Get yer polls, bets, numbers & pretty graphs! Elections 2008

Once again, for those who'd like to pull their noggins out of the immediate horserace and think about the coming general election and the future beyond that, there's an extraordinary discussion with Greg Anrig on his new book "Conservatives Have No Clothes" including Eric Alterman and Hendrik Hertzberg available at the link below. You'll really want to watch the entire thing though it is 1 hour and 13 minutes in duration.
http://fora.tv/2007/09/25/Conservatives_Have_No_Clothes

New Mexico is finally (finally!) going to announce the results! Indications are that it'll be Hillary, right? Just saw a teaser on CNN, not sure when the actual announcement will be.

My god that guy blathered on. His one moment in the limelight and he wasn't going to give it up until he absolutely had to...

Hillary 73 thousand and something, Obama 71 thousand and something.

Expected, but still a little phooey. (An UNexpected boost would've been nice.)

I read that a viewer commented on CNN that the best way to deal with the problem is to divide the FL and MI delegates, half to Obama and half to Clinton. That way, the delegates are seated and still have the punishment of no effect on the campaign results.


Maybe this viewer has hit upon an equitable solution. An equal division of the total delegates from both states includes the states's voters and delegates and it treats each candidate fairly based upon the original rules of no contest in those states. The only loser would be the DNC's power play to control which states "go first" in the election process and they deserve to lose for such a bone-headed punishment of them.

Ohio, a tough nut to crack for Obama?

Based on the admittedly sparse polling data, Soz has her work cut out for her..

In fact, I'd say that Obama has a better chance, perhaps a much better chance, of winning Texas than Ohio. He's polling better there, and the intricacies of the system there work in his favour. Ohio, on the other hand, looks tricky.

First, the straightforward polls. Pollster.com, as always, has the list. Most of the polls that were done over the past year or so were done by Quinnipiac, a pollster with a good reputation, which hasnt done badly so far this year either. There have been few polls by other pollsters - one each by Rasmussen (Feb 08), Survey USA (Feb 08), Strategic Vision (Sep 07) and the Univ of Akron (Jan 07).

Here's the graph:





The Quinnipiac polls showed a relatively small lead for Hillary (10-13%) when Obama was just getting into the race, in Feb-Mar 07. But that lead increased to somewhere around or over 20% in April-July, and settled at a consistent 25-30% between August and December. The last poll out, this month, still has Hillary ahead by 21 points.

In comparison with other states, this looks like a pretty steep hill to climb. Take Michigan, for example. There, Hillary's lead over Obama oscillated between 5% and 16% from February through to July 2007, then increased to around 20% in the remainder of the year. In Missouri, the three only polls done in 2007 (two of which by Hillary-favouring ARG) had Hillary up by 12%, 15% and 25%. And in Wisconsin, coming up next, Hillary's lead last year mostly hovered in the mid-teens. So in all cases, Hillary's lead was smaller than in Ohio.

The preference for Hillary has apparently been more entrenched in Ohio than in Michigan, Wisconsin or Missouri -- or, if barely, Pennsylvania for that matter (more on that later).

There is a more circumstantial sign of that too. You know that I have been tracking these match-up polls (Hillary vs Huckabee, Obama vs MCain, etc) through the autumn and winter. One of the things I was scratching my head about is why Obama was doing badly in Ohio. I'm not up to scratch on January, so I'll leave the Jan items out, but take the data from October through to December.

In a string of nine polls that came out for Ohio, Hillary's score against Giuliani, then still the frontrunner, ranged from leading him by 9 to trailing him by 2. On average, she led by 2.5. The seven polls matching Obama up to Giuliani, meanwhile, varied from Obama leading by 6 to trailing by 8, and on average he trailed by about 1.5. So Hillary did 4 points better against Rudy than Obama did - whereas at the same time in Wisconsin, it was Obama who was doing better.

Same against McCain. In eight polls pitting Hillary up against Mac in Ohio, on average she trailed by almost half a percentage point, with the numbers ranging from a 10-point lead to an 8-point deficit. Obama's numbers in the four polls that pitted him against McCain, meanwhile, ranged from a 4-point lead to a 15-point deficit, averaging out at an 8-point deficit. Doing an apples-to-apples comparison, Obama did 14, 3, 9, and 6 points worse than Hillary in those four polls. Come January, and Survey USA had Hillary trailing McCain by 2, and Obama trailing him by 7.

Again, this was a deviation from the proportions in other states. In Wisconsin, Obama did about as well as Hillary, and in Pennsylvania, he did barely worse. Same against Huckabee - Obama matched up noticably better than Hillary against the Huckster in Wisconsin - but noticably worse in Ohio.

I'd say, in relation to each other, of the upcoming larger states Wisconsin should be easier for Obama than Texas, Pennsylvania and Ohio - well, thats the common wisdom already, of course - but I also think that of those three, Ohio should be the hardest, and Texas actually the easiest.

I just applied for an actual job in the Obama campaign! It's scarily perfect for me -- "coordinate administrative aspects of the Obama for America Ohio web presence" -- but I don't know about hours ("15 to 40") and whether deafness will be an issue.

I so want it though!

We'll see.

Gonna work my butt off anyway -- even if there's no way Obama can win here, will try to shave off every percentage point I can!

As predicted, it didn't take long for the lines to cross.

Pollster.com average of polls.



No sign of the slope decreasing yet, though it should start to soon!

OBAMARAMA

Cycloptichorn

You go girl!!

Gallup Dailies.



Cycloptichorn

Pennsylvania - Democrats

As above for Ohio.

First, the straightforward polls: Pollster.com has the list. Again, most of the polls that were done over the past year or so were done by Quinnipiac. But aside from a smattering of individual polls (ARG, Susquehanna, and three polls by Franklin & Marshall College), there was a series of four polls done within half a year last year by Strategic Vision.

Here's the graph:





We see a relatively balanced race from February through July 2007, with Hillary leading by 11-22% in the Quinnipiac polls, and by just 10% or so in the SV polls. Both candidates are roughly stable, with Hillary mostly scoring in the mid-30s, while Obama is either in the mid-teens (Quinnipiac) or mid-20s (SV).

Things then widen up significantly, with the margin becoming 18% in a September SV poll, and hovering around 30% in the Quinnipiac polls between August and December. Basically, what happened was that Obama kept stagnating, while Clinton's support grew, to the low-40s and then the mid-40s.

A Frankin-Marshall poll in mid-January shows the gap tightening a bit but doesnt deviate all that much from the pattern, with Hillary at 40% and Obama at 20%.

But a Quinnipiac poll this February shows rapid change in the numbers. It has the gap decreasing further, but still very much in the double digits (16%), as both Clinton and Obama gain significant extra support. Obama gains more extra support than Clinton - but Clinton, in her turn, has already breached the 50% border, 52% to 36%.

However, in December Edwards was still in the race. The February results presumably include only Hillary, Obama and undecideds. Given recent trends, it seems reasonable that Obama may capture a large majority of those that remain and, as well, may erode some of Hillary's support. Pennsylvania could be a close race in the Democrat primary.

Sounds perfect. Let us know if you need a couple of hundred references.

Did Obama finally break the working class code in Virginia and Maryland?

Throughout the primary campaign, Obama has faced a popularity deficit among working class voters. He usually does better among those with higher incomes than among those with lower incomes. A bigger chasm opened up from the start between those with higher education and those without. In fact, the higher the education category, the more Obama voters you'd find, the lower the educational category of voters, the more Hillary voters.

Obama and class: theories in abundance

This phenomenon has been blamed on a variety of different reasons, in this thread too. Is there something about Obama's style and rhetorics that appeals more to upper middle class voters than to working class people? Is it just a question of information trickling down slower to lower education groups, who tend not to be as well informed about day-to-day political news - or as intensely interested in it - as college graduates and especially postgraduates?

Is it the contrast in priorities set by Obama, who spends a lot of time talking about postmaterialist and process-oriented causes like changing the tone of politics and tackling the lobbyists, which may be a kind of "luxury issues" mostly valued by voters who dont have to worry about their daily income, and Hillary, who has from the start laid greater emphasis on bread-and-butter issues?

Is it that those who are in a more precarious stratum of society are more inclined to go for the safer choice, to go for the person they already know (well), and tend to attach a greater importance to experience? Whereas those who have the greater self-confidence of those with more education and/or income are more willing to go for the adventure of a relatively new politician, and of a grander mission of changing politics per se - to make that leap of faith?

Is it a different perception / experience of how the world works, which makes those who have had college education more willing to believe that dialogue, respect and a new tone in politics can make good things happen, while those who have experienced their share of gritty struggle in life are more inclined to accept Hillary's appeal not to be naive and elect the proven fighter you will need to really get things done?

Education and income divided electorates

The theories have tumbled over each other - and in some of the above points, you would recognize Sozobe's voice, in some mine. But first let's recap the data.

Before the Potomac primaries, Obama's share of the votes among those without college degree varied wildly, from 19% in Oklahoma to 64% in Georgia. Among those with college degrees, his vote varied from 32% in Arkansas to 69% in Illinois. Obviously, the differences from state to state were greater than the differences between the two groups within any state.

But the difference in his appeal between the two groups within each state were real. In the starkest case, Oklahoma, he got only 19% of those without college degree, and 50% of those with degree - a 31-point gap. In New Mexico, the gap was 24 points. In Massachusetts, Connecticut, Delaware, Missouri, Tennessee and Utah, it was 14-18 points.

The result was that Hillary could win one group, Obama the other - which was true in no less than nine states (MO, DE, CT, AZ, NM, TN, CA, NH and OK).

More tellingly still, there were only five states in all (for which exit polls were held, I mean - they were not conducted in most caucus states), in which he did got a majority of the voters without college degree. And those five included Illinois, his home state, and four states in which African-Americans made up at least 48% of the primary voters. A larger proportion of blacks tend to not have a college degree, whereas of course they are Obama's staunchest voting bloc, so a large share of black voters will mostly cancel out the education gap that exists otherwise.

In fact, there had been just one (1) state in which Obama led Hillary among those without college degree that wasnt either Obama's home state or a state where blacks made up about half or more of the voters, and that was Utah.

Here is what the table looked like:





(Note: what I dubbed the "class gap index" is the difference between the lead or deficit of Obama had on Hillary among those without college degree and that which he had among those with a degree.)

The contrasts were less stark when it came to income groups, but still pronounced.

Roughly dividing the electorate in just two groups - those who earn more or less than $50,000 - the split does not appear in the states where Obama did best. In Georgia, Illinois, Louisiana, Alabama, he did as well among those in the lower income half as among those in the higher income half. But note that again, this is basically Obama's home state plus states where black voters made up about half of the electorate.

There were other states where he did roughly equally well - or in some cases, badly - among both income groups too: Arizona, Nevada, New York, Iowa and Arkansas. (Missouri too, but the numbers for that state are dodgy. I checked both MSNBC and CNN, but the totals dont add up with the actual results, Obama's numbers add up to a much better result than he actually had. So I think the MO numbers were just never adjusted to conform to them, or the final adjustment wasnt uploaded.)

But in many states at the bottom end of his results, the gap opened up. In Oklahoma, he got 40% of those earning over $50,000 - but just 23% of those earning less. A 17-point gap. In California, the gap was 12 points, in Tennessee and New Mexico it was 11. In Connecticut and Utah, two states Obama did well in overall, the gap was 10 points wide, in Massachusetts 9.

Here is what that table looked like:





Virginia and Maryland - harbingers of a new phase?

But the exit poll data for the Virginia and Maryland primaries suggest that something might have changed. Of course, it would be foolish to base any definite conclusions on two primaries in neighbouring states -- there may be regional factors at play. But the comparions is nevertheless striking.

In Virginia and Maryland, Obama received 63% and 60%, respectively, of the votes of those without college degrees. This instantly put both states in the top four of his states in this category, along with only Illinois, his home state, and Georgia.

They are also immediately the only two states in his top 7 on this score that werent either his home state or home to a black population that made up half of the voters. (In Virginia, black voters made up 30% of the Democratic primary voters; in Maryland, 37%. This puts them closer to Delaware and Missouri in that respect than to Georgia, Alabama, Louisiana or South Carolina.)

Mind, they were just two very good states for him overall. He also did very well among those with college degrees, getting 65% and 57% of the vote among those. That puts Virginia into a shared third place for him in that category, but Maryland only in an 11th place.

In short, the "class gap" almost disappeared in these two states. It was just 5 in Virginia - lower than in any previous state except for South Carolina; and in Maryland, Obama actually did better among those without college degrees - which is a first.

Here's the updated table, with VA and MD highlighted:





The same things catch one's eye in an updated table by income. Obama won 62% and 60%, respectively, of the under $50,000 voters in Virginia and Maryland, which instantly put the two states in his top 4 in that category - again, behind only Illinois, his home state, and Georgia. And again, it makes them the only two states in his top 7 in that category that are not either his home state or home to a black population that makes up about half the electorate.

Again, Virginia had a low class gap - Obama led by 26 among those without college degree, by 31 among graduates; and Maryland is actually the first state so far in which my improvised class gap index is negative: his support was larger among lower-income voters than among higher-income voters.

Here's the updated table for income:

Here's a neat graph to illustrate that:





In short: every red dot represents the result of Obama in one of the primary states (for which exit poll data are available); every blue dot represents a Clinton result. The further right a dot is located, the higher the percentage of the vote the candidate got among those without a college degree; the higher up a dot is located, the better the candidate did among college graduates. The candidates' results in Virginia and Maryland are indicated in pink (Obama) and lightblue (Clinton) dots instead.

I fully realize, nimh, that I am a pain in the butt for being critical of your pretty charts sometimes. I admire the hell your knowledge of elections here and everywhere else (Spain and Italy coming up?).
Re your post about voters in VA without college degrees going to Obama, I would respectfully suggest that that may be a function of the support Obama has amongst the youthful voters (18-21), too young to have degrees yet, rather than a shift amongst the older populace without degrees.

Virginia and Maryland: trendbreakers on race, too

Above, I outlined how VA and MD seemed to show Obama breaking through among the less educated, less prosperous voters in a way he hadnt done anywhere else, outside states where blacks made up half the electorate.

If that trendbreak is repeated in Wisconsin and then Texas and Ohio, that is a big thing, which could well play a decisive role in giving him the nomination.

But there is another way in which the Potomac Primary broke through the roles and trends established in earlier primaries, and that was race.

In Virginia, Obama got 52% of the white vote. That instantly made it his fourth-best result among white voters, among the states for which exit polls are available at least. Only in his home state Illinois, in New Mexico and in Utah had he done better among whites.

In Maryland, he got 42% of the white vote. This is less exceptional, but still better than both the average and the median; it's his eight-best result among white voters out of 24 states.

Now you may remember that almost a week ago, about 10 pages back in this thread, I analysed the theory that there was a correlation between how well Obama did among white voters, and how large an African-American population there was in the state:







Right?

Well, Virginia and especially Maryland totally broke the mould.

In Maryland, 28% of the population is black (or was, at the time of the census in 2000). And yet, Obama got a clearly above-average share of the white vote too. In the graph here, that puts Maryland right next to Georgia, previously the one single state with both a large black population and a high score for Obama among white voters.

In Virginia, 20% of the population is black - a smaller share, but still one that's higher than the median, putting the state in the top half of the graph. And at the same time, as noted, Virginia became Obama's fourth-best result among whites.

Here's an update of that graph. Virginia and Maryland are marked out in pink:





An added nuance to the original theory? Or did something actually change in how Obama's appeal is working out? Considering the simultaneous trendbreaks on education and income, I'm assuming the latter. And this must be good news for the Obama campaign, especially with a number of races coming up where African-Americans constitute significant minorities, but not ones large enough to by themselves sway elections - in short, races that fit the tricky middle outlined in the original Daily Dish post.

It is a great analysis, but it has an inherent flaw that you already mentioned. There isn't much representation of the votership demographics from exit polls in the caucus states and that is where many of Obama's successes have been. It makes it difficult to give much validity to any multi-state conclusions that make comparisons based on analysis of the two candidate's successes and failures.

The other missing piece is the effect the timing of Romney's withdrawal and McCain's near victory had on the Potamic Primary demographics. Did those events that other states did not have benefit from, free up crossover votes that skew the trend?

We need more data to lengthen the trend lines....

Thank you!


No, that wouldnt be able to explain it. Even the category of voters aged 17-24 made up just 8% of the voters. So those aged up to 21 probably made up just 4% or so of the voters.

Now these youngest voters did go massively for Obama - those aged up to 24 went to him over Hillary by 78% to 22%. Those are remarkable proportions even if young voters usually broke towards Obama in fairly large numbers in many other states too.

But to even help explain the trendbreak here, the youngest voters in Virginia would have needed to make up an exceptionally large share of the electorate compared to in other states, as well as voting for Obama in much larger numbers than elsewhere. They may have met the second criterium, but if they made up just 4% of the electorate, they just didnt have the kind of weight to impact the numbers in this way - I mean, we're talking about an Obama score among those without college degrees that's 17-20 percentage points higher than it has been on average!

These are good questions. Can I have permission to swipe them for discussions elsewhere too?

True, true. It's a very limited data set of course - it would have been better if America had had 250 states :wink: (I can hear campaign volunteers groaning already.. )

Then again - but this is just an off-the-cuff observation - most of the states that had caucuses were mostly white to lily-white, concentrated largely in the Plains and Rockies. And these are of course states where Obama did very well. So if exit poll data had been available for all these caucuses, most likely you would just have had a lot more dots in the bottom-right corner of the graph - which, if anything, would have appeared to strengthen the thesis.

Except, of course, that it would have been impossible to attach much significance to that, because there are many reasons why Obama does exceptionally well in caucuses. Questions of organisation, of campaigning intensity, questions to do with the more select group of participants, Hillary's near-desertion of those states - all reasons that arguably must have had a stronger role than the racial demographic make-up of the states in question. So perhaps including caucuses would also have for too much of an apples-to-oranges comparison.

(There's already a lot of apples-to-oranges involved in comparing primaries that took place at different times, some when other candidates were still in the mix, etc -- this kind of comparisons would work a lot better with one national primary with a uniform system. <sighs>)


That is very possible. I calculated earlier that in Virginia, 16% of self-identified Republicans and 68% of self-identified Independents voted in the Democratic primaries. And almost three-quarters of those Republicans and over two-thirds of those Indys voted for Obama. And since almost three-quarters of those cross-over voters were white, that will have upped Obama's share of the overall white vote.

Indys and Republicans together made up almost a third of the voters in the Democratic primaries, so that's definitely a group large enough to potentially make for a decisive trend-break.

Of course, to find out whether this is the explanation, you'd have to compare these numbers for Virginia about how big a share of the Democratic electorate Indys and Republicans made up, how large a proportion of Indys and Reps decided to vote in the Dem primary, and what percentage of them broke to Obama, with the equivalents for Maryland and a bunch of pre-Potomac primary states... Yikes!