Back in the day – roughly the third quarter of the 20th century – observers of American politics debated the wisdom of what seemed to be a Tweedledee-Tweedledum party system. Some thought it was pretty good. In the 1960s, political scientist Robert E. Lane hailed an emerging “politics of consensus in an age of affluence.” Government by agreement and expertise would replace divisive, ideological politics. Famed political columnist James Reston explicitly endorsed Tweedledee-Tweedledum parties that disputed only the details of the emerging welfare state. He counseled Republicans that their best route to success was “not by moving to the right and exaggerating the differences” with the Democrats, but by showing that they “can administer [liberal policies] more efficiently.”
Others thought the similarity in positions was terrible for democracy. Conservatives demanded A Choice, Not an Echo. Leftists bemoaned a “choice of a tweedledee as against a tweedledum” and liberals’ timidity to go to a third party.  In 1950, the American Political Science Association complained (pdf) that the parties’ differences were too poorly defined against one another and that they were insufficiently cohesive. Beware of what you ask for.
As is well-known, the political positions of the two parties have divided sharply since those days. This animation
 shows visually how members of the House separated out on a left-right dimension from roughly 1950 to 2000. Most of the shift has been due to the GOP moving right, exactly opposite to James Reston’s recommendation. Early analyses of this ideological polarization stressed that it seemed to be exclusive to politicians and the politically active, that average Americans were not drawn into this ideological fight. Recent work suggests that, while average Americans have still not gotten more ideological, they have become more tightly loyal to their parties as the parties have become more distinct. Party identification has almost become almost tribal. (See this earlier post.) Three new studies underline the power of party loyalty.
The three studies suggest how strong party identification can counteract what people know or even what they experience.
One: Respondents to surveys tend to describe conditions, especially economic conditions, in accord with their political affiliations. When a Republican is president, Republicans tend to say the economy is fine and Democrats tend to say that it is lousy; when a Democrat takes over the White House, the evaluations flip. To which Markus Prior and his colleagues at Princeton responded, “You Cannot be Serious…,” the title of their recent paper (pdf). In 2004 and 2008, a web survey asked a scientifically-drawn sample to describe current economic conditions. As would be expected with George W. Bush in office, Republicans were much more positive than Democrats. However, the researchers offered another set of respondents either some money or some strong encouragement to give the factually correct answers. The gap between Republicans and Democrats in evaluations of the economy was notably smaller for this group. The authors say that this shows that respondents know more than political scientists give them credit for. Another interpretation is that, without the incentive of money or pride, many respondents would rather cast a vote for “their side” than give a correct answer to a survey. (By the way, this is how I interpret the high percentage of Republicans who have said that Obama is a Kenyan or a Moslem — that many do not actually believe it, but saying so is a way of saying that they hate the guy.)
Two: James Druckman and his colleagues (here; gated) set up another survey experiment that pits party and knowledge. They developed pro and con arguments, both strong ones and weak ones for each side, on two issues – oil drilling and the immigration DREAM Act. They presented various combinations of the arguments to respondents. As you would expect, Democrats tended to be more resistant to drilling and more approving of the immigration bill than Republicans were. Still, when presented with stronger arguments for one side or other of an issue, respondents tended to move their opinions in the direction of the strong arguments. Ah, sweet reason. The researchers, however, told another set of respondents who read these arguments that officeholders from each party were very unified in opposition to the other party on the issues. These respondents, alerted to party polarization on the topics, moved their opinions toward the positions of their own parties, largely neglecting the quality of the arguments. Ah, sweet … something else.
Three: Yotam Margalit at Columbia compared party loyalty not to knowledge, but to experience (here, gated). He tracked thousands of respondents across four interviews conducted between July, 2007 and March, 2011. Given what happened to the economy, some folks lost their jobs during this period. A notable portion of the job-losers changed their opinions on a question asking whether they would support more government spending to help “the poor and unemployed” even if that meant raising taxes. Some Republicans became more favorable in the year after they had lost their jobs. However, Margalit also found that once respondents regained employment, their opinions reverted; it was a passing liberal moment. Party position mattered more.
None of these three studies can tell us whether this kind of polarization and party loyalty has increased over time. They do, however, show how deep the party commitments are in era when the parties are so far apart. Makes the Tweedles seem perhaps alluring.
 Lane, “Politics of Consensus,” Am. Pol. Sci. Rev., 1965.
 Reston, “Washington: On Tweedledee and Tweedledum,” New York Times, Nov. 7, 1965.
 http://www.apsanet.org/~pop/poole.gif (from McCarthy, Poole, and Rosenthal)