Tuesday, August 26, 2008

An August Look at the Barr/Nader Effect in 2008

When Zogby released their interactive poll results for the ten states they label battlegrounds over the weekend I considered once again looking at where Bob Barr was doing well. After the 34 state polling release in early July, we saw that Barr was faring quite well -- perhaps too well -- in some states that would make John McCain's job that much more difficult. Instead of doing that again, though, I thought I'd see what kind of data I could gather that would help us gain a better idea of how both Barr and Nader would potentially affect the upcoming general election.

With that in mind, I brought together a few different elements:
1) State polls charting a three or four way race.

2) National polls with either a three or four candidates included.

3) The share of the 2004 general election vote that both Nader and the Libertarian Party nominee, Michael Badnarik, received.

While there is a limited amount of data for the current cycle at both the national and state levels, the picture of a multiple candidate race can be augmented by the 2004 data. That provides a better sense of who does well and where. Since late May when Barr was nominated to represent the Libertarian Party, there have been 30 polls in 15 different states that include Barr and/or Nader in them. And during that same period there have been 18 national polls with either three or four candidates included. Again, this is a limited amount of information, but if we combine that data in a regression with the 2004 election results for Nader and the Libertarian Party, we get much closer to being able to predict if not how well both will do in November, then at least an idea of which states fall where in the pecking order.

Let's look at each separately.

[Click to Enlarge]

After the regression, we can plot the predicted vote share for, in this instance, Bob Barr against the vote share the Libertarian Party received in the 2004 presidential election. For the sake of clarity, I've only included the points for the 14 toss up states in our most recent electoral college projection, but rest assured the model includes all fifty states. On the lower left are toss up states where Barr does not take up too much of the the vote share on the right side of the ideological spectrum. That's good news for McCain in states like New Hampshire and Florida because the conventional wisdom holds that a Libertarian nominee would pull more from the Republican than Democratic nominee.

On the other side of the graphic, though, there are a couple of states where there may be cause for concern for the McCain camp. Both Alaska and Indiana give over three and a quarter points to Barr. That may not sound like much, but when two regularly solid Republican states require some amount of defense, the Republican nominee would be better served if he didn't have to fend off attacks on two fronts. And tucked away there in the middle of the pack are our three closest states, Nevada, Ohio and Virginia. As tight as each of these are no amount of support for Barr would be welcomed by the Arizona senator, and two or more points may be enough to swing any of the three toward Obama and the Democrats.

The situation is a bit different in the Nader context.

[Click to Enlarge]

For starters, the same group who supported Nader in 2004 seem to be behind him again in 2008. Not the same exact people, but at least at this state a similar number of people. That can be contrasted with the Libertarian example above. The party of smaller and less interventionist government appears to have a much firmer footing in 2008 than it did just four years ago. Some or all of that may have to do with disaffected Ron Paul supporters who will hold their own gathering simultaneous to the GOP convention next week. It is a group that is certainly more energized during this cycle. But back to Nader. There is such a small range of vote share values across these toss up states. Nader is close to getting nearly two percent from each of these fourteen states (and for that matter all states since this is the same range in which the lean and strong states would fall as well.). And then there is the question that has been asked since 2000: Who are these Nader voters? Are they people who are only voting for Nader and thus not taking votes away from Barack Obama? Or does Nader represent a refuge for Democrats who won't pull the lever for the Illinois senator anyway? The former seems more plausible than the latter. Nader's share of the vote shrunk from 2000 to 2004 as Democrats, burned by the 2000 experience took a more pragmatic approach into the voting booth with them in 2004. Nader didn't really hurt Kerry; he wasn't even on the ballot in Ohio. And 2008, at least at this early juncture appears to be shaping up similarly to 4 years ago rather than 8 years ago.

I shouldn't short the Nader graphic, though. It is interesting that Indiana is among the strongest states predicted for Barr and is on the opposite end in the Nader example. If Nader were to pull votes away from Obama, the Hoosier state is a place where the Illinois senator would get the best of both worlds: a minimal Nader effect, but a comparatively large Barr effect.

Now, both accounts above come with some caveats. First, polls this time of year, both national and state, tend to overstate the position of third party candidates in the race. As we get closer to November, we'll start to see some movement toward one or the other of the two major party candidates. Diminished or not though, we do get from this a sense of which states are most likely to be affected by these third party candidacies. We can begin, for example, to look on this as a companion to the electoral college spectrum.

Another issue is that this model is far from inclusive. We are dealing with a limited number of variables here, so we aren't dealing with the full world of factors. [Misspecification alert!] However, this does get us moving in the proper direction at least; especially in that it bring more information to the table than simply the minimal amount of three and four way polling that is available.


Recent Posts:
The Pre-Convention Swoon Revisited

The Electoral Map (8/24/08)

Swoon? What Swoon? A Look at the Changes During Pre-Convention August

5 comments:

Allen said...

More useful might be looking at the 3rd party 2000 & 2004 polls vs. the actuals, and using that to predict what might happen this year. I would BTW expect 2000 and 2004 to be different, and for this year to be more like 2004.

Jack said...

Badnarik probably did well in Indiana because he was originally from the state.

SarahLawrenceScott said...

Most of the polls that were done both with and without Barr/Nader show that Nader is actually taking from McCain. That suggests that Nader is a resting spot for Democrats and Democratic-leaning independents dissatisfied with Obama.

Allen said...

I'm not convinced that many voters in swing states (where it matters) are actually going to vote for 3rd party candidates, no matter what they say at this point when asked by pollsters. While I haven't compared 2000 and 2004 3rd party polls vs. actuals, there was a huge drop from 2000 to 2004 in the number of actual votes for Nader in FL. I think voters learned after the 2000 election that a voting for the "lesser of two evils" might be a better option than voting for a 3rd party candidate.

Josh Putnam said...

Excellent points everyone.

Allen,
I should be able to add in the numbers from 2000 pretty easily. Also, in looking at the data yesterday, one thing was clear: being a swing state more often than not means a smaller share of the vote for third parties. Whether Indiana actually is or ends up as a swing state, the fact that it is in the discussion is new for the state. However, I would expect it to come in line with the third party behavior of other swing states if the Hoosier state is close.

For clarity's sake, yes Jack, Badnarik's being a native of Indiana seems to have helped. It gave him the largest vote share of any other state.

Scott,
That is a point I meant to bring up. If anything, the evidence is that Nader is actually hurting McCain. I'll try to find a link for that.