Multi-axis political spectrum

[Ravi Pandya <ravi>]

Your remark to Ed on the phone yesterday about disliking a
single-dimensional analysis of political positions reminded me of an
interesting footnote in the Public Choice book I'm reading:

    Poole and Romer employ a least-squares multidimensional folding
    technique to map the ratings of the House of Representatives by 36
    interest groups into a multidimensional policy space. They find that
    three dimensions suffice to obtain all of the predictive power
    inherent in the ratings, with a single liberal conservative dimension
    providing 94 percent of the explanatory power. [1]

Of course, all this means is the the political positions of the
existing members of the House of Representatives, as judged by 36
existing (presumably mainstream) PACs, can be mostly represented along
a single axis. It says nothing about nothing about the possible space
of variation. It's interesting, nonetheless.
	--ravi


[1] Poole KT and Romer T. "Patterns of Political Action Committee
Contributions to the 1980 Campaigns for the United States House of
Representatives," Public Choice, 1985. 47(1) pp 63-111.  Cited in
Mueller DC. Public Choice II, 1989. Cambridge University Press.