Over the past 40 years, political polarization among elected representatives in Congress has steadily increased. This has presumably been related to the contemporary decline in electoral competitiveness: there are far fewer swing districts in Congress than there were just 20 years ago.
At the national level, it appears that gerrymandering, oft-cited as a cause of increasing polarization and decreasing electoral competitiveness, has in fact had a very small effect. Instead, the main cause is that the national political parties have moved further apart in terms of how they would represent the same district — perhaps the result of the increasing nationalization of politics. Another long-term effect has been the passing of the final vestiges of the conservative southern Democrats (and the weaker echo among liberal northeastern Republicans).
In the past few years, citizens of California have approved two ballot measures intended to address these issues of polarization and competitiveness. In 2008, in an attempt to remove power politics from the redistricting process, Proposition 11 was approved, which created a Citizens Redistricting Commission. The commission consists of 5 registered Democrats, 5 Republicans, and 4 independents, selected from 5,000 applicants to the positions. In 2010, Proposition 20 gave the commission control over U.S. Congressional districts as well. (Democrats apparently used underhanded means to attempt to influence the commission, though it’s not clear how effective they were. Of course, to the extent that gerrymandering doesn’t matter — see above — neither do these shenanigans.)
Then, in 2010, Proposition 14 was passed, which replaced traditional party primaries in CA with “jungle” or “blanket” primaries. Under this scheme, the primaries (typically held in June, before the November elections) would be open to all candidates from all parties. The top two vote-getters in the primary would then square off in the general election, regardless of their party. This could and did result in November runoffs between (for instance) two Democratic candidates, with no Republicans on the ballot. This reform is intended to ensure that candidates in the primaries appeal to a broad swath of the electorate, rather than focusing solely on their political base, liberal or conservative.
Effects on Electoral Competitiveness
I have no way to directly compare political polarization between the 2008 and 2012 elections. At the U.S. level, one could use a legislator’s DW-NOMINATE score to diagnose how partisan — liberal or conservative — their votes were. However, this data isn’t available for state legislators in CA. Furthermore, because the newly-elected 2012 candidates have only been in office for a month, they haven’t had time to establish the voting record required by systems like DW-NOMINATE.
Instead, I consider electoral competitiveness, which I define as the spread between the winning candidate and the runner-up, normalized by total votes given to the top two candidates. Furthermore, I only consider elections for the CA Assembly and the U.S. House of Representatives, in order to make reasonably direct comparisons between districts in 2008 and 2012.
(The assumption of a link between electoral competitiveness and political polarization may not be exactly right, especially in politically balanced districts. But according to the story in the paper just linked, polarization in competitive districts is caused by intense activism by extremists in primaries — which may be attenuated by Proposition 14. Regardless, CA has recently had so few electorally competitive districts that this story is not likely to be explanatory in our case.)
There were 80 Assembly districts and 53 House districts in California in both 2008 and 2012. I define “close” and “medium” as elections where the spread between the top two candidates was less than 10 and between 10 and 25 percentage points, respectively. The figure below shows how electoral competitiveness changed between 2008 and 2012. The most striking effect is the decline in “wide” and uncontested elections, which were mostly replaced by “medium” elections (along with a substantial increase in close elections for the House). Congress and the Assembly had 4 and 7 completely uncontested elections in 2008, while in 2012 they had 0 and 2 respectively.
Finally, the average spread for Congressional elections decreased from 42 to 17 percentage points, while it decreased from 37 to 26 percentage points in Assembly elections. (Removing uncontested elections, the decreases are instead respectively 37 to 17 and 31 to 24 in the cases above.) Competitiveness appears to have increased across the board. These increases in competitiveness are largely consistent with those found in a much more in-depth report by the Public Policy Institute of California.
Some of this may be a one-time effect. If two incumbent legislators were thrown into competing for the same district in 2012, the election would be competitive without necessarily implying that future elections for that same district would be similarly competitive, since in future years only one legislator would have the incumbency benefit.
One might hope to track individual districts, and try to assess why they became more competitive: as a result of the top-two primary format, redistricting, a combination thereof, or some other reason? Unfortunately, the new districts are in many cases not geographically close to their former location. For instance, the 31st Congressional district was previously located in downtown LA, including Hollywood and many Hispanic-majority areas, and was won by Democrat Xavier Becerra uncontested in 2008. However, in the 2012 redistricting it was moved to encompass San Bernadino and Rancho Cucamonga in the Inland Empire, and Gary Miller won 55%-45% over his fellow Republican Bob Dutton. (Becerra now represents the 34th district, which overlaps significantly with the former 31st.) Thus, one would need to invest substantial effort to identify geographic nearness and movement of incumbents to new districts. Maybe a future post…