Sunday, January 9, 2011

American Politics: Electoral Realignment or a Model of Stability?

As part of a larger research project, I've been working on updates to Gerald Pomper's 1968 study of electoral coalitions by studying the linear correlation of state by state election results in paired presidential elections. According to Pomper an electoral conversion or realignment would be evident via a “change in the parties’ bases of support… the geographic distribution of each party’s vote would be different from the past: traditional strongholds would fall, while new areas of strength would become evident.” Such changes or shifts in support would be evident in statistical analyses of the correlation between sequential elections. High levels of correlation would indicate stability, and low levels change.

Though some have openly wondered whether the 2008 election of Barack Obama represented a political realignment, I contend that the political realignment of the late 1960s has been maintained, and is in fact incredibly stable.

Using election data for all states and DC since 1928 I have calculated linear correlations for all successive elections in the last 8 decades.  As shown in the first graph below, the correlations reveal the very stable New Deal realignment as well as it's collapse in the 1960s and the emergence of a new and stable electoral pattern - a realignment. The only disruption of that new stability occurred in 1976 with the election of Jimmy Carter, likely a reflection of fallout from Richard Nixon and Carter's strength in the south, in the midst of the southern realignment toward the Republicans.

What is unmistakable is that 2008 fit comfortably into the electoral pattern that emerged at the end of the 1960s - 2008 was very much a normal election, its correlation with 2004 and the average of 1992-2004 were both quite high. This suggests that no realignment, defined by James Campbell as "a durable and substantial shift in the parties’ national electoral balance of power,” took place in 2008.

I wanted to be certain that I was not missing more subtle evidence of realignment by focusing too much on national trends, so I next divided the states into three groups - Reliably Democratic states, Reliably Republican, and Swing states. Reliably Republican or Democratic states were defined as having voted with their respective party's candidate in at least four of the last five elections, through 2004. The designations were based on an analysis by the Population Reference Bureau conducted prior to the 2008 election.

This resulted in 39 states (including DC) that were not considered to be competitive. The remaining 12 states were coded as swings.

As shown in the following graph, I found a high degree on stability in the Democratic states (light green line)through 2008, lesser but still high stability in the Republican states (red line) and in the swing states (blue line) I found high levels of correlation from one election to the next only when a president was re-elected, or succeeded in office by a member of the same party. Conversely, elections that witnessed the defeat of an incumbent or a shift in party control of the White House was marked by lower correlations - until 2008. Swings state results were highly correlated with 2004.

The swing state finding for 2008 was intriguing given that George W. Bush won 11 of the 12 swing states in 2004 and Barack Obama only 5 in 2008.  I suspect the high degree of correlation reflects the closeness of the results across those swing states in both elections - even though the partisan outcome shifted in nearly half.

It is also possible that the high degree of correlation indicates some changes taking place in swing states. Six swing states (AR, KY, LA, MO, TN and WV) have not gone Democratic since 1996, suggesting that they may no longer be swing states.

I recoded those states as reliably Republican and it reduced the correlation coefficient for 2008 among swing states from the high .845 to a much lower .416 (no graph). The Republican correlation coefficient in 2008 declined slightly from .917 to .795.

The results appear to suggest that the Southern realignment that began in the late 1960s is complete as 4 of the 6 swing states recoded as Republican are southern states. In all, it suggests that the battlefield of swing states may be decreasing.

It's too soon to tell whether Indiana, North Carolina, or Virginia need to be removed from the reliably Republican column and added to the swing state group.

A recent blog post by Professor Andrew Gelman noted that there has been a decline in state by state vote swings, noting "the red-blue map is much more stable from election to election than it used to be." My analysis suggests that much of this stability results from the culmination of the southern realignment, the Democratic electoral coalition has been very stable since 1968 and the hiccups in the Republican coalition have involved southern states in elections with southern Democratic candidates. In recent elections, southern states once considered swing states (because they still voted for southern Democratic presidential candidates) seemed to have completed their journey to the Republican column - witness the failure of southern Democrat Al Gore to carry a single southern state in 2000.

The overall results from my analyses thus far do support my contention that a relatively stable electoral pattern emerged in the late 1960s and there is little evidence of any break in that stability. America remains in the post New Deal electoral system. A highly competitive system in which Republicans and Democrats are near parity and neither party dominates the political scene. Though Republicans have won 7 of the 11 Presidential contests between 1968 and 2008, Democrats have have enjoyed majority control in the House of Representatives in 15 of 22 Congresses between 1968 and 2010. And unified party control of the House and the White House has only existed for 5 of 22 Congresses.

Everett Carll Ladd labeled the post New Deal era the "No Majority Realignment." It's a fitting and effective description of an era that emerged in the late 1960s, and it appears, remained intact through 2010.