[Editor's note: This is the fourth of a six-part series by Paul Lukasiak on what polling reveals about how Americans will vote in the coming election. For more details, see parts 1, 2, and 3. -- DN]
The poisoned landscapes described in the preceding chapters are not merely an academic concern. Sexism and racism are also having an impact on the win/loss margins in individual states – and will play a major role in how the November campaign unfolds as certain states wind up “out of play” because of either sexism or racism (depending upon the candidate.)
Among the more noteworthy traits the data reveals is that as the percentage of black voters in a state increases, the presence of sexist and mysoginist attitudes decline. And that in turn can tell us a lot about voting trends in the coming election.
The chart at left is sorted by the percentage of Republican voters (Grey DATAPOINTS and line), in order to help differentiate between changes that result from an increase of Republicans, and changes that occur for other reasons. Most of the chart itself is unsurprising.
– Red (Clinton) and Blue (Obama) data points and lines/Overall vote percentage against McCain: It will come as no surprise to anyone that the more Republicans there are in a state, the fewer votes a Democrat will get.
These Red and Blue lines do, in fact, run parallel to each other, but that merely reflects overall trends over 50 states. The same parallel lines pattern appears when we sort alphabetically with Obama showing a slight, but consistent, advantage over Clinton, and the data points appear distributed at random.
The Black data points/line represent the percentage of Black voters in the state(s). Somewhat surprisingly, the trend line suggests that the more Republicans there are in a state, the more Black voters there are.
Orange data points and line – White votes for Obama. The Orange data points/line represents percentages of all votes cast, in other words, the Orange line starts at 40%, it means that 40% of all voters in a state were White voters who support Obama against McCain.
The difference between the Orange (White voters for Obama) and Blue (Obama total vote percentage) data points/lines tells you how much of Obama’s support was “non-White” (i.e., the Blue triangle on the far left right below the Blue line that Obama was supported by 53% of voters in that state (Rhode Island), the Orange triangle directly below it tells you that 45.4% of all voters in that state were Whites who cast their vote for Obama, and thus that 7.6% if voters in that state were non-Whites who supported Obama (53% minus 45.4% equals 7.6%).
Pink (Male votes for Clinton) and Light Blue (Male votes for Obama) data points and lines. Unsurprisingly, as the percentage of Republicans increases, the numbers of male supporters of both Democrats goes down These lines almost always move in the same direction; the size/impact of the gender gap can be determined by how close they are – the further away from each other, the bigger the gender gap.
Both the Pink and Light Blue lines are, like the Orange line, based on weighed data, i.e. they represent a percentage of all “votes” cast. The difference between the Pink line to the Red line (Clinton total %), and the Light Blue line to the Blue line (Obama total %), reflects the female votes cast for Clinton and Obama respectively.
While the Light Blue line remains above the Pink line, showing that sexism continues to play a role in all 50 states, the distance between the Pink and Blue lines decreases. As shall be shown, that decrease in distance is related to the trend of less white support for Obama as black voter percentages increase.
When you look at the data from a different perspective, that of the percentage of Black voters, the trends above are confirmed, and new ones can be seen. The traditional strong support of African Americans for Democrats makes the increase in Clinton’s overall totals, and her increase in male votes, unsurprising.
But we see a different trend with regard to Obama’s data – while Clinton’s margins (Red line) increase as the percentage of traditionally Democratic African American voters increases in a state, Obama’s (Blue line) trends lower as the percentage of African American voters increases.
A similar pattern emerges when we look at male support for the two Democrats. Clinton’s male support (Pink line) increases as the percentage of Black males among male voters increases, while Obama’s male support (Blue line) decreases – virtually wiping out any evidence of sexism as the percentages of Black voters gets higher.
And Obama’s White voter % (Orange) line explains why this can occur, despite his overwhelming support among African Americans. Obama’s race is a not a factor for White voters where race is not a factor in their daily lives, but it would appear that the more African Americans there are in a state, the more White voters resist voting for a Black candidate. The resistance of White voters to a Black candidate is so significant that it overwhelms the influence of the traditionally Democratic black vote as the percentage of Black voters rises, leading to a decrease overall vote totals, and overall male (and female) support for Obama.
But not even this provides a full picture of the way in which sexism and racism intertwine to create two different poisoned political landscapes for the Democratic nominee. Breaking the data down further provides even more insight. (Note: Overlapping state data is used in the following charts in order to provide a clearer sense of the dynamic involved.
There are 24 states where the percentage of Black voters is under 6 %. This includes 8 of 10 Mountain/Plains states (four of which are a 0% Black voters), 5 of 6 New England states (3 of which are the other “0% Black voter” states), 6 of Far West states, 5 of 10 North Central states, and no “Mid-Atlantic” or Southern states. These states also include the full range of GOP voter percentages. In other words, a group of states that manages to be simultaneously incredibly diverse, representative of the GOP as a whole, but unrepresentative of the nation as a whole.
By isolating these states, and sorting by GOP percentages, we get a much clearer picture of the impact of sexism on a political landscape where racism is far less of a factor.
· As expected, both Obama (Blue line) and Clinton’s (Red line) vote totals go down as the percentage of Republicans increase, with Clinton’s totals declining a small bit more than Obama’s
· Obama’s overall totals (Blue line) descend slightly more rapidly than his White vote (Orange line) suggesting that Obama loses support among other etchnic/racial groups as the GOP percentage increases in states few black voters.
· The Orange triangles are more widely dispersed than the Blue and Light Blue triangles, which means that Obama’s loss in total votes is more directly related to the increase in GOP voters, than to racism qua racism.
· Clinton’s (Pink line) and Obama’s (Light Blue line) male support also descend parallel to each other as GOP percentages increase, and their associated data points are grouped close to their respective lines, suggesting that in states with few Black voters, male sexism and misogyny is consistent throughout these states.
· The average difference in male support in these states is 4.9 points in Obama’s favor. (see TABLE 1)
It is in the middle range of “Black voter” states where racism and sexism interact. This is by far the most “representative” and “diverse” group of states: Although the Far West is under-represented (2 of 8 states) here, and this group contains all 5 “Mid-Atlantic” states, it also has 5 of 10 North Central states, 4 of 10 Mountain/Plains states, 3 of 6 New England states, and 6 of 11 Southern States
And by sorting these states by Obama’s White vote percentages, we get trend lines that are not very consistent with 50 state trends, but which reveal why those 50 state trends exist.
· There is a slight increase in overall Clinton’s male support (Pink line) as Black males make up a larger percentage of the male vote.
· White voter resistance to a Black candidate results in Obama losing overall male support (Light Blue line)
· As a result of these trends, Obama’s usual advantage over Clinton among males disappears.
Both Clinton and Obama’s overall vote totals decline when there are a “moderate” percentage of Black voters, but Obama’s totals (Blue line) decline more severely than Clinton’s (Red line) in this subset of states, to the point where Clinton does better in the states with higher percentage of Black voters.
· White Clinton’s male totals increased, her overall vote totals (Red line) decreased, suggesting that increased support of Black women was outweighed by decreased support of White women.
· When this data is sorted by female support for Clinton, the decline in female support for Clinton seems more related to the increase in GOP percentages rather than an increase in the Black vote.
· However, the question remains whether GOP party affiliation of women is driven by racism itself.
· The approximate 12 point drop in Obama’s totals (Blue line) is more than twice that of Obama’s approximately 5 point decline in male support, which means that resistance of White women to voting for Obama overcame additional support that Obama receives as the percentage of Black women increases.
· The steep decline in White support for Obama (Orange line) is much greater than the trend line showing the increase in Black voters as part of the overall electorate (Black).
· White support for Obama also drops far more rapidly than GOP percentages (Grey line) in the states rises, making it clear that race itself, rather than party affiliation, is the factor driving Obama’s White support down as black voter percentages increase.
· While it cannot be determined from the data alone whether GOP affiliation itself is driven by racism, the fact that the Black voter % (Black line) and GOP voter % (Grey line) rise at close to the same rate suggests that there is a correlation.
The trends found in the states with the highest percentages of Black voters are similar to those found in the “moderate Black voter percentage group.” This group includes 7 of 11 Southern states, and 2 of 5 Mid-Atlantic states – and unlike the “moderate” group, is neither representative of the US as a whole, nor very diverse.
But in this group, Obama’s white support (Orange line) declines only slightly more than the percentage of Black voters (Black line) or GOP (Grey line) percentages increase. Much of this is due to the already low white support that Obama has in these states.
But the overall “Republican” nature of most of these states means that both Clinton’s (Red line) and Obama’s (Blue line) total votes decline.
It is the impact of racism that leads to the sharper decline of Obama’s totals, especially in the South. And while there is racism evident in all states with more than a minimal Black voter percentage, not all states are the same.
The impact of racism is also felt in the differences in male support.
Once again, we see the decreasing difference in Obama’s (Light Blue line) and Clinton’s (Pink line) male support as the percentage of African Americans increases. But this chart also demonstrates that in the absence of racism, we see the true impact of sexism. Indeed, the two “non Southern” states in this group, Delaware (triangles at the far left) and Maryland (second highest Light Blue triangle, and the Pink triangle below it) show the greatest differences in male support for Clinton and Obama.
SUMMARY:
· While there is a “political” gender gap – women tend to vote more for Democrats, and men for Republican, the gender gaps found in the SUSA poll were heavily influenced by sexism and misogyny, and racism.
· In those states where there are few African American voters, race does not generally appear to be a factor, and as a result the impact of sexism and misogyny is easy to detect – and that impact is significant.
· As the percentage of African Americans increases in the electorate, the appearance of the impact of sexism and misogyny is reduced, as white voters who were willing to vote for a white female candidate switch to McCain when the alternative is a black candidate.
· The evidence of sexism and misogyny also dissipates as Black voter percentages increases, apparently due to the traditionally strong support of African American voters for Democratic candidates. (Exit polling consistently shows a smaller gender gap among Black voters than among White voters.)
· Just as racism disguises the full impact of sexism and misogyny in the electorate, so too is it likely that sexism and misogyny disguise the full impact of racism.
While this study makes clear much of how racism itself affects the electoral landscape, the purpose of this study has been to gauge the potential impact of sexism and misogyny on the 2008 Presidential election. A future study will look more closely at the how racism, and positive racial identification, could impact that election.
Next: Sexism is an Issue, Not Just a Factor
[Cross-posted at Corrente.]
Related posts:
- Can white liberals keep their eye on the prize when racism comes a knockin?
- Recovery? Romer Contrasts GDP Growth with Job Loss
- Election 2009: The (Quasi) National Landscape
- Election 2009: Election Night Thread #1; Polls Now Closed in VA
- Louisiana Justice of the Peace Keith Bardwell Tries to Justify His Racism
Spotlight
Support this site!
Subscribe to the newsletter
Advertise on Firedoglake
Send
us your tips
Make us your homepage
About Firedoglake
Advanced search
Hi Paul
You really need to set this material out in the form of linear regression, which are easier to interpret because they include the controls. I don’t dispute the results of your graphs, just criticizing the presentation.
Paul how does the fact that so many women do not vote effect this. Per the Women Voices Women Votes – http://www.wvwv.org that 20-40 million women do NOT vote.
If this demographic was an active voter, do you think the statistics change?
A high number of this demographic group are of color… how would this effect this too?
Hello Paul.
Thank you for this series.
Owww the charts, the charts. Ok Paul, you know the drill by now…executive sumary please (i’m am statistically challenged)
thank you, Wobbly — those of us who are regressionally challenged appreciate the shout out for us as well.
I’ve been totally fascinated by this series, I just need it broken down. My field is literature/language not math or math like things LOL
O/T..We need a chart for public mass consumption showing that the more Republican influence in the US, EVERYTHING else that is good for us declines. Please don’t let Rich white males see these charts…they will be used to raise money for Republicans candidates.
My guess is that there would be some change, but it would be a matter of degrees.
The sample is of registered voters — do you know what percentage of the women who don’t vote are registered? (and you probably know better than I, but aren’t single women the group least likely to vote?)
is that as the percentage of black voters in a state increases, the presence of sexist and mysoginist attitudes decline.
I would have expected African American voters who tend to be poorer and less educated to have higher sexist and mysoginistic numbers. The expected GOP attack about how African American voters tend to more conservative on social issues has just suffered a hit.
I think this fact will prove embarrassing to White and Hispanic GOP talking heads.
*slapping head* okay I see the summary above. Umm, I’ll go read it now *g*
executive summary.
All that stuff that came before this… is didn’t tell the real story, which is that there is not a linear progression in the way that white vote behavior changes with the increase in registered black voters, and the subsequent decrease in gender gaps.. Rather, there seems to be a “tipping point” in the 4-8% range where race will become a factor, and from there it progresses. (and of course, it doesn’t have to be 4%….there are states with as little as 2% black voter registration that show clear signs of resistance to supporting a black candidate.)
and hey, you think the charts are bad, try reading this stuff without charts — at least this has colors and stuff to keep you from wanting to blow your brains out from sheer boredom! ;)
i still don’t see how this can be so broadly applied when other factors could account for these numbers.
Yeah red states do tend to have lower wages, less education, higher divorce rates, pollution and I wonder if per capita higher gun death rates, higher rates of high school dropouts,less people with healthcare, shorter life spans etc?
Interesting that you say this
being that women have played and continue to play such a prominent and important roles in black communities, i wouldn’t expect to see high mysoginistic numbers.
actually, its the evidence that “declines” in the form of shrinking gender gaps and shifts of votes from Obama (in McCain v Obama) to McCain (in McCain v Clinton). None of this data measures the true levels of sexism and racism, and its impact on overall voter behavior — my personal opinion is that sexists are more likely to be racists and vice versa (just an opinion, no data to back it up) and that you’d see a lot more evidence if one Democrat was a white male, and the other a woman or african american.
(I mean, its not like there is no racism in places like Idaho, its just that these people aren’t going to vote for a Democratic woman or an African American, so there is no “switch” in their vote that shows up in polling data. )
LOL, ummm I’ll take your word for it
its important to keep in mind that this data concerns political behavior/candidate preferences, and its not a measure of overall social attitudes. Exit polls show a consistently smaller gender gap among African Americans that whites — but just because African American men and women aren’t stupid enough to vote for Republicans doesn’t mean there is less sexism in the AA community.
being that women have played and continue to play such a prominent and important roles in black communities, i wouldn’t expect to see high mysoginistic numbers.
I think that having 1 in 10 African American men in prison at some point in their lives has forced African Americans to be less sexist since the women often have to run the family alone.
If anything this trend should only get stronger.
Dang!
I’ve noticed this set of graphs etc for the past week or so, still don’t see it and I do have a background in statistics. These are like some kind of pointillist Rohrschach Test. Christ, a straight line in these point clouds seems to be more a function of a good straight edge than the association of parameters. Maybe it’s a function of trying to illustrate too much in a single chart.
It goes further back than that…..slavery and the breaking up families
Good points about charts…eye catching is always good.
well, you’re free to make your own analysis of the data by downloading http://www.glcq.com/election08/chart_3_users.xls which has all the data, including all the demographic data for each state. you can even change the charts themselves by plugging in different data into the chart data table, and then sorting to your heart’s content. (Please save the spreadsheet before playing with it).
1,813 DAYZ AND THE KILLIN’ GOEZ ON AND ON AND…
Citizen Paul Lukasiak and the Firepup Freedom Fighters:
Let me see if I have this right: 1. Race is both an issue and a factor in our society and in our elections. 2. Gender is both an issue and a factor in our society and in our elections. 3. The impact of each (in elections anywayz) changes based on the percentage of black population. 4.While the impact of gender is smaller as a consequence to elections, the impact of race grows as the percentage of Blacks grows. Now is that right so far?
What is it you are tellin’ us with this calculus of race and gender? Is there anything here that we don’t already know or intuit? How about a straight forward articulation of your thesis…jest what conclusions are you willin’ to draw from all a this? We already know that race and gender are used to divide our pluralistic society and are effective in keepin’ forces in power that do not represent the interests and politics of a majority of people. We already know that race and gender are used effectively to keep any kind of political consensus from bein’ forged electorally. We already know that both race and class are used to keep class interests from becomin’ visible and therefore determining in out elections. What is it that you would have us learn from this that we don’t already know?
I really love the way your mind works and how deep you ken penetrate the overwhelmin’ volume of data but it’s time to come back to the surface and take a deep breath and help us create some meaning out of our understanding of behavior in order to influence it (or not).
KEEP THE FAITH AND DON’T BE AFRAID TO SAY WHAT YOU KNOW AND ADMIT WHAT YOU DON’T!!!
p.s to moderator
‘…We already know that race and gender are used to keep class interests from…” Comment #25
well, its really close for someone trying to interpret my turgid prose ;).
When I’m doing this kind of thing, I try to avoid reaching conclusions — thinking in terms of conclusions can make you see things that aren’t there, and miss stuff that is there.
The thing is that I’m not writing about politics — I’m using political polling to write about the nature of American culture. Unfortunately, the political implications wind up being unavoidable.
And anyway, I don’t write well, and my opinions are far from unique, which is why I generally don’t include my own opinions when I write stuff. Plus, if someone disagrees with your conclusions, they are more likely to disregard the evidence that helped you reach those conclusions.
1,811 DAYZ AND THE KILLIN’ GOEZ ON AND ON AND…
Citizen Paul Lukasiak:
OK Paul, let me try a different approach, sometimes my questions ken force answers into hidin’. First of all, your writing skills are more than adequate to present your own conclusions all ya hafta do is the hardest work…create some meaning from the data. I had a graduate advisor who told me once that the most important skill in research is knowin’ when to quit and attempt to make sense of the data. What you have here is a tremendously important cluster of data and each graph makes a statement of some kind. Understanding that there are a lotta variables that you haven’t included in your calculus (most importantly CLASS) the data you have messaged here should be able to tell us something about mass behavior in elections and political problem solvin that can guide us in our attempts to break through the institutionalized siege walls that the rulin’ elite have created around our electoral system.
So, don’t be afraid, Brother Paul, it’s like a really good dissertation – it’s gunna piss off a lotta people but you have the chops for it. Now what is it that the data tells us that we ken use ?
KEEP THE FAITHAND REMEMBER THAT FEAR IS JEST A WORD AND TRUTH AIN’T GOIN’ NO PLACE!!!
actually, I did check the data against class (at least to the limited extent that the data set allowed, which were three income variables). And Obama’s white vote totals do go down more than Clinton’s as the percentage of the electorate making under 40K increases… but 0-40K is such a wide range — and 40K means something completely different in different parts of the county — that to try and draw any conclusion from the data would be very tough.
well, one use would be to use it to attack the media when it engages in “dog whistle” sexism and racism. (I’m really pissed at Clinton for trying to make an issue of the whole “bitter” thing, because (unlike the 3AM ad) there is a ‘dog whistle’ subtext to the criticism. I doubt if its intentional — Clinton saw an opening and she took it without thinking it through, and the media is playing it up for all its worth.)
Paul;
You must be glassy-eyed poring over all that. I do less extensive research and find it impossible to get blog readers to wade through the specifics. You might want to consider doing a separate page that’s this thorough, then focus the main post on a handful of points, a couple of illustrative graphs, a solid summary, plus a link for those in the wonkerati to explore more if they want to. (Just an unsought suggestion, well-intended).
In the primaries, I’d already deduced that Obama’s won every state that held a population of at least 20% Blacks. And that Clinton won every state with 15% or more Hispanics, except Colorado. (And New Mexico, the highest of them all at 44%, proved surprisingly close).
I understand your study focuses on anti-Black racism and sexism/misogyny, without direct use as predictives (as yet). So to my questions:
1) Typically, there’s a number of states so blue or red that they only shift in landslide years. From your data, do you see any red or blue states that could possibly shift on the strength of racism alone?
2) Also, ME, NH, PA, OH, WI, IA, MO, AR, FL, NM, and OR have the highest potential to swing either way. Some include MI, WA, TN and LA in that list, but it seems WA and MI have been trending fairly steadily blue while LA and TN trend redder. Among other possibles, let’s leave AZ out for this discussion because it’s extremely rare for a guy to lose his home state. However, MN, CT, NJ, VA, NC, CO and NV are not so hard to envision as within reach.
So that’s 22, with CT, NJ, VA and NC the longest shots to shift the other way from their recent patterns.
Have you extrapolated your findings and applied them to these swing states to try and determine:
a) where racism is likely to make Obama more disadvantaged, and
b) where it’s so minimal that it shouldn’t pose big difficulties for Obama to win certain states?
Or is the racism so consistent throughout that Obama could conceivably lose most or all of the swing states?
If you haven’t taken the research that far, do you plan to?
(Thanks for all the hard work. Both my empathy and sympathy are with you!)
1,811 DAYZ AND THE KILLIN’ GOEZ ON AND ON AND…
Citizen Lukasiak:
Thanx for the response…I think that you are bein’ FAAAR too charitable to Mrs. ClintoMcCain in both the 3AM ad and the “bitter” thing. I believe that she knew EXACTLY what the subliminal responses were gunna be ta BOTH and that the corporate media would see to it that those messages got through to any folks who missed it the first time (kinda like a media catapult, don’tcha know).
Further, I don’t think there is a single race or class attack that isn’t intended in this election and I thank you for presentin’ the data that makes it hard to believe that those who make these attack ads don’t understand what the reaction is gunna be.
KEEP THE FAITH AND PASS US MORE A THIS STUFF BUT COME OUT ONTO THE LIMB WITH US PAUL…IT AIN’T GUNNA GET OUT THERE IF YA DON’T KICK IT DOWN THE ROAD!!!
mixing Paul, stats and Norske into one – is working for me. thank you all
I haven’t done that kind of analysis, but I plan to — my focus was on sexism/misogyny because it jumps out at you simply by looking at a few of the state polls in the set. I actually never even planned on looking at “50 state” trends…I was looking at “swing state” first (states that could go either way in theory), then decided to expand it to “battleground” states (states like California and Texas which are all but foregone conclusion — but the GOP will campaign in Calif. and the Dems in Texas, because if they don’t then the other party can concentrate their resources on swing states). At that point, I had already cut and pasted and formatted the data from about 30 states, so I just went ahead and did them all.
the issue for me wasn’t whether there was evidence of sexism, but whether it could be quanified in some way. And it could be…but the results were surprising (like the data showing the South as having the least evidence of sexism).
I never had any intention of writing about racism in relation to sexism — basically, this piece started out as a footnote, then an appendix, to the third (and supposedly final) installment in a series on the evidence of sexism. Basically, the I just wanted to say that I could account for some of the variances in the evidence of sexism — that there was a relationship between the gender gap and the percentage of black voters. The original thesis was that it was the black voters themselves that were closing the gender gap — but while the size of the gender gap could be correlated to the percentage of black voters, the change in male and female voting patterns could not be attributed to black voters themselves. The change was too big for the size of the black vote — which is when I started looking at what happened to the white vote when the there were more black voters.
And that meant that the explanation was too big for an appendix.
In terms of the potential for racism to determine the election outcome, the answer is “in 49 of 50 states”, because except for Hawaii, every state is at least half white, and the maximum % of black voters is about 1/3.
If you exclude the states where racism has already determined the outcome (the south) or where racism will just add to the GOP margins (Mountain states) you’re looking at a whole bunch of states where racism is a very real potential problem.
But the evidence isn’t in this data set — its in the pre-primary polling for South Carolina, and the McCain v Democrat head to head matchups in Ohio that have been done over time. Basically, in SC, white voters only started really coming around to supporting Obama starting in Mid-December, and he had gotten to 29% of the white vote by early January. Then the whole “race-baiting” thing hit, and he lost 7 of those 29 points — and white voters flocked to Clinton — she went from 36% of the white vote to 50% in a two week period. What controversy died down, Clinton lost that white support that went to her as a result of the controversy, but only 1/7 of it shows up in Obama’s exit polls (he was at 22% of the white vote at the height of the controversy, and wound up at 24%.) The rest of that “white shift” went to edwards.
Then, there is Ohio. Obama was lagging well behind McCain in head to head matchups before the primary season focus shifted to Ohio, but by the end of February both Obama and Clinton lead McCain by 10 points. Then the Wright controversy hit… and Obama wound up behind McCain by 7 points in a little over two weeks (there are some weird things about the data that suggest that the real margin was Obama behind McCain by 2 or three points… for one thing, the party distribution of the poll shifted dramatically toward more republicans and fewer democrats, and while Clinton’s own numbers remained the same, McCain picked up 4 additional points, dropping Clinton’s margin over McCain to 6 points.)
Unfortunately SUSA hasn’t done a poll in Ohio sunce that mid-march one, so I can’t say what has happened since then. Other polls suggest that Obama has made up a lot of the ground that he’s lost, but I use only SUSA data because while their methodology leaves a lot to be desired in terms of predicting final totals because they get the overall distribution of the vote wrong, within specific categories (ie, gender, race, age, etc) their polls are remarkably accurate. (They’re polling is automated, which allows for very large samples — but they rely on that data to say what percentage of each group will vote.)
Ohio and SC tell us that Obama can attract the necessary support from white voters to win in November. But they also tell us a whole lot of his white support will evaporate very quickly if race becomes an “issue”