Wednesday, March 16, 2011


Republicans to maintain control of the House in 2012

Unless a historic event occurs, Republicans will still be in control of the House of Representatives after the 2012 election. How can I be so confident even when House re-districting is still occurring?

It turns out that the difference between House election results during Presidential election years are very well accounted for by fundamentals variables*. What I mean by "fundamentals" is simply variables that we know (or can reasonably predict) before the election and do not include polling data. In the 15 Presidential year House elections since 1952 (a common post-war cutoff for political science), these variables** include

-Percentage of seats won by the majority party in the last election. The majority party in the House wins more seats when it previously held more seats. In 2012, this variable is 55.6 because the Republicans won 242 seats out of 435 in the 2010 House elections.

-A dummy (1 or 0) for whether the majority party in the House is the same as in control of the Presidency. In 2012, this variable is 0 because the House is controlled by a different party than the White House.

-The percentage of the vote the party in control of the White House wins in the Presidential election during years in which the majority party in the House is the same as in control of the Presidency. Not surprisingly, when the party controlling both is the same, the majority party in the House gains more seats when its candidate for President wins a higher percentage of the vote. Surprisingly, Presidential vote has little relation to the House election in years when the House and Presidency are controlled by different parties. In 2012, this variable is 0 because the House is controlled by a different party than the Presidency.

-A dummy (1 or 0) that is 1 when the party in control of the White House is different from the majority party in the House, and the President has started or maintained for more than one term an "unprovoked, hostile deployment of American armed forces in foreign conflict" (as defined by Douglas Hibbs) which has resulted in at least 1 fatality during the past term. Interestingly, as if to penalize the party in the White House, the majority party in the House wins more seats when this variable is true (like in 2008). In 2012, this variable is 0 (false) because the Iraq War was started by a Republican President (Bush), and Democratic President (Obama) has not continued it for more than one term.

In simple linear regression equation form, the model reads for 2012,

Percentage of seats won by the current Majority Party (54.8%) = Coefficient for Previous Seat Share (.67) * Previous Seat Share of the Previous Vote share (55.6) + Constant (17.3).

The model is therefore predicting that the Republicans will win 238 seats, more than enough for them to maintain their majority.

We must ask how accurate is this model in explaining past results?

The answer is very accurate. The model is able to account for 95.9% of the difference in the results of the 15 Presidential year House elections since 1952.

What about the chance of error? The root-mean-squared-error (a statistic that measures errors in estimate and penalizes for larger errors) is 1.1% of seat share, which given our small sample size (15) corresponds to a margin of error at 95% confidence of +/- 2.3% seats in Congress. The largest error (and the only with an error of greater than 1%) in our set is the 1988 election, which the model misfit by 2.5% (or 11 seats). For 2012, our margin of error indicates that Republicans winning as many as 248 seats and as few as 228 is a reasonable expectation.

Of course, none of these findings are useful for 2012, unless we know how well the model not only explains but also would have predicted past House elections. To do so, we take out a given election from the model and re-run the regression. The results give us confidence that the 2012 estimate should be a good one. In 2008, the model called for the Democrats to win 252 seats, when they ended up winning 257 seats. In 2004 (thus we take out 2004 and 2008 from our dataset), the model would have forecasted Republicans to win 236 seats, when they won 232.

And what about the last time a Presidential election took place directly after re-districting (1992)? Eliminating 1992 from the dataset and re-estimating the model, we find that the model would have projected the majority party (Democrats) to win 254 seats, when they took 258. Thus, while this year's re-districting may cause the model problems, past history does not indicate that it will.

Given this past accuracy and current Republican majority, it looks like it will still be Speaker Boehner come January 2013.


**As always, email if you are interested in the dataset or have questions.

*To those statistic masters out there, yes it would be proper to include both "original" independent variables that make-up the interaction terms (war and Presidential vote). They were left out to preserve a reasonably low number of variables given the low number of observations as well as for simplicity of explanation. There is no statistically significant difference in the accuracy of the "full" model to explain past results.

Nice to see you back, Harry. Could you post the full dataset on Google Docs or the like?

Thanks for the welcome back. Forecasting winter weather (my other hobby) is pretty much done. Here's the link to the data.

After Walkers assault on the middle class and loyalty to non=taxpaying corporations it has split the republicans. Many republicans are one issue voters...Pro-life. But with the corporations using these politicians as their puppets...who know if they are even really pro-life? It is basically the corporations vs. Planned parenthood.
We need a pro-life democrat. Don't assume the republicans will win. Walker changed that.

Out of curiosity, I have a question/suggestion -- when you discuss how well the model explained the results of the 15 Presidential-year House Elections since 1952, do you just plot the (predicted) fitted_y = fitted_beta*x against the actual_y, keeping the fitted_beta constant?

Have you considered the expanding-window regression, such that you re-estimate beta first for the data-set at {1952}, then at {1952, 1956}, then at {1952, 1956, 1960), ..., and finally full data-set {1952, 1956, ..., 2004, 2008}? This would simulate how the information set (the one you condition on in the OLS) grows over time and how it affects the estimates.

It's one of the simple heuristics to asses forecasting in time-series models (of course, there are usually several orders of magnitude more observations in these, which might be a problem here).

Alternatively, you might also consider in-sample vs. out-of-sample forecasting -- for instance, simply run the regression for the dataset at t={1952, ..., 1980} and see if the fitted_y compares well against actual_y for full dataset.

The point of this is that the relationships might change over time -- e.g., there might be regime changes in beta (perhaps the impact of some explanatory variables has grown/diminished over time), or it may be even continuously time-varying. By doing those kinds of analysis we can gain more confidence in the stability and robustness of the results.

Great question. I have done the latter (i.e. out of model forecast) for all years. I should probably do a follow up post, but what I found was the following. The average seat error was 4.25 seats, and the greatest error as in the residual plot was 1988 at 12.5 seats.

Email me at the address on the side, if you have more questions. I'm usually quick as a lightening strike on there.

The model suggests that in the broad sense of things that Wisconsin doesn't matter. In fact, many political scientists believe the economy may have little impact on Congressional elections (except for how it affects the Presidential election and coattails when the majority party in the House matches the party in the White House).

Now, the model has the chance of error. So it's very possible that we end up towards the lower end of the Republican estimate or even a little lower.

If nothing else, this model gives us a good launching off point and gives an idea of where things could be heading in '12.
Over the last 100 years, you have a sample of exactly 3. Only three times since the 19th century has an incumbent president been on the ballot at the same time as a round of reapportionment/redistricting. During that time, only two incumbents who'd been elected before--Wilson & George W Bush--had close reelection bids. The general pattern is either a blowout win or a blowout loss. The three incumbent reelection bids on redistricting years were all blowouts-- Hoover in 1932, Nixon in 1972 & GHW Bush in 1992--but the net change in the House wasn't consistent; the incumbent's part won the WH and netted 13 seats in the House in 72, lost the WH but gained a net of 9 seats in the House in 92, and lost the WH and a historic high loss of a net 97 seats (!) in the House in 1932. What to make of that? Not much that you can generalize from. The two times Dems lost seats, their presidential candidate--despite in one instance easily winning--failed to top 43%. In the case where they gained 97 seats, their presidential candidate got 57%.

In short, there are a ton of counfounding historically relevant data points that make it hard to say there's any confidence in this kind of model, especially when you don't even know the boundaries or whether the Repubs will nominate someone competitive of someone from the far right who will be a general election disaster.

Thanks for the comment. The model has two examples of redistricting with the incumbent president running for re-election (and a third without the incumbent running in 1952).

One comment I would make is most work I've done as well as that by Abramowitz & Hibbs among others shows it's not the term of the president that matters. It's the amount of terms the party has been in the White House (so that would be 3 for the Republicans in 88).

I want to clarify something. The model isn't predicting seat changes of the party in the White House (yes sometimes it does that). The model is predicting seat change for the majority party in the House. Typically, the majority party will lose a few seats if it doesn't also hold the presidency (the exception being war years). Thus, it didn't matter if it was a blowout loss or win in 72 or 92. The model doesn't believe the prez vote made a difference.

As for the rest, the model is only based on elections since 1952 (a common cutoff for fundamental models for prez year elections). The model won't work for years prior to 52 as many fundamental models won't (I guess I could throw in a bunch of ad-hoc adjustments like Ray Fair... I'd prefer not to). But for the years, I include re-districting didn't prove to be a problem. I wouldn't simply say there are a ton of "confounding" points.

Part of me really wishes the model is wrong, so I can have more fun over the next 2 years.

Write back (though if you email me, I'm much, much faster).
When Harold Macmillan the Prime Minister of Great Brittan, was ask by a news reporter if there was anything that could cause his government’s plans to fail, the Prime Minister using his father’s upper crust English and his mother’s Hoosier common sense replied “Events my dear boy, events.”

Personally I plan to wait a while before I start predicting the outcome of the next election.
Post a Comment

Subscribe to Post Comments [Atom]

<< Home

This page is powered by Blogger. Isn't yours?

Subscribe to Posts [Atom]