Sunday, October 21, 2012

Rasmussen's Weighting

As it happens, Mark Blumenthal wrote about Rasmussen's weighting yesterday just around the same time I was writing my blog post. It was a piece about the national polls generally and Blumenthal noted Rasmussen's importance as the one firm that has done state-level snapshots before and after the second debate and the fact that they show slight gains for Romney. But Blumenthal explains these gains are related to the "dynamic" party weighting Rasmussen is using:
Rasmussen is one of the few pollsters to routinely weight its samples so they match predetermined targets for the percentage of likely voters that identify as Democrats or Republicans. The catch, as Rasmussen Reports confirms to The Huffington Post, is that its weighting targets are now adjusting on a weekly basis to match the average party identification for likely voters measured on their last six weeks of calling (after weighting for demographics, but not for party). So the party weights for the past recent week may be slightly different than the party weights the week before.

More important, the weight targets for Rasmussen's national samples grew slightly more Republican in mid-October. Although the data are published on pages available to paid subscribers only, Rasmussen indicates that the national interviews for the week of Oct. 8 to 14 gave Democrats a 1-point edge over Republicans (38 to 37 percent). The party balance for the two prior weeks, Oct. 1 to 7 and Sept. 24 to 30, was a 3-point Democratic advantage (39 to 36 percent).

In the past, Scott Rasmussen has explained that the state-level party weighting targets are derived, in part, from national numbers and the "national shifts appear to provide a good indicator" of mid-year changes at the state level.

A 3-point shift toward a more Republican identification would more than explain the one-point shift to Romney in the five states Rasmussen surveyed this week.
Now, here's the thing. IF you're going to weight by party, the shift that Rasmussen appears to have made does make some sense. There was reason to believe, in the wake of the first debate, that a poll should have a few more Republicans in it and a few less Democrats than before. But then, this is why weighting by party ID is a bad idea in the first place.

Anyway, what this tells us is that Rasmussen is shifting their party ID weights slightly over time but they are always doing so in response to last week's polling. And that's problematic. If you are looking for a snapshot of where the public is today (the point of a poll), using their mood from last week as a determinant is really problematic.

Saturday, October 20, 2012

Let's Say Rasmussen is Right

Rasmussen is not a terrible pollster but you have to understand what Rasmussen is doing in order to interpret their results.

First, Rasmussen is a robo-pollster. They don't do live-interviewer polls and this means they also don't call cell phones. This is one reason why Rasmussen has a bit of a Republican lean. In general, Nate Silver has argued Rasmussen has about a 2-point Republican lean. Their national tracking poll currently has Romney ahead by 1 point (49/48). So, a 2-point lean would put Rasmussen right about where some other national polls are.

Second, Rasmussen weights by party ID. This has two implications for how we read their polls. First, Rasmussen's polls are going to be less volatile than other polls. When something happens that changes public attitudes about the race, poll respondents don't just change their response on which candidate they support. They also sometimes change their response on their party identification. If you have a fixed number of Republicans and Democrats you're looking for in your poll, you are not going to pick up that shift even though it is real.

How does this play out? Since late September, most pollsters have shown a shift towards Romney in the national polls and, more recently, a tiny shift back towards Obama. Rasmussen's national tracking poll has remained within a very narrow range between Obama ahead by 2 and Romney ahead by 2. For the last couple of weeks, Rasmussen's tracker has been within an even more narrow range between a tied race and Romney ahead by 2. These same things are true at the state level as well: Rasmussen polling is very stable and has a slight Republican lean.

So, let's say all of Rasmussen's polls are right. Who wins? It looks like Obama wins.

Rasmussen's "Electoral College Scoreboard" has Obama ahead in states totalling 237 electoral votes and Romney ahead in states totalling 235 electoral votes. 7 states are listed as tossups. But Rasmussen has done polling in these tossup states (very recently in most of them). Here's what those numbers show:

CO - Obama +1
IA - Obama +2
NH - Obama +1
NV - Obama +3
OH - Obama +1
VA - Romney +3
WI - Obama +2

All of these polls are very close but Obama is winning in more than enough of them that if Rasmussen is right, Obama wins. If Nate Silver's arguments about Rasmussen's lean is right, Obama wins a bit more handily.

Again, we come to the same place: Is it close? Yeah. Is Obama winning right now? Yeah.

Friday, October 19, 2012

Romney's Binder Is Full of Women

At least it is according to Chris Cillizza. I saw Cillizza's article yesterday and meant to write about it but didn't get the chance. I like Chris Cillizza and I think he generally has some smart and interesting things to say about electoral and legislative politics. But not this time.

He reviews data from the ABC/Washington Post poll and finds that Mitt Romney "doesn't have a woman problem." Cillizza finds that over the last 3 polls (a fair measure) Romney has been trailing the President among likely female voters by 7 points on average. Cillizza asks us to assume Romney loses among women by 7 ...
That would be a better showing among women than John McCain made in 2008 (lost women by 13), George Bush made in 2000 (lost women by 11) and Bob Dole made in 1996 (lost women by 16). It would be roughly equal to the eight-point margin that George H.W. Bush lost women to Bill Clinton in 1992.

So, for all of the chatter about Romney’s women problems, he is currently positioned to do as well or better than every Republican presidential candidate among female voters save one: George W. Bush in 2004 who lost among women by just three points to Massachusetts Sen. John Kerry.
Cillizza is arguing that Romney "doesn't have a woman problem" because he's doing as well or better than ... the last 4 Republican nominees to lose the national popular vote. This is the literal equivalent of arguing that the Cubs are doing just fine this year because they're on pace to win more games than the Houston Astros ... who also had a miserable season.

There's another problem with this insane logic. Cillizza points out that McCain lost women by 13 and Romney's doing better than that so he "doesn't have a woman problem." But here's the problem. McCain lost nationally by 6.5. Romney is doing 6 points better among women by this metric so let's say he's doing 6 points better nationally. That leaves Obama ahead by 0.5 points. And given Obama's advantage in key swing states, a national popular vote win almost definitely means a win for Obama in the Electoral College.

But wait, there's more. Here's a different way to calculate what a 7-point deficit among women means. In 2008, the gender gap was 12 points. Barack Obama won by 13 among women and by 1 among men. If you assume a 7-point deficit among women and hold the gender gap constant, then Romney wins by 5 among men ... and loses the election by about 1 point. But the most recent ABC/Washington Post poll has Romney losing by 7 among women and winning by just 1 among men. In other words, Romney seems to have a "man problem" too! He's losing by 3 overall in that poll and it is because he's underperforming among men.

Here's the bottom line. There are slightly more female voters than male (Thank God!). Romney doesn't want to be in a situation where he's losing in the high single digits among the larger group AND there is a smaller gender gap.

Cillizza is right about one thing. Romney is not in a worse position than the four most recent Republican nominees to lose the national popular vote. Like the Cubs, he will not come in last. But he ain't on track to make the playoffs either.

Thursday, October 18, 2012

State-Level Polling

More evidence of trouble with that Gallup poll comes in the form of state-level polling that is WAY out of whack with Gallup's national numbers. This afternoon, NBC/WSJ/Marist released two state-level polls that are VERY bad for Romney. The President leads in Iowa by 8 and in Wisconsin by 6. Folks, there are two possibilities on these numbers:

1) They're wrong
2) Romney is in very deep trouble

It is entirely possible they're wrong. I would actually be surprised if Obama was ahead by 8 in Iowa right now. I do think he's ahead in both of these states but my guess would have been something more like 4-5 points in Wisconsin and perhaps a little less in Iowa.

If these polls are accurate, here's why Romney's in trouble:

1) States don't usually move by themselves. They reflect national trends a lot of the time. If Obama is ahead by 8 in Wisconsin, he's winning by something like 4-6 points in Ohio and he's winning in Virginia and so on.

2) If the Obama campaign knows they have leads like that in those states, they can move at least some of their resources from those states towards Florida and Ohio and Virginia, etc.

3) Obama's lead is the same in Iowa as it was a month ago. There has been no movement or whatever movement there has been has cancelled itself out. Obama was winning in September and this would mean he's winning now.

4) Early voting appears to be as big a disaster for Romney in Iowa and Wisconsin as it is in Ohio. 34% of likely voters in Iowa say they've already voted and Obama is winning among these voters 67-32. The poll reports that Romney is winning among likely voters planning to vote on Election Day but not by nearly enough. Similarly, 156% of likely voters in Wisconsin say they've already voted and Obama is winning there 64-35. But the race is about even there among those planning to vote on Election Day. So, Obama appears to be winning right now in these states and locking in large numbers of votes right now.

Anyway, coming back to that Gallup Poll, the situation is simple: The two sets of numbers (Gallup and NBC/WSJ) are not consistent or the result of statistical noise. Nate Silver agrees and advises that people "be careful" with the Gallup numbers.

There is something very different going on in the sampling and one of them (Gallup or NBC/WSJ) is just dead wrong. I don't think Gallup is right.

Gallup Getting Freaky

I've mentioned in the past that Gallup is not the gold standard in polling it once was. One reason is that Gallup's tracking poll has played a bigger role in defining their brand than before and there are problems with any tracking poll.

One problem is that Gallup's tracking poll includes 7 days of data while some of their other numbers (like their approval rating data) includes just 3 days of data. So, there is very often an asymmetry.

But there's other problems. As of this moment, Gallup has Mitt Romney ahead among registered voters by 1 point and they have Romney ahead among likely voters by 7 points. Both of these sets of numbers are outliers among other polls but so is the gap between the two. Romney will surely do better among likely voters than among all registered voters but a net 6-point difference between likely voter numbers and registered voter numbers is a bit on the high side.

PPP just released their tracking poll and they have the race 48-48 among likely voters (didn't see registered voters numbers but surely they have Obama slightly ahead among registered voters) and Tom Jensen indicated that Obama did a tiny bit better on the last night of the tracker (the one night that was fully after the second debate).

All of that "feels" about right to me and is more in line with where most other polls are at this point. I wouldn't get too freaked out by Gallup's tracker.

Wednesday, October 17, 2012

If Women Are Going to Work ...

... then we've gotta make some changes. That's totally what Mitt Romney said last night ... which was also in 2012.

The more I've thought about this, the more staggering it is. Everyone is focusing in on the "binders full of women" line. That was awkward. But what followed was painfully backward and ignorant. Romney said:
I recognized that if you're going to have women in the workforce that sometimes you need to be more flexible. My chief of staff, for instance, had two kids that were still in school.

She said, I can't be here until 7 or 8 o'clock at night. I need to be able to get home at 5 o'clock so I can be there for making dinner for my kids and being with them when they get home from school. So we said fine. Let's have a flexible schedule so you can have hours that work for you.

We're going to have to have employers in the new economy, in the economy I'm going to bring to play, that are going to be so anxious to get good workers they're going to be anxious to hire women.
"IF" you're going to have women in the workforce??? We do! I think there might be several! And, apparently employers are only hiring them because they're desperate. You know, because we've run out of men to hire. I'm not making this up. He said, in the "new economy," employers "are going to be so anxious to get good workers they're going to be anxious to hire women." Good God! Not that!

Wait a second Mr. 1950s caricature ... who will take care of the children (the job women are supposed to be doing)? Mitt responds it'll be okay. We'll just need to be more flexible. Women working! What's next?

I must say I'm a bit perplexed why this is not a bigger deal.

Will the Second Debate Affect the Polls?

It depends on how you understand how the first debate affected the polls. I argued here that the effect of the first debate was to energize Republican voters and to depress Democratic voters. As post-debate polls rolled in, like the Pew poll that had Romney ahead by 4, we saw more Republicans and fewer Democrats identified as likely voters.

One thing that is not happening as a result of last night's debate is some big swing among "undecided" voters. There just aren't very many undecided voters and, if you're undecided at this point, you're a low-information voter who is not paying any attention at all to these debates.

So, what did last night's debate do for Republican and Democratic voters? Republicans who were already energized about Romney will remain so. Romney made mistakes last night but he does a very good job of delivering Republican talking points related to the question asked. Republican viewers came away satisfied with that. The question is whether the debate and, more importantly, the day-after media discussion will energize previously-depressed Democrats/Obama supporters.

My sense is that the debate will do that at least to some extent. The President was feisty, was hitting the key themes Democrats want to hear about, and hit Romney in three ways that Democrats had been worried about:

1) The President reminded everyone of his commander-in-chiefyness with the Libya exchange
2) The President helped fire up women on a number of answers (Ledbetter Act, contraception/abortion, Romney's silly binders answer)
3) The President emphasized over and over that Romney is going to cut taxes for the most wealthy, will favor the most wealthy, and finished the debate by finally raising the 47% issue and, because it was the last response, Romney couldn't bat it down.

How much will this move the polls? Very, very little in my view. But, to the extent it has an effect, Obama will gain a little ground.

Second Debate - Two Key Moments

There were two key moments in the second presidential debate in my view. One was the exchange on Libya and the other was Mitt Romney's response to the question on equal pay for women. First, let's take a look at the key part of the Libya exchange:



There is a lot of focus on the instant fact check done by Candy Crowley there. But I think the key moment is actually what precedes the fact check moment. Crowley asked the President about where the buck stops on the Libya issue. Obama's response here was simply his best moment in either debate because in that moment, he demonstrated what a President looks like and made Romney look small. That Romney went on to challenge it further and actually get his facts wrong and had to be fact-checked by Crowley only emphasized it. That was a TERRIBLE moment for Romney.

The other key moment in the second presidential debate was Romney's response to the question on equal pay for women. Romney's response was very strange and awkward and the twitterverse exploded with messages about binders like this one.


But the focus on binders misses the bigger problems with Romney's response. Here's the whole answer:



Notice Romney's language here: "I recognized that if you're going to have women in the workforce, that sometimes, they need to be more flexible." He goes on to explain that flexibility means those women might need certain hours home with the kids. This was just not good at all. "'IF' you're going to have women in the workforce???" Not good.

Remember that the effect of the first debate had to do with key moments in the debate being replayed and reinterpreted as we moved along. The same will be true here and I would argue these clips are going to be playing over and over in the next 24 hours. And that will make those bad moment worse.

Tuesday, October 16, 2012

Mark Halperin ...

... is not advancing the state of knowledge.

Here's his scoop this morning:
Here’s why some Democrats are worried tonight. If Romney wins the three Southern battlegrounds (FL, NC, and VA) and OH, he is at 266 electoral votes. Leaving the other five battlegrounds unallocated, that means Obama would be at 237 and Romney would only need to win one of the remaining five states to get to 270+.

Halperin is right. Alternatively, if Romney wins CA, NY, and IL, he will similarly have a shot of winning. Or he could win CA, FL, and VA! There are LOTS of combinations!

I am dumber for having read this article.

This Is Bad News

It is one poll so I'm not freaking out about it but PPP has a poll out this morning that has Obama down by 4 nationally.

Below the topline, PPP says they found the following sub-samples on the 3 days of the poll:

Friday (38%) Obama 47, Romney 49
Saturday (39%) Obama 49, Romney 47
Sunday (24%) Obama 43, Romney 55

So, obviously, Romney's lead comes entirely from the last day of polling and that could well be just random noise. Also, 44% of respondents in the poll identified as conservative and 16 percent identified as liberal. That can be either evidence of a bad sample or just the way real Romney voters and real Obama voters are identifying themselves. There's no way of knowing for sure.

I'm not going to get all upset about one poll but I do think PPP is generally a good pollster.