Numbers, arranged haphazardly (blog.contegix.com).
Tony Petrangelo
by Tony Petrangelo
May 22, 2012, 7:00 AM

Which numbers are the right numbers?

» Post updated below

The question in the title of this post comes from a Twitter discussion I had with @NorthernMNer yesterday. He asked:

A little background, Kory Kath announced his retirement from the Minnesota House yesterday, prompting Twitter chatter about the competitiveness or lack thereof of his seat.

So who’s right? Well, they both are of course, they are just using different numbers. Which is the crux of NorthernMNer’s question, “which numbers are the right numbers?”

The answer to that is, it depends. A great example of the differences in these numbers can be seen in Kory Kath’s old House district, 26A.

In 2008 in that district, Barack Obama lost to John McCain by six points 46%-52%. At the same time Kory Kath beat his Republican opponent, Tom Kuntz, 57%-43%. Rep. Kath over-performed President Obama by 11 points.

In 2010 in that district Mark Dayton lost to Tom Emmer 37%-47%. Kory Kath beat his GOP opponent, David Thul, 62%-38%. A 25 point over-performance.

In 2006 though, a good year for Democrats nationally and in the state, Republican Connie Ruth beat DFLer Kathy Muellerleile 56%-43%. That same year incumbent GOP Governor Tim Pawlenty won the district 56%-37%.

What you can see when looking at those numbers is that the only time Democrats do well in the district is when that Democrat is named Kory Kath.

This means that its hard to know if a district is voting the way it is because of its partisan lean, or because of candidate quality. Or even both. And because the quality of candidates will be different in every legislative race in the state, you can’t compare those districts to one another based on past legislator performance, because you don’t know if the votes were for the candidate or the party.

To be clear, you still have these issues with candidate quality in statewide and nationwide races, but the difference is that every person in the state is given the chance to vote on those candidate options.

That means that to the degree candidate quality is skewing the numbers, that skew should effect all districts about equally. This is why Cook PVI uses Presidential numbers to measure the partisan nature of congressional seats instead of the results of the actual congressional elections.

In a partisan metric such as PVI, you want to control for candidate quality as a variable and using numbers from the legislative races themselves doesn’t do this, instead, it amplifies the problem.

To answer NorthernMNer’s question then, you want to use numbers from races that everyone in the state had a chance to vote on.

—

» Update

A few people have taken issue with this sentence

What you can see when looking at those numbers is that the only time Democrats do well in the district is when that Democrat is named Kory Kath.

Both Tim Walz and Amy Klobuchar are DFLers not named and Kory Kath and both of them won this district. Additionally, I suspect that people exist who are not named Kory Kath who could also win the district as a DFLer.

But I didn’t say that Democrats not named Kory Kath don’t win, just that they don’t do well. Let’s take Amy Klobuchar as an example. In 2006 she won district 26A 53%-42%, an 11 point margin, while winning statewide 58%-38%, a 20 point victory.

If we were to calculate a PVI out of just these numbers, in the standard way by subtracting Amy’s top line number in the district by her statewide top line, we would get an R+5 district.

The advantage here is that the statewide numbers act as the baseline from which the individual legislative districts are measured. This controls for the candidates themselves (barring any regional quirks which do exist), allowing comparisons between districts to be apples to apples as opposed to apples to Kath’s.

You can actually do the same thing with Tim Walz if you want, although congressional races are not usually used for this purpose. Tim Walz won 26A 49%-44% and won CD1, 49%-44%. Meaning the district would rate out as even.

The only numbers you can look at that paint the district as a DFL leaning one are the ones from Kory Kath elections, and those are non-contextualized numbers.

The fact the Amy Klobuchar and Tim Walz won in district 26A in 2006, despite the fact that their names were not Kory Kath, is irrelevant to the point of the post itself. And using the numbers from Klobuchars 2006 race helps to illustrate exactly why using statewide and national numbers for statewide legislative district metrics is the superior method.

Thanks for your feedback. If we like what you have to say, it may appear in a future post of reader reactions.