Bell Curve The Law Talking Guy Raised by Republicans U.S. West
Well, he's kind of had it in for me ever since I accidentally ran over his dog. Actually, replace "accidentally" with "repeatedly," and replace "dog" with "son."

Friday, May 23, 2008

More on November 2008

Hi Again,

Part of the worry that Democrats have is that Hillary Clinton's version of events in the last few months is correct. That the party is hopelessly divided along class and racial lines and that only she can bring everyone together. In particular, Obama simply cannot win working class voters and cannot beat McCain in swing states where there are not large number of African American voters to boost his numbers.

However, with the world FINALLY acknowledging that Hillary will not be the nominee, polls are just now being put in the field that make no mention of Hillary Clinton. This is important because including questions about Hillary can prompt voters. Professional pollsters are well aware of this "Heisenberg Effect." Asking several questions about someone's preferences between Clinton and Obama then asking them whether they would vote for McCain or Obama is likely to effect the responses.

Recent polls in swing states that put Obama directly against McCain without including Hillary in the mix of questions have Obama beating McCain in several important swing states. You can see these polls at RealClearPolitics.Com.

Here are some highlights of the averages of the recent polls:

Obama beats McCain in Pennsylvania 46.3% to 40.5%

Obama beats McCain in Ohio 44% to 42.7%

Obama beats McCain in New Hampshire 45.7% to 45.3% (48% to 43% in the latest poll)

Obama beats McCain in Virgina (according to the latest poll): 49% to 42%

Obama beats McCain in Iowa 47.3% to 41.7%

Obama beats McCain in Colorado 48% to 42%.

Virginia, Ohio and Colorado all voted for Bush in 2004. If these poll results hold or improve as Obama gets into his campaign, it would mean 42 electoral votes switching from the Republicans to the Democrats. That's a huge swing!

Bush won the electoral vote in 2004 by a count of 286 to 251. If 42 of those votes switched, the Democrat would have won by a count of 293 to 244.

These observations are also interesting because Ohio and Pennsylvania are alledged by the Clinton Campaign to be unwinnable without her. These polls show that, thankfully, Hillary's analysis is incorrect. Obama has the edge in both states despite not having done any general election campaigning there. Just wait until Obama gets 20,000 or 30,000 people to show up to a rally in Columbus (75,000 showed up to his rally in Oregon).


Dr. Strangelove said...

I smell a little cherry-picking going on here. Let's look at the RCP averages for the big three: OH, PA, FL. (I notice you omitted Florida from your list.)

Florida: McCain wins by 8 against Obama, but Clinton would win by 3.

Ohio: McCain is in a statistical tie with Obama (1.3 point gap), but Clinton wins by 8.

Pennsylvania: Obama leads by 6 points, but Clinton has twice the lead (12).

So for the big three we have a stronger showing for Clinton in every state, and the sum is:

Clinton goes 3-0
Obama goes 1-1-1

Now let's look at the other "wins" you have for Obama.

Obama only ties McCain in New Hampshire. The RCP average shows a 0.4 point advantage--clearly a statistical tie.

Obama only ties McCain in Virginia. To claim a win for Virginia, you had to forgo the RCP average and instead look at the "latest poll." But the RCP average shows McCain up by 1.3 points--a statistical tie.

And by the way, Obama is--like any Democratic nominee--losing big time in North Carolina too.

So basically, Clinton clearly has a better chance to win each of the big three than Obama. And Obama has not shown he can crack the South. That leaves you with Iowa and Colorado, which are too small.

I hope you are correct that Obama's numbers will improve when he is the only Democratic candidate mentioned. And it makes me wonder how much better Clinton might be doing if Obama were not mentioned.

Dr. Strangelove said... contains maps showing present polling data around the country. Currently, McCain is defeating Obama narrowly, by about 30 EV, while Clinton is trouncing McCain.

Raised By Republicans said...

I disagree.

The point you made (and that point is the standard Clinton party line) is that these states, espcially Ohio and Pennsylvania are unwinnable for Obama. It's not that both Obama and Clinton would win them but Clinton would win them by more. That second argument is meaningless given our electoral system.

The RCP averages I reported showed that these critical states were winnable. Also, Florida is not the holy grail of American politics. If the Democrats win every state Kerry got plus Ohio, they win the Presidency. If they win Colorado and Virginia too, Flordia just won't matter. There is more than one way to skin a cat or win an election and it is clear to me that Obama can win - and if the election were held today - would win.

Finally, Clinton started out with large leads in every primary state before campaigning began. But as teh campaign progressed, she lost most of those states. Obama is very good at winning people over to his side. He'll also have plenty of money to spend on doing just that. If I'm McCain and I look at the fund raising advantage Obama has and these poll numbers in Ohio and Pennsylvania, I'd chew my finger nails down to the quick.

So again, the assertion that Obama can't win in November without Hillary, is just false. He can.

By the way, the only shows the most recent polls, not the RCP averages. Besides, it is not true that the map on that website shows McCain beating Obama by 30 electoral votes. It says that McCain has 272 electoral votes (assuming the most recent polls hold and he wins every state where he has even a tenth of a point advantage) to Obama's 242. But there are 24 electoral votes from states where the polls are tied. Even so, if the argument is that Obama can't win, these numbers don't show it. They show him neck and neck with McCain despite both Clinton and McCain attacking him for months with no one attacking McCain.

The Law Talking Guy said...

The maps at are interesting, but somewhat academic. First, they sort of cherry-pick their polls over there. Many polls reported by other sites aren't in their calculus. Figure them all in, and the picture becomes clearer: in most states, it's way too early to tell, and the polling results are scattered and contradictory. Hillary wins Arkansas and N. Carolina? Obama wins Indiana? These are not realistic outcomes.

These maps also indicate lingering antipathy by Clinton or Obama voters to saying to pollsters "yes, I'd vote for the other guy/gal." That sort of thing just fades. Look at McCain's numbers. The GOP base is getting over its Huckabitis and Romneysia. Also, neither C nor O looks like a winner now. Once Obama is crowned the winner, that aura of success will follow. With it, poll numbers.

The maps do emphasize the fact that Clinton needs to stop injuring Obama's chances with white working class voters in OH and PA, and needs to start building him up there. We really shouldn't have to wait another couple weeks for this. Just get the hell out now, before the last primaries, and let Obama take a public victory lap.

Dr. Strangelove said...

RbR cherry-picked his polls in this post, and I called him on it. The RCP averages actually show that Clinton does much better in the big three swing states, while Obama can claim no real lead in Virginia, etc.

I think Obama can pull it off. I think he can win. But right now the polls show Clinton would have the advantage in the Electoral College. That's what the conventional strategy buys you.

The site is not as good as RCP in my view, in that they only average polls from the previous week--which at this stage is usually just a single poll (no averaging) for most states. But the maps are pretty and the site also backs up what I was trying to say: Clinton does much better in the big three swing states, while Obama has no real lead in Virginia etc.

Will these numbers change? Absolutely. Will the "Heisenberg effect" and the "aura of victory" that LTG and RbR dicuss influence the polls? Sure--I trust you both know what you are talking about. But for now, the averages are as they are. You're not entitled to your own facts, remember.

Raised By Republicans said...

I did not "cherry pick" the polls Dr. S. Saying that did assumes you knew my motive.

I was trying to refute a specif argument by Clinton that Obama can't win swing states with lots of rural, working class white voters.

Speaking of people being entitled to their own facts, the Clintonistas are most guilty of that.

By the way, have you seen the website today?

Obama beats McCain.

The Law Talking Guy said... has finally updated its map as of 5/25. It now is more realistic for Obama. These maps remain fluid.

Just think of this: even without a clear opponent, McCain still can't make out a victory. Wait until the Dems unite and the "Clinton-or-bust" folks start acting rationally when they talk to pollsters!

Dr. Strangelove said...

Electoral-Vote updates their maps pretty much every day--sometimes several times during the day.

The Law Talking Guy said...

Here's the more interesting news. CA took its vote on 2/5. The polls now show Clinton and Obama virtually identical in heats against McCain. Same thing will happen everywhere as the summer goes by.

Also, does not update its polls every day. Cross-check polls available on RCP with the results on EV-Com for examples.

The Law Talking Guy said...

Dr.S - I'm not disagreeing that the polls say what they say. I'm not disagreeing that Obama has ground to make up in the polls if he is to win.

Rather, I'm disagreeing with the conclusions you draw. You say of the current map "that's what the conventional strategy gets you." No, that's not a fair copnclusion to draw from polls today. If it were October, yes, but that's not a proper inference to draw from polls in May. May polls (not to be confused with maypoles) have only a thin relationship to November voting. They're accurate to within 20 points or so. A couple months ago, Obama was doing much better in some states, and Clinton in others. These things will change around. To hold up the map today and say it represents the likely outcome of a strategy in the Fall is cherry-picking.

All the map today reflects is the result of four months of seesaw campaigning. Clinton should no more count on winning Arkansas (which has a hefty # of disgruntled Huckabites currently futzing with the numbers) than Obama winning Kansas (although he leading in some Kansas polls not long ago).

Dr. Strangelove said...

To hold up the map today and say it represents the likely outcome of a strategy in the Fall is cherry-picking.

Um, no. To "cherry-pick" means to select a biased subset of results from a larger set, usually intentionally so as to make a point. But what you are actually arguing is that current RCP polling averaging--which I quoted without selection bias--is a poor predictor of November results.

Likewise, you accused of cherry-picking their polls too, but they don't do that either. They use a different set of polling data and a different algorithm to compute their averages--but there is no selection bias. What you are actually saying is that their current maps are volatile and a poor predictor of November results. (Although when their maps changed to favor Obama, I notice you called them, "more realistic.")

I chimed in on the comments to correct selection bias in RbR's "highlights." I did not introduce any selection bias of my own. Since you certainly understand all this, I have to wonder why you would accuse me of "cherry-picking" other than for the joy of sticking in an ad hominem attack. An apology would be nice.

The Law Talking Guy said...

Here's what I see about the ev-com maps. There are actually dozens of daily poll snapshots of winter and spring '08, all over the map. None of them has any real predictive value, in the sense that there is no study I know of showing any strong correlation of poll results to November's election for presidential races before at least June.

If you select just one or two that show Clinton doing well (the recent ones), then say that it must be predictive of the "conventional strategy" in the Fall, you've made a leap I can't buy. That's a selection bias, selecting one snapshot of dozens.

You don't see it that way because you obviously don't consider today's tp be just one felicitious snapshot out of dozens of poll snapshots that could have been generated over the past few months, all with equally little predictive value for November. But that's my reading. This is an honest disagreement, not an ad hominem attack.

(In fact, I don't think ad hominem was really what you meant anyway, since the "cherry-picking" is an attack on a data selection method, not on the data because of who selected it. A true ad hominem attack would be saying that because Dr.S is a military scientist, he can't be trusted to select data accurately).

The Law Talking Guy said...

For the record, I do think the maps at ev-com are unrealistic. The idea that Clinton will sail to victories in WV, KY, NC, and AR in the Fall is hard to swallow. That seems like a residual afterglow of the primaries. By that I mean that the Democrats have campaigned heavily there, but not McCain yet.
The idea that Clinton is weak in Washington, Hawaii, and Connecticut also suggests lingering antipathy from primaries, not November's vote.

Similarly, the idea that Obama sails in Colorado and is competitive in Indiana and South Carolina strikes me as the fact that the GOP hasn't really started campaigning there yet.

The Law Talking Guy said...

See the April 1st map showing McCain walloping Clinton 304-203. There is no more reason to pick that map than this one while the Dem primary process is still unfinished and unresolved and the national campaign has not yet begun.

The Law Talking Guy said...

Or the April 12 map showing Obama winning North Dakota, competitive in Texas, SC, and Nebraska, but losing New York.

Dr. Strangelove said...

ad hominem is exactly what I meant. When you accuse someone of "cherry-picking", you are accusing them of deliberately misrepresenting statistics to make a point. You are accusing them of being intellectually dishonest in presenting data. (If "cherry-picking" means something else to you, then explain it please.)

All I did was to quote--accurately and without bias--the current RCP averages and the current electoral-vote maps. There was no bias in my selection of those websites at all. None. Those are the two resources we all look to for meta-polling data. I did not choose them because they supported my views.

I don't think the electoral-vote maps are as good as the RCP averages: I said so clearly in my original comment! But neither site cherry-picks their data either, by the way--at least, I have seen no evidence to that effect anywhere.

Those sites have some value, not a whole lot. But if you are going to quote them to try to make a point, then you should do so fairly and without bias. That was the point of my comment.

Raised By Republicans said...

You admit that the key to your accusation of me is what point you think I was trying to make, right?

I was NOT trying to make a point about Florida. I was focussing on Ohio and Pennsylvania and states like them.

Your obsession with Florida is what caused you think I was "cherry picking."

The Law Talking Guy said...

I'm gonna get persnickety here. But it ain't an "ad hominem" attack to accuse a person of bias in selecting data.

An "ad hominem" attack "consists of replying to an argument or factual claim by attacking or appealing to a characteristic or belief of the person making the argument or claim, rather than by addressing the substance of the argument or producing evidence against the claim. The process of proving or disproving the claim is thereby subverted, and the argumentum ad hominem works to change the subject." (From wikipedia). The accusation of "cherry-picking" goes to quality of the data, not the quality of the person promoting the data. That is to say, it may be insulting to say "I believe you have chosen data to fit your biases - see the other data out there that you didn't pick!" but it's not an "ad hominem" attack. An "ad hominem" attack would be "because you are biased, I doubt the data you have chosen" The crucial part of the attack was not pointing out your bias (indeed, that was not even mentioned) but the availability of alternative data.

In fact, Dr.S., you seem to have taken a different view of the phrase "cherry-picking" than I would, which may be a matter of our professional lives. In my world, as a lawyer, I expect people to deliberately choose the best evidence to support their position. The motives are not relevant. Whether it is intellectually dishonest or not is not relevant. "Cherry-picking", in my world, is not about the intellectual honesty or dishonesty of the presenter, rather it is an accusation that the evidence so adduced is not representative of the whole story.

Perhaps scientists take the term differently. But then on this blog, we always give just some of the evidence, because we have limited time and space. I never assume I'm getting the whole story. Rather, I ask whether that evidence is really representative of the universe of possible evidence. If it appears to be biased, I would say it is 'cherry-picked.' But picking cherries is not inherently bad, unless done deliberately, which we can't know and don't care about anyway.

Dr. Strangelove said...

Let's go back to the start here. When I looked at the list RbR had compiled, two things jumped out at me. First and most important, RbR selectively presented either the "latest poll" or the average, whichever suited his argument better. Second, his choice of states was also curious and amazingly, all the states he included just happened to be ones that (sort of) supported his claim. But I use the RCP site too, so I know for a fact that RbR saw both average and "latest poll" data for every state he mentioned and for many states he omitted. RbR's original list was clearly a "fluffing up" of the statistics: the RCP data did not in fact support his contention. That is what I wrote in my comment.

LTG now says he flung the term "cherry-picking" back in my face innocently, but here's what really happened: I said, "RbR biased his data," and LTG replied, "Oh yeah? Well you do it too--so there!" LTG replied to my factual claim by changing the subject into an attack on my methods (and by extension, me). If that is not technically an "ad homimem" argument, then I apologize for using that phrase. But once agian, for the record, that argument is also wrong: I reported the data accurately, without bias. The data is flawed, which I noted up front, but that is a different contention entirely.

Dr. Strangelove said...

Oh, and "limited time and space" is a lame excuse. LAME! It takes no more time or space to consistently present either the averages or the latest polls (instead of a mish-mosh.) And if adding an extra line for Florida would have been too much, one could have swapped out New Hampshire--which has only 4 EV and shows no advvantage to either side anyway.

When a supporter of X picks only data that supports X, I cry foul. That is, in fact, inherently bad if your goal is to find the truth rather than to win the argument.