Tuesday, October 19, 2010

The Hutchinson Effect

This is a footnote of sorts to The Horner Effect series, part three of which will be done soon.

A few people have questioned why I left the 2006 Governor's race out of The Horner Effect series. I suppose I should have more explicitly addressed why it was left out, so I'll do that now and while I'm at it I'll go ahead and dispell some zombie myths about the 2006 election, many of which are conveniently contained in a diary on Swing State Project:

If recent Minnesota gubernatorial elections are any indication, the great equalizer will be the late October televised debates. Polls moved in double digits in both directions in a matter of a week based on Minnesota gubernatorial debates. If you impress there, you're golden. If you fail to impress, you're ruined. And that's true almost wherever your poll numbers may currently be. The Independence Party's articulate 2006 candidate Peter Hutchinson was not in a position to win, but he nonetheless impressed in the debates and managed to surge at Mike Hatch's expense, handing victory to Pawlenty.

This is simply not true. There is no evidence that Peter Hutchinson cost Mike Hatch the election, despite popular belief.

The Hutchinson non-factor

The easiest explaination for why I left the 2006 Governor's race out of the Horner series is that I'm trying to determine the impact Tom Horner will have on the election with the assumption that he will finish with somewhere around 15% of the vote on election day, if not more. In doing this I'm looking at past elections where the IP candidate played a significant role and I didn't include the 2006 governor's race because Peter Hutchinson didn't play a significant role.

The races I focused on in the Horner series were the ones in which the IP candidate got a decent chunk of the vote. Those candidates that got the aforementioned decent chunk are in bold in the following chart, from part one of the series:

Image Hosted by ImageShack.us

The DFL - GOP vote share correlations in the non-bolded races are very strong, meaning the race was decided by the vote share distribution between these two candidates and that distribution was not influenced very much by the other candidates in the race.

To show this graphically let's compare a few scatterplots, these are from the 2006 and 2008 Senate races respectively, in 2006 the IP candidate got less than 3%, in 2008 the IP candidate got over 15%.

Image Hosted by ImageShack.usImage Hosted by ImageShack.us

Here is the 2006 Governor's race where the IP candidate got just over 6%.

Image Hosted by ImageShack.us

While it is not quite as tight as the 2006 Senate scatterplot, it resembles that distribution much more so than the 2008 Senate race. But this doesn't show how Hutchinson's vote was distributed. Again, well look at the 2006 and 2008 Senate races first, then the 2006 Governor's race.

Image Hosted by ImageShack.usImage Hosted by ImageShack.us

Image Hosted by ImageShack.usImage Hosted by ImageShack.us

Image Hosted by ImageShack.usImage Hosted by ImageShack.us

Again, which race does it more resemble. Hutchinson's vote came in a small band throughout the state with little direct effect on either candidate. It's important to point out again that correlation is not causation, we have no idea where those Hutchinson voters would have gone had he not been in the race, but it's highly unlikely that they would have fled, en masse, to one candidate to another.

Even in a race as close as the 2006 Governor's race was, Pawlenty won by just over 21,000 votes, Hatch would have had to collect at least 67.5% of Hutchinson's votes to have overtaken Pawlenty, quite the tall task in a race decided by 1 point. And that's under a best case scenario too, assuming none of Hutchinson's voters decide sit out the election if he's not on the ballot.

About that Hutchinson surge

Looking at all the polls done in the last month of the campaign that included Hutchinson, he was at 7,7,6,9,4,7,8 for an average of 6.85%. He finished at 6.43%, so pretty much exactly what he'd been polling at the whole time.

On the other hand, averaging out all the polls in the last month for the other two candidates gives you 45.5 for Hatch and 42.5 for Pawlenty. If anyone surged at the end it was Pawlenty, not Hutchinson and this was most likely because of Hatch's public blowup which drove undecided voters to Pawlenty.

The 2006 Governor's race was, for all intents and purposes, a two person race between Mike Hatch and Tim Pawlenty, Peter Hutchinson just wasn't a factor no matter what kinds of elaborate reasoning has been constructed to try and make him into one after the fact.

But Peter Hutchinson got 10% in Ramsey county

Peter Hutchinson scored 10% of the vote in the DFL stronghold of Ramsey County, doing much better than that in some of the city's most liberal precincts. If I remember correctly, Hutchinson's haul in Ramsey County alone was the difference between Hatch and Pawlenty.

While this is not untrue, it's also not relevant. Hutchinson got 19,527 votes in Ramsey county and the difference between Hatch and Pawlenty was 21,108, so it was close, but not quite enough. But what if it was, let's say Hutchinson got 22,000 votes in Ramsey county, what then? Certainly 100% of those votes wouldn't have gone to Hatch had Hutchinson not been in the race, in fact it's not even clear that Hutchinson's good Ramsey showing came at Hatch's expense, which is what the implication is of stating that Ramsey is a DFL stronghold.

Let's look at the difference between the percentage of the vote a candidate got in Ramsey versus Statewide.

Image Hosted by ImageShack.us

In 2006 Mike Hatch's vote share difference between Ramsey county and statewide was almost the same as it was for the other DFLers in the chart in other years. The big difference you can see is that Tim Pawlenty did worse in Ramsey in 2006 than past GOP candidates have and Peter Hutchinson did better. This would seem to be rather conclusive evidence that Peter Hutchinson's Ramsey county votes did not have an adverse effect in Mike Hatch, even though he got almost 10%. If anything it appears that Hutchinson actually helped Mike Hatch in Ramsey county by soaking up moderate Republicans.

Occam's Razor

Hutchinson's generally impressive showing in the debates made it possible for him to score those kinds of votes out of left-leaning areas and ultimately threw the election to Pawlenty. I don't think liberal voters in St. Paul would have been any more likely to vote Hutchinson merely out of protest for Hatch than other regions of the state if not for the fact that they liked what they saw in Hutchinson.

While the above theory is certainly plausible, as we have seen there's no data that backs it up and more to the point we have a glaringly obvious reason for why Tim Pawlenty surged at the end and Mike Hatch lost, there is no need to construct elaborate "Peter Hutchinson pwned the debates" theories.

When faced with multiple possible explanations for why a thing happened Occam's razor suggests you go with the simple one and the simple one in this case is that one week before the election Mike Hatch lost his mind and called a reporter a "Republican whore" and that was all that got talked about until election day. Peter Hutchinson was a non-factor.

Putting 2006 to bed

This is the problem with theorizing about the results of an election without knowing the facts. I have had this very same debate, albeit in much less detail, with many well intentioned DFLers who are convinced that if it wasn't for the damned IP candidates, whether its CD3 and CD6 in 2008 or the 2006 Governor's race, the DFLer would have won. When it comes to 2006 at least, that's simply not true.

The 2006 Governor's election was about Tim Pawlenty and Mike Hatch and there is no evidence that Peter Hutchinson had any impact one way or the other. That is why the 2006 governor's race wasn't included in The Horner Effect series, in 2006 there was no Hutchinson Effect.

No comments:

Post a Comment