Are Democrats Less “Disciplined” than Republicans?

In response to my post on Drew Westen’s latest, a few commenters took issue with a secondary point. (My primary point, that Westen mischaracterizes the partisanship of the mass public, attracted less dissent.) My secondary point was that, despite this stereotype that Democratic politicians are less disciplined than Republicans—more fractious, harder to coordinate, etc.—Democrats and Republicans in Congress have essentially equivalent levels of unity on roll call votes.

I honestly believe that most people who say that Democrats are less disciplined than Republicans do not know this fact about unity on roll call votes. That’s why I pointed it out.

Is roll call voting the entire story on party discipline?  Of course not. Let’s review some other evidence:

  • As I noted in the first post, Democrats and Republicans in the mass public vote for their party’s candidates at the same (high) rate. In presidential elections, party loyalty is approximately 90%. Here is data from the 2008 exit polls, for example.
  • Democrats are frequently portrayed as less ideologically cohesive than Republicans. Is this true in the mass public? It would sure look that way, given that there are more self-identified conservative Democrats than liberal Republicans. But as we also know, many people who identify as conservative support an ostensibly liberal position: government spending on various social programs. (See also this.  Or this.)  So I don’t know of an easy way to score this one for either party.  (I’m not prepared to take a dating site’s data as representative of the broader public.)
  • Networks of interest groups. As it turns out even though Democratic groups are supposed to be factionalized—e.g., unions, minorities, feminists, environmentalists, etc.—networks of Democrats groups are actually denser than GOP networks, as measured by patterns of endorsements and financial contributions.  See this earlier post featuring research by Matt Grossmann and Casey Dominguez.

Okay, but quit dodging the issue, Sides.  Westen specifically mentions parties “in Congress.”  One commenter noted that roll call votes were not a good measure of unity because leaders will only bring bills to the floor when the party is unified.  Yes, that happens.  Although not always. But let’s say, for the sake of argument, that party unity scores overestimate party discipline.

If they do, then how else might we measure discipline?

Some commenters seem to think that the Democratic caucus is more ideologically heterogeneous than the GOP’s.  One holds up taxes as an example.  The GOP will oppose tax increases in lock-step; but the Democrats do not support them at similar levels.  But contrary examples also emerge, notably the GOP’s attitude toward government spending—which, as in the mass public, is conflicted at best.  Which may explain why, despite Grover Norquist and “starve the beast” and “drown in it the bathtub” and so on, Republicans disagree about levels of government spending.   So you get stories like this.  Ultimately, I know of no systematic evidence—beyond what we could learn from roll call votes (and I don’t even know what those would show)—that the GOP’s congressional delegation is systematically more homogeneous across time and across issues.  I welcome someone to provide some evidence, as long as it’s more than just another anecdote.

Others think that the Democrats can’t coordinate on a message, which means, I guess, that if we put 5 Democrats on the Sunday talk shows, they’ll say 5 different things while Republicans will all recite the same scripture.  Do we have systematic evidence for this?  I don’t know of any either.  Again, I welcome it.  (UPDATE: See Sebastian’s comment to my earlier post, which he posted after I published this post.  He found a paper that examined the parties’ message discipline on the estate tax and found comparable levels of discipline.)

Ultimately, my sense is that the stereotype—Democrats are less disciplined than Republicans—is just not well-supported by systematic evidence that looks at all facets of parties, including voters, networks of groups, members of Congress, etc.  In other words, people are too sure of this generalization than the evidence can support.

Which brings me to a final, and perhaps more interesting question: where does this stereotype come from?  A colleague and I discussed this today.  The gist of our conversation:

  • Republicans were in the minority in Congress for many years, which meant that they had a greater incentive to stick together and oppose the Democrats.  This gave the misleading impression that they were all the same and Democrats were always infighting.
  • After the McGovern-Fraser primary reforms, it seems liked the Democrats were all helter-skelter, whereas the GOP always had a designated frontrunner.  I think 2008 and now 2012 should start to call that impression into question.  The GOP is certainly not immune from uncertainty about its nominee.
  • Republicans won more presidential elections than they lost for several recent decades.  Democrats blamed bad messaging, bad campaigns, bad candidates, etc.  There was a tendency to reach out, well, to people like Westen to tell them The Message that would enable them to win again. Needless to say, this overstated the role of messaging.

I welcome other explanations or reactions.

19 Responses to Are Democrats Less “Disciplined” than Republicans?

  1. Kevin October 31, 2011 at 8:23 pm #

    The question of message discipline in media seems like a good candidate for the application of automated text analysis, no?

    • Trey November 1, 2011 at 11:04 am #

      I believe Nick Beauchamp at NYU is working on something along these lines. Also, Justin Grimmer at Stanford has also worked on this. Not the exact question you’re posing, but their work seems like a starting point for a relatively straight-forward application.

  2. David Shor October 31, 2011 at 8:33 pm #

    Two things that jump out:

    1) Plot DWNominate scores of GOP senators and Governors against state PVI. You’ll see a linear relationship with Democrats and a flat relationship with Republicans.

    2) Look at ideological composition by party. There are many more conservative Democrats than Liberal Republicans, and the Republican party is much more dominated by conservatives than the Democratic party by liberals.

  3. db October 31, 2011 at 8:35 pm #

    Hacker seems to argue (haven’t read the book yet) that there is evidence this this effect:

  4. frankcross October 31, 2011 at 9:57 pm #

    I’m suspicious that it is only liberals who think Democrats are less disciplined, speaking from their ideologically influenced perceptions. I have heard a lot about RINOs over the years, and Republicans seem to have plenty of their own gripes about more liberal party representatives.

  5. david October 31, 2011 at 11:03 pm #

    In response to the question of where this stereotype comes from, I can offer two names: Will Rogers and V.O. Key.

    Rogers, “I belong to no organized party, I’m a Democrat.”

    And Key, who found a bewildering level (and variation) of factionalism and interest heterogeneity in the democratic parties of the south.

    The point being twofold: 1) it is a stereotype that is much older than McGovern-Fraser and 2) it was a trope that was formed in respect to southern democracy, both in their internal state politics and in their role in a national coalition increasingly dominated by northerners.

    My intuition for where this stereotype emerges is this: it begins as a characterization of the democratic party in the south post-1900, when most politics occurs under the democratic tent. Subsequently, this trope of a disorganized political party is extended to the national level to explain the growing divergence between a stable northern wing that came out of nowhere in 1933, and the old southern democracy that no longer constituted the core of the party. While Key was interested in how southern factionalism turned into the solid south in congress, it is often forgotten that he found higher (and increasing) levels of cohesion among northern Democrats. This trope resonated at the national level with the new deal through to the 1980s. Not sure why it continues to resonate, but I suspect it began in the genuinely greater heterogeneity of the democratic party in the south, which was the democratic party from 1890-1920s, and the greater heterogeneity of the national party from 1935-1970s.

  6. Scott Monje November 1, 2011 at 1:30 pm #

    Here’s “another anecdote” that I think highlights a potential classification issue. John Boehner negotiated a debt-ceiling deal with President Obama and was then forced to retreat owing to pressure from his own backbenchers. The result was a solid GOP position, but it wasn’t because the Republicans lacked factional differences and it wasn’t party discipline if “discipline” means the leadership imposing its views on its members.

  7. Seth November 1, 2011 at 4:44 pm #

    This is just a quick-and-dirty analysis, without having had much time to think it through, but I’ll post it as a piece of evidence for the idea of, well, heterogeneous heterogeneity. Ideological heterogeneity doesn’t necessarily translate to party discipline, but it should be easier to produce better discipline among a more ideologically coherent group.

    As far as ideological heterogeneity goes, we might expect the size of a majority in congress to play a role. Generally, the larger your majority, the more heterogeneity we’d expect to see, since it requires winning more moderate districts (e.g. Blue Dog Dems). Significant minorities should produce greater homogeneity, since it means a party mostly won on its home turf, so to speak.

    To look at this, I calculated the variance in each party’s 1st Dimension DW-NOMINATE scores by congress (house only), and compared them to the size of a Democratic majority (or minority) in a given year. Since 1964 (I wanted to start after the Civil Rights Act), there is a very strong relationship between the size of the Democratic majority and the variance in NOMINATE scores among the Democratic House caucus. Specifically, the larger the percentage of congressional seats held by Democrats, the larger the variance in scores (and the more heterogeneous the Democrats appear to be). Variance ranges from an average around 0.022 in congresses with significant Republican majorities to around 0.040 in years with significant Democratic majorities.

    The Republican caucus, however, does not change much. Strangely, they appear to become slightly more heterogeneous in years with Democratic majorities. But the variance ranges from about 0.023 with large Republican Majorities to 0.026 with large Democratic majorities.

    Tossing the data into an OLS, the coefficient for Democrats (0.078, SE=0.019) is a little over 5 times the size of the coefficient for Republicans (0.015, SE=0.006). The R-Squared for the Democratic model is also much higher (0.413) than the R-Squared for the Republican model (0.154).

    I also tried running it starting in 1982 (i.e. post-Reagan Revolution), and found an even larger difference, with the Democratic model basically unchanged (beta=0.074, SE=0.017) and the Republicans showing basically no relationship between size of majority and party heterogeneity (beta=0.007,SE=0.005). So, in essence, the last 30 years have seen great heterogeneity among Democrats when they hold large majorities, but more homogeneity when they are in the minority. By contrast, the same period has seen fairly consistent homogeneity among Republicans, regardless of the size of their majority.

    Scatterplots here:
    and here:

  8. William Ockham November 1, 2011 at 5:30 pm #

    John Sides and I seem to be talking past each other and I’m don’t understand what the problem is. I described his position as being that “Party unity is completely, accurately, and precisely measured by the data on roll call floor votes.” In response, he says that I’m engaging in hyperbole. But I think my interpretation was completely fair, given the flat statement of fact that he makes in his earlier post and the links he includes to his previous posts on the subject. To quote one (and the upper case is in the original):


    The graph in question is a chart of the parties’ cohesion on floor votes. I’m really not sure how one is to interpret that statement in way that doesn’t include the notion that the data in the chart completely answers the question. Who is engaging in hyperbole?

    My point is really pretty simple. Party unity on roll call votes is something that both unified parties and disunified parties can achieve. I’ve tried to make that point with logic, parables, and examples from my personal experience of other types of organizations that achieve final vote unity which I know sometimes reflects true unity and sometimes just papers over serious core disagreements. I know these aren’t serious political science studies. They are attempts to suggest an alternative model for understanding the effective cohesion of political parties that doesn’t rely floor votes as the core data element.

    I’m willing to let that go. I’m not willing to let go of a complete mischaracterization of my point about taxes. Again, I don’t think this is hard to understand. There are certain types of position that are asymmetric with respect to party cohesion. That is, by their very nature these positions invalidate the notion that measures of party unity are the only important measures of party cohesion. The Republicans actually have lower party unity with respect to the Norquist pledge than the Democrats, but their position on taxes is more cohesive because the pledge is “always oppose higher taxes on anyone” and the alternative is “sometimes prefer higher taxes on some people”. I can tell that Sides still doesn’t get that because his counterargument is completely inapposite. Democrats certainly do not support the position that we should always have more government spending. Even Keynesians don’t support that. The Democrats don’t even coherently support the Keynesian position. I actually think the Republican position on spending is pretty clear: always support higher defense spending, only support higher domestic spending when there is a Republican president, and never support higher spending to benefit the poor. That’s more coherent than the Democratic position on spending.

    I’m not invested in proving that the Democrats are less cohesive than the Republicans, although I suspect that is the case. I’m trying to encourage Sides to stop yelling at political pundits when his facts are irrelevant to the issue at hand. It’s just not productive.

    • John Sides November 1, 2011 at 7:48 pm #

      William: Roll call voting in congress is not “irrelevant” to the question of party unity. Period. It’s not a perfect indicator. I never said it was. If you interpret my yelling at Gail Collins otherwise, so be it.

      In response to yours and other critiques, I’ve summoned a bunch of other evidence — yet more in this post— that Westen’s thesis lacks compelling evidence. You haven’t really addressed any of that. In fact, you seem happy to agree with me regarding Westen. Great!

      Nevertheless, we are apparently we are talking past each other. You think the Republicans’ position on the Norquist pledge is more cohesive than the Democrats’ position, even as you acknowledge that they are less unified than the Democrats on the pledge. The latter seems much more fundamental to me. Who cares about the “cohesiveness” of the Norquist pledge if it’s just something to which many Republicans pay lip service? Or to put it differently, why is this particular type of “cohesiveness” important if Republicans aren’t “unified” in the policies that they actually make?

      • William Ockham November 1, 2011 at 9:54 pm #

        “they are less unified than the Democrats on the question of government. “

        Is this a typo? If not, can you explain what you mean? I do not understand what you are claiming I that I acknowledge.

        • John Sides November 1, 2011 at 10:24 pm #

          Typo. Fixed the comment.

      • William Ockham November 1, 2011 at 11:25 pm #

        Floor votes are irrelevant to the claim that Westen made. That is a fact. Westen did not use the term “party unity”. He made a specific claim that you quoted:

        Because of their attitude toward authority and hierarchy, Republicans in Congress are more likely to follow their leaders… Democrats on the other hand react so strongly against taking “marching orders” that they can scarcely stay on message even if their political lives depend on it (which they often do).

        Where does he mention floor votes? Where does he mention ideological homogeneity? Where does he mention rank and file voters or networks of interests?
        I haven’t addressed your arguments because my point is that you are missing the point. You made the leap from attitudes and messaging to floor votes as a proxy for your definition of party unity. Perhaps there is a case to be made that the data you cite disproves Westen’s thesis, but you haven’t made it. You have just assumed it.

        • John Sides November 2, 2011 at 9:43 am #

          William: Westen’s claims have testable implications about how voters, party networks, and members of Congress should behave. Party unity votes are one measure of how well rank-and-file members follow their leaders. They are absolutely a relevant piece of data. In response to your comments and others, I have presented many other pieces of available data, including the best measure we have of the “attitudes” of members of Congress — none of which support Western’s portrayal of the two parties. Rather than grapple with any of that evidence, you dismiss it out of hadn. Rather than respond to my (sincere) questions for you about “cohesive” vs. “unified,” you prefer to launch further accusations at me about “leaps” and “assumptions,” etc., etc., etc. I think we have passed the point where further conversation is fruitful.

  9. Seth November 1, 2011 at 7:42 pm #

    Okay, this is what happens when you scramble to post something before an appointment. Here are the variances in NOMINATE scores by party (with majority noted) since the Gingrich revolution:

    Congress Year DemVar RepVar Dem Maj
    105 1996 0.02252228 0.02311929 0.4761905
    106 1998 0.02305096 0.02379706 0.4874142
    107 2000 0.02120617 0.02406902 0.4829157
    108 2002 0.02069020 0.02280014 0.4738041
    109 2004 0.01754058 0.02315547 0.4622426
    110 2006 0.02039880 0.02383918 0.5403587
    111 2008 0.02273833 0.02558051 0.5887640

    This seems to go against what I was saying before, at least in terms of the absolute differences. The Dem variance in their largest majority year is still smaller than the Republican variance in any year in the sample. Obviously, small sample caveats apply, but this does at the very least suggest that recent congresses (or houses, at least) have seen no greater homogeneity among Republicans than among Democrats.

  10. John Jay November 1, 2011 at 7:45 pm #

    “Republicans won more presidential elections than they lost for several recent decades.”

    Going back 40 years, the count is 6-4 for Republicans (5-5 in the popular vote). Does that small difference in results really explain why one party would be more likely than the other to overestimate the importance of The Message?

  11. Sebastian November 1, 2011 at 8:58 pm #

    “Which brings me to a final, and perhaps more interesting question: where does this stereotype come from? “

    Are we sure this is actually a widely held stereotype or is this a stereotype among liberals?
    My sense is that both sides of the political spectrum in the US describe their opponents as powerful political machines, ruthless and coldly calculating in their actions.
    At the same time they fret about the weakness of their leaders, the lack of unity among them, they (verbally) tar and feather moderates etc.
    If you’re a liberal and a significant part of your news on political infighting come from Washington Monthly, TNR, Thinkprogress, Huffpost etc. you’re going to think the Democrats lack unity and leadership.
    If you’re a conservative, and a large part of your news on political infighting come from Red State, NRO, Weekly Standard you complain about Boehner’s weakness, about RINOs in Congress etc.

    (Whenever I feel dispirited, I spent 30mins on NRO reading about a country in which a socialist President has pressed through radical social reforms and is transforming the country into a European Social Democracy and how his Lackeys in Congress shirk all rules and convention to help him).

    • Scott Monje November 1, 2011 at 9:50 pm #

      If I recall correctly, Robert Jervis said years ago (in “Perception and Misperception in International Politics”), that people tend to view their enemies as more cohesive, more unitary, more purpose-driven than their own side, or words to that effect.

      That would also be supported by psychology’s “fundamental attribution error,” whereby people see their own negative actions–and those of their close allies–as forced upon them by circumstances but interpret the negative actions of others (especially enemies) as indicative of their true nature.

  12. Sebastien November 2, 2011 at 1:59 pm #

    James Fowler has a couple of articles building a network of Congress based on cosponsorship data. Basically, you create a network where a link’s strength is a function of how many times two congressmen cosponsored the same bill, or cosponsored a bill sponsored by the other.

    You could think of building a discipline indicator based on that, e.g. comparing the average link’s strength among the two parties, etc…

    (I tried my hand at building the network here: