Customer Reviews

A Unified Theory of Voting: Directional and Proximity Spatial Models

5 star:		(0)
4 star:		(1)
3 star:		(0)
2 star:		(0)
1 star:		(0)

 

 

In theorizing about why voters vote the way they do, two spatial types of models have become dominant: proximity-models that reason that voters vote for candidates nearest to them, and directional models that reason voters vote for candidates who are most like them in the general direction they prefer. These two model-types have been seen by some as mutually exclusive.

Merrill and Grofman succeed in devising a formula that combines the two models. The mathematical reasoning to justify the formula is not very difficult: any undergraduate should be able to follow it. But to be able to actually think of the formula from scratch, that's impressive.

Furthermore, the authors show how their combined formula holds up empirically when tested against the other models. Obviously, their model does pretty well - but what's more important for the reader (at least, for me) is how they travel from mathematical reasoning to applied research. That's what science is for, and it sure beats many of the authors these days who contend themselves in dreaming up myriads of possibility or impossibility-theorems. It actually makes their ideas relevant.

On the downside: the book could have used a better proofreading. To name two examples: on page 42 in footnote 5, their derivation of the damped directional utility function seems to be wrong: there's an "=" sign where there should be a "*" sign. Second, it is not entirely clear from the text what values the mixing parameter beta is allowed to take. On page 45, they show an example of it varying between 0 and 1, but elsewhere they state (p 70, footnote 4) they state beta is not restrained to lie between 0 and 1, even though it is a footnote to a sentence stating that the values of 0 and 1 are actually theoretically relevant values.

These are minor errors though. The biggest question I have about the book is more theoretical: they assume that what the proximity models see as 'policy position' is conceptually equal to what directional models see as 'policy intensity'. To my mind they do not justify this enough - this is not necessarily criticism from my part, I'm actually curious about this and it was a thought-provoking step for them to take.

So, all in all: a mustread for anyone studying voting behavior. Despite some editorial errors, it sums up the debate in voting behavior and provides an accessible and very useful addition to it.

ADDED LATER:

Someone who is more familiar with the economics literature on a similar subject (utility functions) pointed out to me that ever since Tversky and Kahneman's 'Prospect Theory', economists have been modelling functions along similar lines - to such an extent that what Merill and Grofman come up with as a mathematical formula even looks like a very close approximation of what Tversky and Kahneman described as how an asymetric utility function would look like. It's a pretty big ommission for Merill and Grofman to not even discuss Tversky and Kahneman (even though I missed it at first, too, but I'm not writing a book about it) and one has to wonder how much more interesting their book would have been if they would have made this comparison.