Is Pennsylvania getting bluer?

In a wide-ranging discussion at Althouse, commenter Eli Blake suggests that "it's too late [to save Rick Santorum's seat]. He's too conservative for a blue leaning (and more importantly becoming bluer) state like Pennsylvania." I think Eli is probably right that Santorum is in pretty dire straights, but I'm not sure I agree with the theory that Pennsylvania is a blue state leaning bluer.

Ely relies on Scott Elliot's analysis, which can be found here. It is this: "[b]ased on voting patterns since 1992, Pennsylvania is trending DEMOCRAT." An ipse dixit is no substitute for actual numbers, though, and it is numbers that this post is about.

I f Eli and Scott are right, there ought to be some sort of quantitative data to back up that trend / drift / leaning. I suppose that, if you want to be very narrow about it, you can say that Kerry got a higher percentage of the vote in 2004 than did Gore in 2000, and this constitutes a drift, but I think that is a specious conclusion. If you narrow your data range sufficiently, you can label almost anything a trend (see D. Huff & I. Geis, How to Lie With Statistics). I would suggest that a more longitudinal survey of election results in Pennsylvania demonstrates that, put in context, the drift theory fairly tenuous.

We will look at some numbers in the twenty year period for which we have numbers, 1974 through to 2004. Because we at Stubborn Facts believe in citing sources, I should begin by saying that the Presidential numbers are from Dave Leip's excellent U.S. Election Atlas, the numbers for the Pennsylvania gubernatorial and senatorial elections are from data provided by Wilkes University Prof. Harold Cox, and the numbers for Pennsylvania's U.S. House seats are from Wikipedia. In addition, we are using the now-traditional red for Republican, blue for Democrat colorscheme.

So, without further ado, to the numbers. Cue graphs.

As you can see, this is not the stuff that serious, compelling evidence of drift is made of. In the graphs for the three statewide contests, we see a competetive state that is moving back and forth; I include the distribution of house seats largely for comparative value, since I have little doubt that gerrymandering and safe districts makes it a poor freestanding measure, but it is notable that it shows much the same fluctuations as in the statewide races.

Is Pennsylvania "blue getting bluer"? Not on this evidence. But this evidence excludes the most relevant piece of information, which would be to consider the number of seats won in the Pennsylvania state legislature in the same time period. I can't find those statistics, but if any readers want to help out, go ahead and email me and I'll run up some graphs for that too. So I'm not ruling out the possibility, and as always, past performance is no guarantee of future returns - but I think that Eli is letting his optimism (optimism from his perspective, I hasten to add) get the best of him.