break for Clinton so heavily while voters outside Vegas did not? Obviously, the ideal explanation is one that accounts for not just Nevada, but also Iowa and New Hampshire. I see three hypotheses that could connect these dots:
(a) It is a matter of GOTV organization. Obama beat her in Iowa. Clinton beat him in New Hampshire and Las Vegas.
(b) It is a matter of income. Whites who make more money tend to support Obama. Whites who make less money tend to support Clinton.
(c) White voters in racially uniform areas are more attracted to Obama that white voters in racially diverse areas.
Any of these could be true. Each of them has evidence to support them, and none of them excludes any other. I am sure that there are other potential explanations as well. Unfortunately, we cannot arbitrate between them. We could if the media chose to release raw exit data numbers, or at least more detailed cross-tabs. But they don't, so we can't. [The biggest difficulty is with the second explanation. We clearly saw income play a role in Iowa and New Hampshire - but we would need to see data on income controlling for race in Nevada, which is more racially diverse than Iowa or New Hampshire. The media does not provide that kind of data.]
The situation in South Carolina is not nearly as mixed as it was in Nevada. The fact of the matter is that Clinton won a strong plurality of white voters. Once again, gender was a critical factor. The entrance poll shows that white women broke decisively for her (42% to her, 36% to Edwards, 22% to Obama). Edwards won the white male vote, and Clinton and Obama were in a statistical tie for second. So, Obama's victory was dependent upon black voters. We can confirm this with a look at the county-by-county results. I ran an ordinary least squares (OLS) regression analysis on the South Carolina results. This is a way to predict a dependent variable based upon an independent variable. I tested whether the percentage of white residents per county could explain the percentage spread between Obama and Clinton per county. It does. In fact, it explains about 70% of the difference between Obama's share of the vote and Clinton's share. As the white population in a given county increases, Obama's margin of victory over Clinton decreases.
This lopsided result among white voters is consistent with the second and third hypotheses I listed. It could be a matter of economics. South Carolina as a whole tends to be less wealthy than Nevada, Iowa or New Hampshire. It could also be an issue of racial mixing. Counties in South Carolina are much more heterogeneous than the other three states. But that does not mean that either theory is true. All it means is that they are consistent with the data we have. These two theories help us link the results of Iowa, New Hampshire, Nevada, and South Carolina. But I am sure there are other plausible theories that could do the job.
The bottom line - this is a real puzzle, and it does not admit of any easy answers. White voters in Iowa and New Hampshire embraced Obama. White voters in South Carolina did not (though he tied Clinton among white men). And in Nevada, what they did depended on where they were located. Minimally, it should be clear that Obama can win white voters - so, the answer to our initial question is yes. Accordingly, breaking this race down by a simple dichotomy of white and black is oversimple and empirically unsatisfactory. There is more going on here than what appears at first glance. What it is, we cannot know.
This is what we do know. Clinton has done well among Hispanics. Obama has done well among African Americans. Depending on where and when, white voters vary their support. How will that play out on Super Tuesday? We can get a sense from the following table, which reviews the states on Super Tuesday, their pledged delegates, and the percentage of their residents who are white, African American, and Hispanic:
The final row is worth taking careful note of. This is an average of each demographic group weighted by the delegates per state. As you can see, white voters make up a majority of the Super Tuesday population - but African Americans and Hispanics are important minorities. And remember that in most states the electorate will probably oversample from African Americans and Hispanics, who tend to vote Democratic more than whites.
And so, it seems to me that Super Tuesday depends upon three variables:
First, will white voters follow the pattern they followed in Iowa and New Hampshire, in Nevada, or in South Carolina? And remember that viability, favorability, and vote choice can go go hand-in-hand in the primaries. The current polling that shows Clinton with a large lead among whites could change as a consequence of Obama's South Carolina win. After Obama's victory in Iowa, Clinton's margin among white voters shrunk to just 8 points, according to ABC News/WaPo.
Second, will African Americans "out perform" Hispanics? This is an interesting question. The Nevada entrace poll found that African Americans and Hispanics each comprised 15% of the total electorate. But Nevada as a whole is 6.6% African American, and 19.7% Hispanic. If Super Tuesday African Americans "out perform" Hispanics as they did in Nevada - then Obama will be in a better position than what the above table suggests.
Third, what happens to Edwards' voters? It is unclear what they will do if Edwards drops out. It is also unclear whether Edwards can sustain his current support. Voters can be brutal with their evaluations of viability. If they see Edwards' candidacy as hopeless, it is quite possible they will abandon it even if he stays in. If they do, where do they go?
-VG
January 27, 2008
Demography and the Democratic Race
There is a simple explanation for Obama's victory last night: he won African American voters. They constituted 53% of the vote, and 80% of them went for Obama.
This is an incredible result. Of this there is no doubt. But it begs a question - can Obama win white voters?
Since he lost Nevada, pundits have been suggesting that he cannot. But I think the picture is much more complicated than it first appears.
First of all, we have to acknowledge that Clinton is currently beating Obama in the polls among white voters nationwide. The latest LA Times/Bloomberg poll tells the story pretty succinctly. Clinton leads Obama among white voters by 19 points. Compare this to Clinton's 9 point lead among all self-identified Democratic primary voters - and it should be clear Clinton is doing better with white Democrats than with the Democratic electorate at large.
So, the more precise question is: can these numbers change, or has Obama "maxed out" among white voters?
Let's take a look at recent results to see if we can get an answer.
Obama won Iowa by a decisive margin. And, as Iowa is 93% white, it should come as no surprise that he won white voters by 6 points. Clinton won white women in New Hampshire; Obama won white men; and, as more women voted than men, Clinton won the state. But both of these states indicate that Obama can win sufficient numbers of white voters.
But wait a minute - what about Nevada? Clinton beat Obama by 18 points among whites. She even won white men by 6 points. Doesn't that indicate that the dynamic of the race has changed, that Obama has perhaps been marginalized as the "black candidate?"
Well...not so fast.
Clinton won white voters statewide. Of this there is no doubt. But we might ask which white voters did she win? The Nevada entrance poll clearly indicates that she won white voters in and around Las Vegas - but that she and Obama at least split them outside Vegas:
These results are really intriguing. Consider that, according to the Census Bureau's 2006 population estimate, Clark County (where Vegas is) is home to about 93% of the state's African American population. So, if Obama won Reno and split the rural counties, he must have been strong among white voters outside Vegas. There are simply not enough African American voters outside Clark County for Obama to do so well having lost white voters.
Let's expand on this. Unfortunately, the Nevada Democratic Party does not release raw vote totals. But we can look at delegates, which are a sufficiently unbiased estimate of how the votes went in each county. The following table reviews counties in Nevada according to: the percentage difference in delegates between Obama and Clinton, the percentage of whites, and the estimated population from 2006.
Obviously we cannot know the exact votes per county. These are delegate totals, not raw vote totals. Above all, we do not have breakdowns of vote by race per county. Nevertheless, this should make it clear that the entrance poll was certainly on to something. Obama won Reno (Washoe County), which is largely white. He won Carson City, which is largely white. And he did very well in the rural counties, which are also white.
It should be clear from this that it is insufficient to say that Clinton won "the white vote" in Nevada. It is better to say that she won a certain type (or types) of white voter. But what type? Why did white voters in Vegas break for Clinton so heavily while voters outside Vegas did not? Obviously, the ideal explanation is one that accounts for not just Nevada, but also Iowa and New Hampshire. I see three hypotheses that could connect these dots:
(a) It is a matter of GOTV organization. Obama beat her in Iowa. Clinton beat him in New Hampshire and Las Vegas.
(b) It is a matter of income. Whites who make more money tend to support Obama. Whites who make less money tend to support Clinton.
(c) White voters in racially uniform areas are more attracted to Obama that white voters in racially diverse areas.
Any of these could be true. Each of them has evidence to support them, and none of them excludes any other. I am sure that there are other potential explanations as well. Unfortunately, we cannot arbitrate between them. We could if the media chose to release raw exit data numbers, or at least more detailed cross-tabs. But they don't, so we can't. [The biggest difficulty is with the second explanation. We clearly saw income play a role in Iowa and New Hampshire - but we would need to see data on income controlling for race in Nevada, which is more racially diverse than Iowa or New Hampshire. The media does not provide that kind of data.]
The situation in South Carolina is not nearly as mixed as it was in Nevada. The fact of the matter is that Clinton won a strong plurality of white voters. Once again, gender was a critical factor. The entrance poll shows that white women broke decisively for her (42% to her, 36% to Edwards, 22% to Obama). Edwards won the white male vote, and Clinton and Obama were in a statistical tie for second. So, Obama's victory was dependent upon black voters. We can confirm this with a look at the county-by-county results. I ran an ordinary least squares (OLS) regression analysis on the South Carolina results. This is a way to predict a dependent variable based upon an independent variable. I tested whether the percentage of white residents per county could explain the percentage spread between Obama and Clinton per county. It does. In fact, it explains about 70% of the difference between Obama's share of the vote and Clinton's share. As the white population in a given county increases, Obama's margin of victory over Clinton decreases.
This lopsided result among white voters is consistent with the second and third hypotheses I listed. It could be a matter of economics. South Carolina as a whole tends to be less wealthy than Nevada, Iowa or New Hampshire. It could also be an issue of racial mixing. Counties in South Carolina are much more heterogeneous than the other three states. But that does not mean that either theory is true. All it means is that they are consistent with the data we have. These two theories help us link the results of Iowa, New Hampshire, Nevada, and South Carolina. But I am sure there are other plausible theories that could do the job.
The bottom line - this is a real puzzle, and it does not admit of any easy answers. White voters in Iowa and New Hampshire embraced Obama. White voters in South Carolina did not (though he tied Clinton among white men). And in Nevada, what they did depended on where they were located. Minimally, it should be clear that Obama can win white voters - so, the answer to our initial question is yes. Accordingly, breaking this race down by a simple dichotomy of white and black is oversimple and empirically unsatisfactory. There is more going on here than what appears at first glance. What it is, we cannot know.
This is what we do know. Clinton has done well among Hispanics. Obama has done well among African Americans. Depending on where and when, white voters vary their support. How will that play out on Super Tuesday? We can get a sense from the following table, which reviews the states on Super Tuesday, their pledged delegates, and the percentage of their residents who are white, African American, and Hispanic:
The final row is worth taking careful note of. This is an average of each demographic group weighted by the delegates per state. As you can see, white voters make up a majority of the Super Tuesday population - but African Americans and Hispanics are important minorities. And remember that in most states the electorate will probably oversample from African Americans and Hispanics, who tend to vote Democratic more than whites.
And so, it seems to me that Super Tuesday depends upon three variables:
First, will white voters follow the pattern they followed in Iowa and New Hampshire, in Nevada, or in South Carolina? And remember that viability, favorability, and vote choice can go go hand-in-hand in the primaries. The current polling that shows Clinton with a large lead among whites could change as a consequence of Obama's South Carolina win. After Obama's victory in Iowa, Clinton's margin among white voters shrunk to just 8 points, according to ABC News/WaPo.
Second, will African Americans "out perform" Hispanics? This is an interesting question. The Nevada entrace poll found that African Americans and Hispanics each comprised 15% of the total electorate. But Nevada as a whole is 6.6% African American, and 19.7% Hispanic. If Super Tuesday African Americans "out perform" Hispanics as they did in Nevada - then Obama will be in a better position than what the above table suggests.
Third, what happens to Edwards' voters? It is unclear what they will do if Edwards drops out. It is also unclear whether Edwards can sustain his current support. Voters can be brutal with their evaluations of viability. If they see Edwards' candidacy as hopeless, it is quite possible they will abandon it even if he stays in. If they do, where do they go?
-VG
