Sunday, April 15, 2007

An introduction to political sabermetrics

Before I started blogging, a good chunk of my time online was spent at what was then Baseball Primer (now the blogs section of Baseball Think Factory). While I'm a fan of a lot of sports, baseball's sabermetric revolution made for a particularly interesting study: first a small group of fans led by Bill James and SABR, then the conventional wisdom within the sport, recognized that many of the existing means of evaluating the performance of both players and franchises could be improved by leaps and bounds through a better appreciation of how to interpret the game's statistics. (And it didn't hurt that the level of political discussion on that site was on par with any that I'd yet discovered - with at least one regular going on to be a contributor at a current political favourite.)

So what does this have to do with current Canadian politics? Over the past week, two posts have used calculations of each party's expenses per vote as a measure of how much growth potential they have in that riding - highlighting the potential to analyze Canadian politics based on a return-on-investment approach similar to the dollars-per-win type of analysis that's now in vogue in sports.

Let's note that even on a surface look, the claims currently being made don't reflect the numbers being used to support them. While moving from 4th to 2nd in total votes in the London North Centre by-election, the Greens simultaneously went from being the most efficient spenders per vote to the least efficient - meaning that Jim Harris' "we spend money better!" claim isn't supported by the evidence. Nor does there appear to be any other consistent party ranking based on this small sample: the Libs were the least efficient spenders in Central Nova, but the most efficient in the London North Centre by-election (and well ahead of the Cons in London North Centre in the general election as well).

In sum, the data now being cited doesn't support any generalizations as to who's likely to spend their money most effectively, and how other parties can improve their return. But while the current simplistic claims don't stand up, there seems to be plenty here worth analyzing.

What return do political parties normally receive on their riding investments? Is there a diminishing return on every dollar put into a riding? How do past investments - or any change in those investments - affect a party's vote outcomes in a riding? When is party money better spent on national advertising than on local organizing? And how does the vote-per-dollar calculation and resulting federal funding fit in with the core purpose of a party to win seats?

As is the case in baseball, there are still individual factors at play which can't be fully captured by the numbers. In what might make for an analogy to the importance of stadium adjustments in baseball, the effect of any riding expense presumably needs to be weighed against a party's national standing. And individual candidate personalities could make a huge difference on a single-riding level, but their effect may be impossible to measure with any precision.

That said, it's still worth trying to figure out what we can deduce about how Canadian parties are spending their money now, and how they can invest it better in the future. And over the next little while, I'll plan to do just that.

No comments:

Post a Comment