Monthly Archives: October 2012

Different Forecast, Same Result: More Political Science Models

Politico’s Dylan Byers created a minor dust-up in the twitterverse today when he posted an article that appeared to take a shot at the New York Times’  Nate Silver’s prognosticating skills. Byers writes, “Prediction is the name of Silver’s game, the basis for his celebrity. So should Mitt Romney win on Nov. 6, it’s difficult to see how people can continue to put faith in the predictions of someone who has never given that candidate anything higher than a 41 percent chance of winning (way back on June 2) and — one week from the election — gives him a one-in-four chance, even as the polls have him almost neck-and-neck with the incumbent.”  Byers’ article almost immediately created pushback from others who pointed out that the win probability is not the same as projecting a popular vote percentage.

That distinction is likely lost on many Romney supporters who have been criticizing Silver’s forecast for some time now.  But while Silver’s highly publicized work has attracted many Republicans’ ire, it is important to realize that several political scientists have developed their own state-based forecast models that are every bit as good as Silver’s and have the added virtue of being completely transparent and which have been vetted by other political scientists.  Those models are, as of today, also forecasting an Obama Electoral College victory.

In my last post I discussed one such model – the one developed by Emory political scientist Drew Linzer and featured at his Votamatic website.  As of today, Linzer’s state-based polling model continues to forecast an Electoral College victory for Obama with 332 Electoral College votes to Romney’s 206. I’ve discussed some of the assumptions built into Drew’s model in a previous post.

Today I want to discuss a second state-based forecast model created by political scientists Tom Holbrook and Jay DeSart.  Their model is even simpler and more parsimonious than Drew’s. Essentially, they look at three variables: the average Democratic vote share in all trial-heat polls in the field during October, the average Democratic share of the two-party support in national polls taken in the October prior to the election, and the average Democratic two-party state vote share in the four previous presidential elections. Like Drew’s model, then, they are incorporating both a long-term factor – the previous state-level election results that provide a window into a state’s ideological leanings – and short-term factors captured by the October state-level polls during the current election year together with the candidate’s standing in the national polls.

In 2008, the model successfully called all but 3 states: Missouri, North Carolina and Indiana.  (It’s not clear to me what they did for the split vote in Nebraska.) This prediction isn’t quite in Sam Wang territory (Sam nailed everything but Nebraska, as I recall), but it’s not too shabby either.

For those of you interested in playing at home, here’s the equation for their 2012 forecast model:

VOTEi = -29.454 + .575(POLL)i + 0.57(PRIOR VOTE)i + .44(NATIONAL POLL).

As of today that model is also projecting an Obama Electoral College victory, but by a closer margin, 281-257, than the Votamatic projection. Despite the closer Electoral College projection, they still estimate that Obama’s win probability is more than 86%.  Here are the current state-by-state projections:

Obama

Romney

State

Win
Probability

State

Win
Probability

Ohio

67.8

Colorado

50.5

New Hampshire

72.0

Florida

52.7

Nevada

74.4

Virginia

55.0

Iowa

76.8

North Carolina

81.4

Wisconsin

83.6

Arizona

92.0

Pennsylvania

88.7

Missouri

95.0

Michigan

93.6

Georgia

97.5

Minnesota

96.1

South Dakota

97.8

Oregon

97.4

Montana

99.3

New Mexico

98.4

South Carolina

99.7

Maine

99.5

Indiana

99.9

Washington

99.7

Kentucky

100

New Jersey

99.8

Tennessee

100

Connecticut

99.9

Texas

100

California

100

Mississippi

100

Delaware

100

West Virginia

100

Illinois

100

Kansas

100

Maryland

100

Nebraska

100

Massachusetts

100

North Dakota

100

New York

100

Louisiana

100

Rhode Island

100

Arkansas

100

Hawaii

100

Oklahoma

100

Vermont

100

Alabama

100

DC

100

Alaska

100

Wyoming

100

Idaho

100

Utah

100

Note that according to these win probabilities, it is more likely that Obama would “win back” Virginia, Florida or Colorado than it is that Mitt can take Ohio. Indeed, those are the three states that Drew currently has in Obama’s column.  That’s a total of 51 Electoral College votes in those three states alone, and it is the difference between Obama winning 281 Electoral College votes versus 332.  So clearly these forecasts are amenable to change, even with the dwindling number of undecideds.  Given the close nature of the race, several states can tip in either direction.  However, both models suggest that, based on current state polling, Romney has a bigger hurdle ahead of him if he is going to pull this out.

I want to stress that these models use the latest polling in each state to project the winner.   As such, they tend to be more accurate than the structural forecast models political scientists issue by Labor Day based on the “fundamentals” that I’ve discussed in several previous posts.  The drawback, of course, is that these state-based models don’t help us understand why people are voting for a particular candidate.  In effect, they use current support to predict future support.  That works well if all one is interested in is predicting the election outcome, but they aren’t very theoretically satisfying.

Note that both the Linzer and the Holbrook-DeSart models are premised on the assumption that state-based polls this late in the year are generally accurate.  Is that a safe premise?  In fact, as John Sides discusses here citing research by Robert Ericksen and Christopher Wezlien, they are.  As this graph shows, the share of the Democratic vote based on polls in the last week of the election closely aligns to the actual vote percentage received by the Democratic candidate in the period 1952-2008. (If the alignment was perfect, the data points would fall directly on the diagonal line.)

So, using conceptually simple (and transparent!) methods, political scientists still see this race as Obama’s to lose. This does not mean, however, that this election is over.  The key – and as yet unanswered – question is whether Mitt can, through a combination of winning over undecided voters and gaining a turnout advantage, rope in the 1-2% more support he needs to flip Ohio, or some other combination of swing states.   Your answer to that question may depend on whether you think there continues to be movement toward Romney, however slight, during these last eight days. If there is, both models should pick this up, and adjust their projections accordingly.  If there is not, however, and this race has entered a stable equilibrium, the odds seem to be in Obama’s favor.

P.S If you are coming to this site for the first time (there’s been a lot of traffic of late) I encourage you to follow me on Twitter at @MattDickinson44 – I tweet all new posts there.

 

How True Drew? Linzer Still Sees Obama As A Heavy Favorite

Emory University political scientist Drew Linzer, who created and runs the Votamatic website, paid a visit to Middlebury College last Wednesday to discuss why, based on his forecast model, he believes President Obama is still the heavy favorite to win the presidential election.  To construct his model, Drew uses his colleague Alan Abramowitz’ “Time for A Change”  forecast model as his baseline starting point.  You will recall from one of my earlier posts that Abramowitz’s forecast model uses three factors — the incumbent president’s net approval rating at the end of June, the change in real GDP in the second quarter of the election year and a first-term incumbency advantage — to predict the winner of the national popular vote.  However, during the current election cycle Abramowitz updated his traditional model to include a “polarization” variable that, in effect, reduces the advantage enjoyed by a first-term incumbent running for reelection by about half – from a bit more than 5% to closer to 2.5%.  Under his “new” model, Abramowitz projects Obama’s share of the two-party vote to be about 50.3%.

As I’ve discussed before, not everyone accepts Abramowitz’s rationale for updating his model.  Drew is one of the skeptics, and so his forecast model starts with the “old” Abramowitz model which is decidedly more bullish regarding Obama’s chances.  Without the polarization variable, Abramowitz’s structural baseline component has Obama winning 52.2% of the two-party national vote.  That’s a much stronger starting advantage for the President than the “new” Abramowitz model suggests.

The second component in Drew’s model is the state-based polls, which he uses to “update” the Abramowitz baseline forecast. As we get closer to Election Day, state-level survey data influences his projection more and more, and Abramowitz’s structural component becomes correspondingly less important.  At this point, 10 days out, the state-level polling component is really driving his forecast almost entirely.

So, where does the race stand, according to Drew’s model?   As I discussed in my Economist post, as of today, he projects Obama to win 332 Electoral College votes, or 62% of the 538 Electoral College votes, compared to Romney’s 206.

Note that Drew makes several assumptions in his model.  First, he makes no effort to adjust for the “house effects” of individual polls in the belief that in the closest states that are polled most frequently, polling biases will largely cancel out.   Second, he essentially assumes that the “undecideds” will break in rough proportion to the distribution of the vote, as indicated by the polls, in each state.  Third, since he is interested in forecasting the Electoral College vote, he pays no attention to national tracking polls.

It is doubtful that Drew, or anyone, could have constructed such a forecast model even eight years ago.  But the proliferation of state polls, particularly in contested states, now allows political scientists to adjust their structural models in light of recent polling on a state-by-state basis.   Of course, this type of modeling is in its infancy; Drew only has one previous election cycle, in 2008, to calibrate his assumptions.

I asked Drew what would happen if he changed his baseline starting point by, for example, substituting Doug Hibbs’ Bread and Peace forecast model, which predicts that Obama will win closer to 47% of the two-party vote – or about 5% below the Abramowitz projection.  Drew acknowledged that this would shift the baseline parameter enough to move several swing states into Romney’s column.   Nonetheless, based on a state-by-state electoral projection and given the current polling, even with this shift Drew does not believe that Romney would gain enough Electoral College votes to overcome Obama’s current projected advantage.

Is Drew right?  Remember, for some time now I have been arguing that the state-level polls will gradually align with the national tracking polls, which as of today are showing a much closer race, with some indicating that Romney has pulled into a narrow lead.   How much must Romney gain in the key swing states to overcome Obama’s polling lead? Using the RealClearPolitics state polling averages, Owen Witek created the following table showing the state of the race in the 12 closest state races as of today (Electoral College votes are in parentheses). The last column contains an estimate of the number of undecided voters in each state.

State

Obama

Romney

Obama Margin

Undecided

Michigan (16) 48.8 44.8 4.0 6.4
Ohio (18) 48 45.7 2.3 6.3
Pennsylvania (20) 49.5 44.8 4.7 5.7
Virginia (13) 46.8 48 -1.2 5.2
New Hampshire (4) 48.3 47.2 1.1 4.5
Colorado (9) 47.8 47.8 0 4.4
Iowa (6) 49 46.7 2.3 4.3
Florida (29) 47.1 48.9 -1.8 4
Wisconsin (10) 49.3 47 2.3 3.7
North Carolina (15) 46.5 50.3 -3.8 3.2
Nevada (6) 49.7 47.2 2.5 3.1

Source: Real Clear Politics, 10/27/12

Based on the RCP averages (and remember that Drew’s state-based polling formula includes a structural component and thus is different than the RCP simple averaging), Obama right now has 201 Electoral College votes in strong and lean states, compared to Romney’s 191. That leaves 146 electoral votes in the 11 swing states listed in the table still up for grabs.  How likely is it that Romney can pick up the additional 79 needed to reach 270?  If he holds his “lead” in Virginia, North Carolina and Florida, he picks up 57 more, leaving him with 22 to go. Assuming a small shift in voter sentiment, he might also squeak to victory in Colorado, earning another 9 electoral votes, leaving him 13 to go.

But here is where the math becomes difficult for Romney, and why Drew – as of today – believes Obama will hold on.  Obama leads by more than 2% in all the remaining battleground states.  His smallest lead is 2.3% in three states: Ohio, Iowa and Wisconsin.  Looking only at these three, Romney needs to win either Ohio, or Iowa and Wisconsin, to reach the 270 mark.   But, as Witek shows, there are not very many undecideds left in either Iowa or Wisconsin, so they would have to break strongly in Romney’s favor for him to eke out a victory in both states.  That means Ohio may still be Romney’s best path to victory, and that assumes Obama loses the other states – no sure thing.  Remember, Romney got perhaps a 2.7% boost from the first widely-watched debate that was generally viewed as a convincing win for him.   How likely is it that he will be able to almost match that total in the last 10 days among the much small number of undecideds in the absence of a similar focusing event?

Keep in mind that all these calculations are based on polls that are, by nature, very squishy, so we ought not treat the RCP averages as having more precision than they do.  In this respect, Drew’s model, which uses a different algorithm to analyze the polls and predict the final outcome in each state, presents a slightly different picture in the battleground states.  Here are Drew’ state-based calculations, complete with the 95% confidence interval.  (Note the cool color coding!). As you can see by the vertical line in the middle demarcating the 50% threshold, he projects that Obama will do slightly better in key states than a simple reading of the RCP polling averages might suggest.  In contrast to my simple RCP average, he has Florida, Virginia and Colorado all leaning toward Obama (although with a confidence interval that suggests Romney might be leading in all three states).  That means Romney has that much of a bigger hill to climb among the undecideds.

Of course, if Romney does begin to close the gap in the swing states, Drew’s model will pick that up, and it will adjust the forecast accordingly.   Regardless of the outcome, however, there is one overriding reason why you should be visiting his site for the next 10 days: in contrast to other forecasters (you know who I mean), Drew’s methodology is completely transparent.  Anyone can see and utilize his modeling equation, which he describes at length in this journal article  (which has been peer-reviewed and accepted for publication). Moreover, as an academic, Drew’s rooting interest in this race is to see whether his forecasting tool is validated. He has much less at stake if the model turns out to be wrong. If it is wrong, however, he can tell us why.  I call that progress.

P.S If you are coming to this site for the first time (there’s been a lot of traffic of late) I encourage you to follow me on Twitter at @MattDickinson44 – I tweet all new posts there.

Ten Days And Counting: Where The Race Stands

My post today examining the State of the Race is up at the Economist’s Democracy in America website, but I will try to amplify my remarks in a post at this site later today.  In particular, I’ll try to reconcile the state-level polling which suggests Obama retains a slight Electoral College lead with the national polls that indicate Romney has been gaining ground.

Meanwhile, Bert Johnson and I have our last pre-election “Professor Pundits” taping up as well.

Also, if you are in the Middlebury area, I’ll be giving  my Election Forecast on Monday, Oct. 29th, at 7 p.m. (Details here.)

More at this site later today…..

Live Blogging The Third Presidential Debate

8:40 Welcome all to another installment of Live Blogging The Presidential Debate!   I hope you can join in- we set a record for participation and visits on this blog during the last debate, and I hope we can break it tonight.  And this gives me an opportunity to remind viewers that if they want to get on the distribution list for the regular posts here, just shoot me an email at dickinso@middlebury.edu  And tonight, I promise (Anna are you listening?) that we will be employing the new live blogging software so – in theory – you don’t need to refresh your screen to catch the latest comments.

Yes, Jim Fallows, Debates Do Matter (But Not As Much As You May Think)

In the aftermath of the first presidential debate, as swing state polls began to converge with the national tracking polls, former Carter speechwriter and Atlantic magazine correspondent James Fallows tweeted, “Maybe acads will stop saying debates never matter MT @ppppolls Obama down 6 overall, -6 MT, -5 NV, -5 WI, -4 MA, -2 VA vs pre-debate” .  He followed up on this theme in an interview on NPR in which he argued that debates do, in fact, matter.

Let me be clear: no political scientist that I know of believes that “debates don’t matter” (although I don’t doubt that a cursory or short-hand reading of some of their comments may lead one to believe this.)  Instead, what they argue is that the persuasive impact of debates is small, and therefore debates themselves are rarely consequential in terms of independently altering the outcome of an election.  But this is different from saying they don’t matter.  In fact, debates do matter.  I have argued that the first presidential debate served as a focusing event, allowing many voters to make a side-by-side comparison of the two candidates for the first time.   In so doing, a small number of viewers may have been persuaded that Mitt Romney is not nearly as extreme as he was being portrayed by the Obama campaign, and that given the economic fundamentals driving this campaign, they might take a chance on the challenger.  In short, the debate accelerated a process that was already underway, in which polls, particularly at the state level, moved closer into alignment with what the fundamentals driving the vote dictated.

But notice that I said a small number of viewers.  For all the talk about what a disaster the first debate was for Obama, the impact on his polling support was not very large. Middlebury College student Anna Esten, building on research by political scientist Tom Holbrook regarding the impact of previous debates, calculates that Obama lost 2.7% support nationally after the first debate. (She based this on a comparison of the average of national polls in the field in the week before the debate with the average of polls in the week after.) As you can see in the chart below, which incorporates Holbrook’s data, Obama’s 2.7% drop is relatively large impact for a first debate – about .5% larger than the second biggest – but not overwhelmingly so.

Incumbent Party Percent of Two-Party Vote During Debate Periods, 1988 to 2008
Debate Pre-Debate Post-Debate Bump Total Change During Debate Period

1988

First

52.9

53.15

0.25

Second

53.72

55.32

1.6

2.42

1992

First

41.57

41.74

0.17

Second

42.76

40.73

-2.03

Third

41.73

42.13

0.4

0.56

1996

First

60.27

58.8

-1.47

Second

58.64

58.83

0.19

-1.44

2000

First

51.07

50.13

-0.94

Second

49.66

48.47

-1.19

Third

48.92

47.55

-1.37

-3.52

2004

First

52.8

50.54

-2.26

Second

50.05

50.97

0.92

Third

50.59

50.85

0.26

-1.95

2008

First

48.11

46.76

-1.35

Second

46.66

45.73

-0.93

Third

45.86

46.18

0.32

-1.93

2012

First

49.29

46.57

-2.72

?

Moreover there is some evidence that the second debate, which some Obama supporters claim is his best debate performance ever!, may have at least arrested Obama’s polling decline, and may even have given him a small polling boost. This is a reminder that there is one more debate to go, this one tomorrow night dealing with foreign policy.   So we shouldn’t jump to the conclusion regarding the impact of the debates this year quite yet.  Fallows may be tweeting a different tune come next week.  (“My bad – acads right again.”  Ok, maybe not.)

For a variety of reasons, however, I would not expect tomorrow’s debate to have nearly the impact of the first presidential debate, or even the second one, for that matter.  First, and I don’t want to make too much of this given the small “n”, the average impact of the four third debates dating back to 1992 on the incumbent party’s polls is a miniscule -.10.  That compares to -1.19 for the first debate, and -.24 for the second.   Second, tomorrow’s debate centers on foreign policy. This is a topic much in the news of late due to the controversy over the killing of a U.S. diplomat and his security team in Benghazi, and more recently the reports  – denied by the White House – that Iran has agreed to bilateral talks with the U.S. regarding its nuclear program.   Nonetheless, most polls indicate that voters’ concerns over the economy far outweigh their interest in foreign policy questions.   Finally, we have had two relatively widely viewed debates so far, with the audience for the second, at an estimated 65.6 million viewers, only slightly smaller than the 67 million who tuned into the first debate.  (This does not count the millions more who viewed the debates through other means, such as computers and tablets.)  I expect that the audience for tomorrow’s debate will be smaller than that for the first debate – and perhaps even less than the second, since it will be occurring opposite Monday night football.  And there likely will be fewer undecided voters within the viewing audience.   Finally, with the election polls now more in line with where I have been predicting they would be since Labor Day, I think there is likely to be a bit less volatility in the polling in the remaining 17 days before the Election.

Again, this is not to say that Monday’s debate “won’t matter.”   It should help at least some of the remaining undecideds begin considering their vote choice in light  of the fundamentals, and in this regard should continue the process of bringing the polling closer in line with the actual vote come November 6.   But I would be very surprised if it led to a bump, or slide, in Obama’s polling anywhere near the 2.7% impact of the first debate.

That doesn’t mean it won’t  be entertaining, however!  As always, I’ll be on, live blogging, and yes, the live blogging software is now installed and operating properly. (Really.  I mean it.)  So join in. It’s scheduled to start at 9, and I’ll be on slightly before.

And, remember, you can’t ever be sure who is watching the debate, and what they will learn (hat tip to Kate Hamilton)!