The Center Cannot Hold, and It's A Center-Left Center

Another cool chart from the Monkey Change which compares the ideological distribution of the electorate to that of the House and Senate. All three curves are bimodal, but the voters clump closer to the center than do the members of congress. Were I David Broder I would argue that this shows the wisdom of the masses and the baleful influence of special interests in pushing party leaders to extreme positions, but realistically it probably reflects the fact that members of congress are much better-informed about politics than are average voters, and therefore members of congress tend to have more coherent ideological viewpoints.

The other interesting point is that the electorate seems, on the whole, slightly left of where the congress is. The "trough" of the voters' bimodal distribution is to the left of the House and Senate troughs, and the left peak in the electorate is substantially higher than the right peak but that's not true in the congress. You should probably expect congress to be somewhat to the right of the public thanks to the fact that the current gerrymander mostly favors Republicans and the apportionment of the Senate tends to overrepresent conservative parts of the country.

Comments

I think that there are tails in the population that
are non-zero but FAR more extreme than anyone actually in the house or senate. The curves show the opposite.

Posted by catclub | July 21, 2008 12:43 PM

This finding fits with endogenous party equilibria in which candidates must first win a party primary before contesting the general election.

Posted by Martin | July 21, 2008 12:50 PM

realistically it probably reflects the fact that members of congress are much better-informed about politics than are average voters

As Martin says, this reflects that we have a two-party system.

Posted by Tyrone Slothrop | July 21, 2008 12:54 PM

This sort of thing is fine as a parlor game, but any study that starts from the premise that most voters have consistent ideological beliefs is badly flawed. Converse back in the day showed that most voters do not think ideologically, and to my knowledge no recent studies have refuted him. Voters may be able to use the terms "liberal" and "conservative," but the extent to which self-described "liberals" hold beliefs that are liberal, or self-described "conservatives" hold beliefs that are conservative, across the board, is very limited.

And those damned DW-NOMINATE scores for members of Congress basically conflate party voting with ideology and thus are actually worthless--worse than worthless, actually, because they mislead folks like Matt.

It might be hard for folks reading this blog to believe, because almost all of them (you) think in ideological terms, but that (you) is (are) the exception in American politics.

Posted by Number Three | July 21, 2008 1:06 PM

So, MY, now I can say that the charts are from MY, because you posted them?

How 'bout something vaguely resembling proper credit--that chart ain't from Monkey Change [sic]. It's like linking Sullivan's weekly poem and calling Andrew a poet.

Posted by zaleriana | July 21, 2008 1:23 PM

...but realistically it probably reflects the fact that members of congress are much better-informed about politics than are average voters, and therefore members of congress tend to have more coherent ideological viewpoints.

I suspect it reflects that the only citizens motivated enough to run for political office in the first place are people who are more passionately ideological to begin with. The two-party system influence and their greater access to information likely hardens their political viewpoints over time, but I doubt it's the root cause.

It would be interesting to test this, maybe with the distribution of all the people who have run for political office but lost, or the distribution of people when they first register to run vs. once they've been in office for a year.

Posted by Troy | July 21, 2008 1:28 PM

I think there was somebody, somewhere, who once asserted "we are not as divided as our politics suggest".

But it is believed this man was a lunatic, because there is no sense in what he said.

Posted by eyelessgame | July 21, 2008 1:49 PM

...realistically it probably reflects the fact that members of congress are much better-informed about politics than are average voters, and therefore members of congress tend to have more coherent ideological viewpoints.

Matt, I don't think that statement is warranted at all. OTOH I think Troy's explanation makes a good deal of intuitive sense.

I also agree that the two-party system "hardens their political viewpoints over time". In other words, someone with a strongly Democratic position on a few issues runs as a Democrat. Now they have a bunch of other issues on which they feel less strongly, but precisely for that reason they're more subject to influence from their peers. Over time their position moves closer to "the team's" on all those other ancillary issues. (Of course the same happens with Republicans.)

Our political discourse is tragic because it makes everything into a two-sided issue, and ignores those issues (ahem, the War on Some Drugs) where a significant segment of public opinion isn't in line with either the Red Team or the Blue Team. Really, those of us on neither team get the message very quickly that our input is not desired or valued. What a country!

Posted by Equal Opportunity Cynic | July 21, 2008 1:53 PM

Oh dear. Somebody's "estimate of ideological positions" suddenly becomes "the ideological distribution of the electorate" on Matt's blog.

Why don't you get your red crayon and draw your own curvy-thing and label it "the ideological distribution of the electorate" and compare it to other curves you made up. It would have about as much authority as the one you posted.

Posted by Oh Dear | July 21, 2008 2:24 PM

Voters may be able to use the terms "liberal" and "conservative," but the extent to which self-described "liberals" hold beliefs that are liberal, or self-described "conservatives" hold beliefs that are conservative, across the board, is very limited.

I actually find it depends on how you ask the questions and what the emphasis is. Let's consider your typical "conservative" or even "independent but leaning Republican" voter. I'm no pollster and the plural of anecdote is not data, but allow me to generalize from my own experiences:

If you ask a question focusing on the "morality" of something social conservatives oppose and then ask "should that be legal?", they will give the social conservative answer. If you ask the question in terms of "even if you think [X] is immoral would you support actually putting people who do [X] in prison?" or otherwise getting them to think of the practical consequences of what it means (in terms of how a moral law would be enforced, etc.) they will take a socially liberal point of view.

Where do such voters fall on the spectrum? I dunno ... How do they vote? We all know how ...

Typically, with white voters, the GOP has been able to get them so frightened of gay married terrorists having abortions that they vote for the GOP. OTOH, various elements of the GOP appeal, for reasons we have discussed on this blog (the implicit racism, GOP anti-intellectualism not playing well in a religiously oriented community in which your (stereo)typical pastor has a Ph.D. or equivalent doctoral level degree, etc), simply do not resonate well within the African-American community so people in that conservative mode still typically vote for Democrats.

Of course the challenge for Democrats then is to make sure that these voters think in concrete terms of "what the law will mean practically" rather than "it's icky, so let's make it illegal". Of course, one challenge here is that, due to religious and cultural "anti-legalism" (re-enforced by the anti-lawyer attitude of the right), such people are not wont to think practically in terms of law which they find inherently stupid (which also leads them toward more "libertarian" views on, e.g., government regulation even as they support expansive moral legislation -- whence the GOP platform).

OTOH, and to get back to your point, I'd actually say that, for many voters, the opposite of your point holds. I know people who, for various reasons, hate to be labeled as conservative or liberal -- and certainly not as Democratic or Republican -- but who regularly trot out partisan talking points.

They may claim to only be, e.g., social conservatives but when push comes to shove (and as I described in the previous thread on this vis-a-vis, e.g., their inclination to support Joe Lieberman, who is certainly to the right of, e.g., Obama on economic issues but not so far to the right, modulo a lot of moralizing -- which they claim to dislike when a real Democrat does it, but IOKIYAR or a DINO I guess? -- of Obama on social issues) they prefer (in a "which of these two blurry lenses is better, #1 or #2?" sense) more economically/foreign policy-wise conservative Dems. to more liberal Dems. even though they claim economics/foreign policy is where they are more sympathetic to the Dem. position.

Do y'all have any other explanation for this besides that even though such voters claim to be "social conservatives but more liberal on other issues" (or, similarly, "economic conservatives, but more liberal on other issues"), their preferences indicate an accross the board conservatism?

FWIW -- others have pointed out a while back that the GOP is somehow better at getting one-issue voters to drink their kool-aid. Perhaps part of the issue is that a social conservative accepts a certain "Pauline" critique of "the law" that renders said social conservative sympathetic to GOP "government action in economic issues is futile" arguments -- even though social and economic conservatism seem quite irreconcilible, they actually fit together better, in terms of the underlying world view, than the different "faces" of liberalism? Or is the GOP simply better at selling (because they are more "business like"?) kool-aid than the Dems. are?

because they mislead folks like Matt. - Number Three

And more-so Mixner who'll argue that Joe Lieberman is a liberal, so why are we moonbats so unhinged we even hate Holy Joe.

Posted by DAS | July 21, 2008 2:30 PM

... apportionment of the Senate tends to overrepresent conservative parts of the country.

By that I assume you mean the smaller states. That comment made me wonder, so I went and counted the party affiliation of the senators from the 25 smallest states and found 22 democrats, one socialist, the indefinable Joe Lieberman who was once elected as a democrat and 26 republicans. So if we call it a 24-26 split, I don't think you can really blame rural voters for not supporting democrats. I was very surprised at how even the split was given 20 years of neglect smaller state parties have suffered under the DLC 50+1 strategy. I suspect in coming years, if the Dean-Obama 50 state strategy is maintained, that the Senate will look a lot bluer after 2012.

Posted by justawriter | July 21, 2008 2:41 PM

Looking at senators from the 25 smaller states hardly seems like the way to measure ideology. Bush won 31 states in 2004, which means that 62 senators come from states won by Bush. The Democratic senators in those states tend to be not particularly liberal. States that Kerry won are represented by Democrats (or Democratic aligned independents) at 31-7. States that Bush won are represented by Republicans at 42-20.

In essence, any Senate majority for the Democrats requires a good number of Democrats from Republican leaning states, making for a cautious conservative Democratic majority. Any Senate majority for the Republicans requires very few Republicans from Democratic leaning states.

Beyond that, looking at the bottom 25 is misleading - the bottom 25 states only represent about 15% of the population of the country. If you look at the top 15 states, which represent 65% of the population, the seats are split 19-11 in favor of the Democrats. The other 35 states, representing only 35% of the population, are split 38-32 in favor of the Republicans.

If you look only at the top five states, which represent more than twice as many people as the bottom 25 states, you find the split at 7-3 in favor of the Democrats.

So, sure, if you cherry pick your data, you'll find that Democrats do nearly as well in the bottom half of states as Republicans do. This is because of a few generally Democratic leaning small states (Vermont, Rhode Island, Delaware, Hawaii, Connecticut), and because Democrats have (often fairly conservative) Senators in a number of small Republican leaning states (North Dakota, South Dakota, Montana, West Virginia, New Mexico, Nevada, Arkansas, Nebraska), who outnumber the Republicans from small Democratic leaning states (Oregon, New Hampshire, and Maine). Note also that 18 of the bottom 25 states voted for Bush, and only 7 for Kerry, whereas the bigger states split almost evenly - 13 for Bush and 12 for Kerry.

Posted by John | July 21, 2008 3:21 PM

The interpretation that the chart shows that the members of Congress are "more extreme" than the public misreads the data. The chart shows that members of Congress are more conservative than members of the public. The highest peak for both House and Senate membership is the right-hand peak - well to the right of the great majority of voters.

It would be helpful to see a bar graph dividing congress and the public into five categories - more liberal, liberal, centrist, conservative, more conservative - and assigning 20% of the populations to each. Then you'd clearly see how much more conservative congress is than the public.

Posted by Bloix | July 21, 2008 3:25 PM

Why do we even bother with this pretty-but-point-less graphs? I realize that any-thing other than one-dimensional becomes more difficult (but, only slightly) to graph, but, what about libertarians? Populists/Authoritarians/Statists (depending on whose Nolan Chart you employ)? Paleo-conservatives? Left-libertarians? Who's "more conservative" - Ron Paul or Sam Brownback? "More liberal" - the (allegedly) libertarian-leaning Mike Gravel, or Ted Kennedy?

Posted by Nathan P. Origer | July 21, 2008 3:40 PM

Why do we even bother with these pretty-but-point-less graphs? I realize that any-thing other than one-dimensional becomes more difficult (but, only slightly) to graph, but, what about libertarians? Populists/Authoritarians/Statists (depending on whose Nolan Chart you employ)? Paleo-conservatives? Left-libertarians? Who's "more conservative" - Ron Paul or Sam Brownback? "More liberal" - the (allegedly) libertarian-leaning Mike Gravel, or Ted Kennedy?

Posted by Nathan P. Origer | July 21, 2008 3:41 PM

Nathan P. Origer,

It would seem to me that, when you ask people about their beliefs on various issues, they indeed do have viewpoints which are not reducible to a 1D ideological spectrum.

The question, however, is whether their voting behavior is similarly multidimensional or whether it can be reduced to a simple, linear ordering. If voting behavior can be so reduced, it makes sense to have a 1D ideological spectrum of elected officials (even if one can quibble with for what the NOMINATE ranking actually measures and doesn't measure).

There are at least two phenomena to consider:

(1) what people support depends on how the issue is framed: this sometimes reflects an underappreciated subtlety in the views of American voters but sometimes reflects a failure to consider fully consequences of, e.g., proposed legislative actions (e.g. someone who wants to ban abortions because they are icky but doesn't actually want anybody to be punished under an abortion ban -- nu? what is it they want? a ban? or not a ban?) ... sometimes what is reflected is merely American voters forming opinions based on their self-interest while lacking a full appreciation of where their self-interest lies.

Nu? People's voting behavior is not entirely rational, so even if their political belief system (in terms of answers to questions carefully worded to avoid bias based on asking questions in a leading manner) can be represented on a 2D spectrum (or even a 1D spectrum), their voting behavior might be very much more complicated.

(2) OTOH, sometimes people might answer questions about their political beliefs in a 2D (or nD, with n > 2) manner yet their voting preferences are perfectly 1D and consistent with the "standard" political spectrum. Many Democrats are consistently frustrated with this phenomenon -- we'll find a Christian social conservative who claims to be at least somewhat economically liberal or an economic conservative who claims to be at least somewhat socially liberal, yet their voting preferences are thoroughly one dimensional.

Posted by DAS | July 21, 2008 4:05 PM

DAS--

It's true that voting behavior may be reduced to a single dimension, both in Congress (NOMINATE, mostly, at least in 2-pty eras) and, perhaps, in the electorate. BUT, and this is a big but, there's simply no reason to label that dimension "ideology." This is the leap that reliance on unidimensional measures like this call for.

It would be better to call it "partisanship."

#3

Posted by Number Three | July 21, 2008 4:31 PM

What Oh Dear said. Why do we care about some random dude's "estimate of ideological positions of voters"? I'm sure anyone could pick some selection of votes to make the graph turn out any way they wanted.

Posted by Al | July 21, 2008 4:35 PM

OTOH, sometimes people might answer questions about their political beliefs in a 2D (or nD, with n > 2) manner yet their voting preferences are perfectly 1D and consistent with the "standard" political spectrum.

(Raises hand.) I have certain conservative positions (abortion, gun control, school choice, aversion to Al Sharpton) but almost always vote for Democrats, reason being that (a) regardless of my feelings on the other issues, they're in general outweighed by the general odiousness of the Republican positions on most other things, and (b) the general odiousness of Democrats on abortion is matched by the ineffectiveness of Republicans to decrease the number of abortions committed measurably more than Democratic social policies reduce them.

Posted by Adam Villani | July 21, 2008 4:57 PM

Number Three,

Fair enough. I guess the point is that people's ideological views can generally not be represented as a 1D spectrum. The question then becomes why does, in some cases, voting behavior require a higher dimensional representation than ideology whilst in other cases, voting behavior can be reduced into a 1D spectrum of partisanship?

Particularly unusual are those many people who claim to be "beyond partisan thinking" yet who reliably express preferences that can be represented on a single axis of "partisanship".

Posted by DAS | July 21, 2008 5:00 PM

Part of this is that the republicans are a coalition of Religious right and Economic conservatives that very often has folks that are one and not the other (but care a lot about that one!).

It would be useful if the article showed the same plot for several election years, say 2004 and 2000 as well to see if there was a trend.

Posted by joshspark | July 21, 2008 8:33 PM