The Monkey Cage: Left-right ideology of voters, congressmembers, and senators

Andrew Gelman of the Monkey Cage says:

Comments

[James Grimmelmann]

This graph cannot possibly be correct. Out at the tails, it claims that significant portions of Congress take ideological positions that essentially no one holds. There are plenty of crazy Republicans in Congress, but they have constituencies; that's why they're able to be crazy. There just aren't enough people in those tails for these results to make sense.

[David Lloyd-Jones]

You'd think there ought to be some facts that could be brought to bear on a notion like this -- even if it's only something flaky like an indicator cobbled together out of lobbyists' favorable/un- voting tables.

The reason I find this factless empty speculation annoying is that in the real world I often find that things which are intuitively bimodal to me turn out to have strong central tendencies. I think that the tendency to see things as drifting into extremes or divisions may be one of those Tveskyesque dementities.

[sm]

Don't you think this left/right baloney is a waste of electrons?

[A]

I assume "ideological position" is defined in an abstract and meaningless way. On issues that I care about, such as leaving Iraq, I understand that a majority of voters is for it (presumably the liberal position), but Congress apparently against it, as shown by continued votes for Iraq war appropriations without a time line for withdrawal.
I suspect this news item is propaganda telling Democrats in Congress to be more 'centrist' (and forget about all these 'liberal' ideas); although the right edge of the graph indicates that there are many 'conservative' congress critters who are more conservative than the voters (perhaps due to well-moneyed contributors).--

[Jonathan]

The statements above that the graphs cannot be right are themselves baseless. I find it quite plausable. My basis for this is the book "Off Center" by Jacob S. Hacker and Paul Pierson. It documents that Republicans especially are far from the norm ("positions essentially no one holds"), Democrats less so, and goes into detail about how this has happened.

Even after 2006, which moved Congress to the left, my take on the graph is that the Republicans are farther right than the Democrats are left.

I don't know what Herron based this on, but remember that political scientists have been collecting data on these things for decades.

[Ken Houghton]

Click through to Gelman and you'll see the graphic that should be here, which is 8.10.

[MaryCh]

As an old DFH, I have to ask: where would 1968, 1980, 1992, or 2000 fall on this spectrum? Maybe Paul and Jacob's book will tell me?

[JoeV]

The camel is in the tent!

[John Emerson]

Conservative (Republican) voters are fewer than liberal (Democratic) voters but conservative congressmen are more common than liberal Congressmen. Furthermore, conservative Congressmen are more conservative than the voters who elect them, and by and large conservative Republicans in Congress dominate the moderate Republicans.

Republicans work the system better.

[wolfgang]

I think there could be an easy explanation for this double peak.
The overall population is roughly described by a normal distribution,
proportional to e^(-x^2), where x is the 'ideological parameter'.
However, I assume the motivation to go into politics to be proportional to
const + x^2, in other words the more extreme a position one holds (left or right) the more likely one is motivated enough to go into politics.
Therefore, while the distribution for the voters is roughly normal, the
distribution in congress is something like (const + x^2)*e^(-x^2) and
exhibits a double peak if const is small enough (i.e. moderate people are
not very motivated to go into politics...)

[wolfgang]

I should add, that according to my 'theory' the small double peak of the general public is a 2nd order effect - the backreaction of the double peak distribution of the politicians on the people via media etc.

[ogmb]

There's all kinds of radical cranks going into politics, but one should assume them to fall by the wayside through the force of the Median Voter Theorem (which also seems to fuel the doubts about the graph expressed above). The cause of the emerging radicalization of the U.S. Congress is more likely that individuals with radical viewpoints are more inclined to get engaged in politics -- via primary voting and campaign contributions -- rather than as politicians.

On the question how this looked in 1968 or 1982, the trend towards a polarized Congress started in 1981, kicked into overdrive around 1995 and peaked in the late 90's. Poole & Rosenthal have a number of interesting animations on the development at http://www.voteview.com and discuss their hypothesis for the causation (income inequality) at http://www.polarizedamerica.com.

[MaryCh]

Thanks, ogmb

[Stephen Purpura]

Gelman is a stand-up guy. So let's assume the graph has some basis in reality. There is a simple economic explanation for Congressional members to take positions more polarized than the electorate (that's what this graph shows) -- the multiplayer, multi-round game of negotiation in the Congress rewards polarized positions.

FYI -- I can't take credit for this. Zeckhauser had a lecture about this during API-320 at Harvard. ;)

[Ginger Yellow]

It's interesting that the House skews to the left of the Senate, for both liberals and conservatives. I'd have expected it to be more extreme at both ends. Is this a legacy of the 2006 elections or was it the same before?

[baileyman]

Ok, so take healthcare, where nearly all the voters want it, or Iraq, where nearly al the voters don't want it, or impeachment, one of the two (with Iraq) top issue so of the 06 election. then compare these against the pols positions, and the voters ain't in the middle. How many issues to i have to go through like this before it becomes obvious this is a big lie?