Numbers, arranged haphazardly (
by Tony Petrangelo
Jun 14, 2012, 7:00 AM

And now for hPVI

A few weeks ago I discussed what PVI, Partisan Voting Index, is and how it’s calculated. Today I’m going to discuss my own version of PVI that I use for Minnesota legislative races and that I very creatively call hPVI (hybrid Partisan Voting Index).

But before we get into that though, one question that should be answered is, why not just use PVI for state legislative races?

My main issue with PVI, is that at this point in time, right before a Presidential election, it is still using data from 2004 for it’s calculations. That’s more than seven years ago at this point and we were in a whole different political climate then.

When you’re measuring congressional districts though you don’t have a choice, Presidential elections are the only elections that all of the different congressional districts have in common.

When we’re talking about measuring state legislative districts instead, we have more options at our disposal, as any statewide race will be common among all of the legislative districts. You can use Governor’s race results, US Senate race results, even the statewide constitutional officers races would work.

My personal preference is to use the most recent Presidential race and the most recent Gubernatorial race. There are a few reasons for that. For one, the highest level races will be the races that have the least amount of local calibration, so to speak.

Candidates for President and even Governor will not be tailoring what they say to match the concerns of Minnesota Senate district X, whereas candidates running for that district will have (or should have) a message almost completely tailored for the district. So the higher level the race, the more likely a vote reflects a voters actual partisan tendencies.

The reason I like to use Gubernatorial numbers as opposed to say US Senate numbers is because in Minnesota there is a Gubernatorial election two years apart from a Presidential election, meaning you get a nice set of staggered data. If one was to use Senate numbers the most recent election you could use is from 2008, but we already have Presidential numbers from that year.

There are however, arguments to be made for using different numbers, constitutional offices specifically. They are on the same cycle as Gubernatorial elections but have the advantage of not attracting high profile Independence party candidates who may or may not skew the results.

Additionally, since they are lower information races the votes people cast are likely to be based more on peoples personal political leanings than any feelings about the candidates specifically.

I have a slightly different take on this argument, which is that the presence of a viable Independence party candidate serves to give us a better understanding of the actual partisan dynamics in the state, then a heads up race does.

This is because partisans will not likely vote for the Independence Party candidate and those who do vote for the Independence party candidate are likely not partisans. In this regard the Independence Party candidate pushes soft support away from the parties, leaving mostly the partisans.

That’s the theory at least.

There’s no correct answer in this regard, just preferences, and I’ve explained what my preference is and why.

So with all of that out of the way, here’s how hPVI is calculated:

PA = Democratic share of the two-way Presidential vote, 2008
PB = Democratic share of the two-way Gubernatorial vote, 2010

pA = Democratic share of the two-way Presidential vote in the district, 2008
pB = Democratic share of the two-way  Gubernatorial  vote in the district, 2010

    \[ hPVI = (\frac{pA+pB}{2}-\frac{PA+PB}{2})*100 \]

What you can see is that the formula for hPVI is the same as the formula for PVI, it’s just that different data is being used in the calculation.

Next week I will post the actual hPVI numbers of the new legislative districts.

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