Vinod B
Published in

Vinod B

Political Asymmetry

Political polarization has become a household term for explaining the current state of US politics. It refers to the level of overlap between the two parties on various issues. High political polarization leads to little agreement between the two parties, fostering an environment of heated arguments and little consensus. It can lead to the election of extreme candidates and gridlock that reflects the public’s level of polarization.

We see clear evidence of polarization today in US politics, but the level of polarization on the right and the left in the broader public and between the two parties has not been symmetric. Many sources have noted that the the Republican Party has become more polarized over time.

This asymmetry between the two parties, however, also appears among the general public. We can use the ANES (American National Election Survey), which is a yearly survey that seeks to explain why election outcomes occur to see this. It goes all the way back to 1964 with results up to 2012, but it is only accurate at the national level.

The thermometer question in this survey asks respondents how they feel about various groups on a 0–100 scale (referencing temperature) with higher numbers indicating a greater affinity.

With this data we will (1) examine how Democrats and Republicans (including leaners, meaning we include both those who lean towards either party and those who are registered as being in each party) separately rate conservatives and liberals on the thermometer scale and (2) do the same for the Democratic and Republican parties.

In the plot below we show the data for thermometer rating by survey year for liberals (in blue) and conservatives (in red) of each other.

ANES Thermometer Rating from 0–100 yearly by Democrats (Blue) and Republicans (Red) of Liberals (Left) and Conservatives (Right).

Focusing on the left plot of ratings for liberals, we see that Democrats perception of liberals has improved over time as they have rated them higher on the thermometer scale, indicated by the blue regression line. By contrast Republicans have rated liberals lower over time as the red line in that plot is downward sloping.

We see similar but reverse trends in the conservative thermometer plot on the right with Republicans rating conservatives higher over time (upward sloping red line) and Democrats rating conservatives worse over time (downward sloping blue line).

This all squares generally with what we know: Democrats have grown to dislike conservatives more over time and Republicans have grown to dislike liberals more than in the past, driving polarization.

The interesting thing in this graph though is the asymmetry that appears. The degree to which Republicans dislike liberals seems much higher than the degree to which Democrats dislike conservatives. This is because the red line in the left graph has a much steeper slope than the blue line in the right graph.

We can examine this asymmetry more rigorously in a series of regressions where we regress either the liberal or conservative thermometer on party identity, year, the interaction of party and year, and demographic controls such as race, education, age, and income. The base group here is Democrats.

For the liberal thermometer, the Republican time trend is -0.263 (-0.392+0.129) with demographic controls, meaning that each year the average thermometer rating of liberals by Republicans was about -0.263 points lower than the previous year.

Similarly, for the conservative thermometer, the Democratic time trend is -0.134. Note that this is about 1/2 the size of the Republican time trend, reflecting the asymmetry we saw in the plot above.

A simple hypothesis test for the difference between these two coefficients suggests they are statistically different, meaning that this asymmetry appears to be real. Republicans have grown to dislike liberals at 2x the rate that Democrats have grown to dislike conservatives.

We can compare the above, which focused on general feeling of other people, to feelings about each political party by Democrats and Republicans. We redo the regressions above, swapping in party thermometers as the outcome of interest. The base group is again Democrats.

Cross ratings (Democratic party by Republicans and Republican party by Democrats) are both negative overall and over time, but there is again an asymmetry. This time though Democrats appear to have grown to dislike the Republican party more than Republicans have grown to dislike the Democratic party.

The time trend for Republicans rating the Democratic party is -0.591 (-0.093-0.498) while the time trend for Democrats rating the Republican party is -0.653. This difference is statistically significant too.

Democrats have grown significantly less fond of the Republican party than Republicans of the Democratic party. Note though that there is less asymmetry here than we saw between Republicans and Democrats in rating people of the other side.

The asymmetry in polarization we see talked about anecdotally and within elected officials in the two parties also holds for individuals in the electorate. Republicans have grown to dislike liberals twice as much as Democrats have grown to dislike conservatives with the asymmetry reversed when we focus on feelings towards each political party.

This asymmetry has interesting implications for understanding current trends and debates. If Republicans dislike voters of the other side much more than the reverse, it suggests that political arguments appealing to tribalism and identity can be much more effective on the right because it plays upon this unfavorable view and animosity. That is kind of what we see today in the debates about gun policy, political correctness, and many other contested issues.

The debates appear to center more around taking something away or infringing upon an identity issue than about the best policy to solve actual problems.

To get back to a less polarized and tribalistic politics, we need to get the two sides on a less asymmetric footing, in addition to reducing negative feelings of the other side for everyone.

Data Sources:

ANES survey cumulative file

Github code



Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store
Vinod Bakthavachalam

Vinod Bakthavachalam


I am interested in politics, economics, & policy. I work as a data scientist and am passionate about using technology to solve structural economic problems.