We’ll be hitting few different topics in this polling update, most of them related to Mitt Romney’s selection of Representative Paul D. Ryan as his running mate.
I’ll look at the the effects that Mr. Ryan could have in Wisconsin specifically. And I’ll describe the short-term effect that Mr. Ryan could have on Mr. Romney’s national standing — vice presidential announcements often produce temporary but discernible bounces in their ticket’s polling numbers. (The more interesting question, of course — and the hardest to predict — is what impact Mr. Ryan will have on the race in the long run. You can find my thoughts on that in the article we posted earlier on Saturday.)
But first, time for Saturday’s regularly scheduled look at the latest polls.
Saturday’s Polls
One new national poll was published on Saturday, although it was conducted prior to Mr. Romney’s selection of Mr. Ryan. That poll, from the firm TIPP for The Christian Science Monitor and Investor’s Business Daily, gave Barack Obama a 7-point lead over Mitt Romney in the national race.
Long-time readers may remember that I am not terribly fond of this polling firm — their numbers, for whatever reason, have tended to be pretty volatile in past elections — but this is nevertheless the latest in a series of national polls showing good numbers for Mr. Obama. And as in some of those other national surveys, the trend-line is favorable for Mr. Obama in the TIPP poll, as their previous survey showed a 4-point lead for him.
The national tracking polls, also published on Saturday, haven’t been joining in the fun. The Gallup national tracker continues to show a tied race, and Rasmussen Reports puts Mr. Romney two points ahead — although Mr. Obama’s position was improved in their survey from Friday, when he trailed by four points.
Why has Mr. Obama been on a hot streak in the one-off national polls but not the national tracking polls? I wouldn’t overanalyze it — it’s probably just a statistical quirk. Mr. Obama has had a pretty good month in the polls, but he probably hasn’t really gained 3 points on Mr. Romney, as some polling firms seem to imply.
Unfortunately, we’re not going to get a clean test of my hypothesis that Mr. Obama’s national polls are due to revert to the mean, because of the presence of a new factor in the race: Mr. Ryan.
Mr. Ryan’s Effect on Wisconsin
Although it varies a lot from candidate to candidate, the typical vice presidential pick has given his ticket a 2-point bonus in his home state.
We can incorporate this knowledge into our forecast model by applying a relatively simple adjustment: I added two points to Mr. Romney’s total in all the Wisconsin polls conducted before Mr. Ryan joined the ticket (which is to say, all of them so far) on the assumption that they will improve slightly with Mr. Ryan on the ballot.
We also shifted our estimate of Wisconsin’s partisan lean two points toward Mr. Romney, which is reflected in the “state fundamentals” score that also informs the forecast. Mostly, the state fundamentals number reflects how the state voted in the prior two presidential elections, and you’d expect Republicans to do a bit better with Mr. Ryan on their ticket than in their previous election without him.
Incidentally, this is not a free lunch for Mr. Romney, or an ad hoc adjustment: We had already been incorporating this type of adjustment into the state fundamentals index for the other states that have a native son on the ballot. Delaware’s state fundamentals score is two points more favorable to Mr. Obama than it otherwise would be, since Joe Biden is Mr. Obama’s vice president. And Mr. Obama and Mr. Romney get a 7-point bonus — presidential candidates produce a larger effect than vice presidents do — in their home states of Illinois and Massachusetts, respectively. (It is not necessary to adjust the polling in these states, since the polls there already reflect the presence of Mr. Obama, Mr. Romney and Mr. Biden — but the state fundamentals score, which is not based on polling, is subject to the adjustment.)
Adding two points to Mr. Romney’s figures in Wisconsin makes him more likely to win the state, of course, although he remains the underdog there: the model now projects Mr. Romney to lose the it by 5.2 percentage points. Still, his chances of pulling off a win in Wisconsin improved to 20 percent in the forecast, from just 12 percent on Friday.
Keep in mind that we now have Mr. Romney trailing by nearly 3 points in the national race. The spread between being behind by 5 points in Wisconsin (with the projected boost from Mr. Ryan on the ticket) and being behind by 3 points nationally isn’t all that great.
In other words: Wisconsin has moved up the rankings as an important state. It was just on the boundary between being a critical state and not before, but it moves into the former category now.
In fact, the model now rates Wisconsin as the fourth-most important state based on its tipping point index, up from tenth place before, and trailing only Ohio, Florida, and Virginia. Mr. Romney would have realized a larger benefit by picking a candidate from a state that was already at the top of the list, like Senator Rob Portman of Ohio. But it’s not like Mr. Ryan is from … well, Delaware, or some other state that plays no role at all in the electoral math.
Of course, Mr. Ryan may not actually produce a 2-point bonus for Mr. Romney in Wisconsin. Mr. Ryan represents just a corner of the state rather than Wisconsin in its entirety, and his favorability ratings are lukewarm in other parts of the state. Based on a more sophisticated method that accounts for the favorability ratings, I estimated that Mr. Ryan would be worth about 1 point to Mr. Romney in Wisconsin instead.
But we’re using the default 2-point bonus in the official forecast, since that assumption had already been built into the design of the model. If the actual bounce that Mr. Romney gets from Mr. Ryan is 1 point instead, or 0 points, or 6 points, we’ll know soon enough once we start to see some new polls from Wisconsin.
A National Polling Bounce?
I wouldn’t be too eager to consume the first few polls that we’ll get after Mr. Ryan’s selection, however. They may be a bit unsettled, and reflect a temporary burst of excitement for Mr. Romney and Mr. Ryan that may not persist.
Gallup has found, for instance, that candidates typically get about a 5-point bounce in their polls after naming their running mates. But the effect seems to fade out after a week or two.
Mr. Ryan is an atypical choice for a running mate in some ways — he’s a deeply strategic pick by Mr. Romney, rather than your usual crowd-pleaser. About half of Americans didn’t know who Mr. Ryan was prior to Mr. Romney’s announcement, but the other half were about equally divided in their views of him.
And this has also been an atypical election, in that very little has moved the polls at all. So perhaps Mr. Ryan will not produce such a bounce.
At the same time, the initial steps of a vice presidential roll-out are usually staged effectively by the campaigns. Mr. Romney and Mr. Ryan will get a lot of free exposure in the news media, and Republican voters will be excited by the pick and perhaps become more likely to respond to pollsters.
In short, Democrats should probably not worry too much if the polls move slightly toward Mr. Romney over the next week or two. For that matter, Republicans probably should not worry much if the polls fail to move toward Mr. Romney, because of the circumstances of the pick. (What if the polls actually move discernibly against Mr. Romney? Well, that might be a bad sign for him.)
All of this is complicated by the fact that we will soon be moving into the period of the party conventions, which almost always do produce polling bounces. The convention bounce is predictable enough that our forecast will actually build in an adjustment for it, taking some points off Mr. Romney’s numbers when he’s basking in the convention afterglow, and the same for Mr. Obama after the Democratic convention.
The model does not make such an adjustment for the vice presidential selection — or for any other type of event apart from the conventions. Some of these things have relatively predictable effects on the polls — an unexpected attack on an American military installation on foreign soil might temporarily improve Mr. Obama’s numbers because of the rally-around-the-flag effect, for instance. But you have to draw the line somewhere or pretty soon you’ll be making an ad hoc adjustment for the fact that Mr. Romney wore an ugly tie one day.
The better solution — and this is incorporated into the design of the model — is to take a relatively long-term view of the polls at this stage of the campaign, rather than trying to chase down every trend from day to day. Apart from the statistical noise in polls, there can be short-term shifts because of news events that won’t persist for very long. This is also a reason to hedge forecasts based on polling against a steadier forecast based on economic fundamentals, as our model does.
None of which is to say that Mr. Romney’s selection of Mr. Ryan is unimportant — it may be the most important event in the election to date. But it’s simply going to take some time to properly judge its impact.
No comments:
Post a Comment