Behind Realignment--House/Presidential Shifts Over Time

by: Paul Rosenberg

Sun Nov 01, 2009 at 08:30

Three weekends ago, I wrote a diary, "Reagan Did NOT Win A Realigning Election In 1980".  I wanted to show conclusively that the 1980 election didn't look anything like the commonly-recognized realigning elections, such as 1932 and 1896. In addition to the macro-political event of a political party disintegrating (the Federalists in the 1820s, the Whigs in the 1850s), I identified two metrics to identify a realigning election--one for the House, the other for the Presidency. (The Senate was not directly elected for more than half our history, so it's not suitable to use.)  The idea of both was simple: a realignment ought to show a significant before/after shift--and it ought to show up in both metrics. If there is no such shift, then there is no realignment.  (Unless, of course, the complete collapse of one party makes use of the metric superfluous and/or problematic.)  The comparisons of 1980 to 1896 showed quite clearly what this shift looked like in the House (1896), and what its absence looked like (1980):

 

Of course, simply presenting these in isolation raises the question of how significant the 11+ point shift in House balance for 1896 really is, historically speaking.  And the same can also be said about my presidential metric. This was actually a rather atypical election, since the GOP didn't pick up seats that year--it lost a good number in a counter-swing election after winning two consecutive wave elections.  While realigning electios generally involved two consecutive wave elections paired with a decisive party-changing Presidential election, this is the only time that the Presidential election came after both wave elections.   And yet, a strong GOP majority remained even after this anomalous House election.

For me, this leads toward shifting focus to the longer patterns of political ebb and flow, a perspective from which the nature of specific biennial and quadrennial elections is more easily understood in comparison to others.  Instead of cherry-picking whatever particular fact pops out of us to prove or disprove a favored or disfavored argument, we look to find a consistent guide for analyzing all in a common framework.

The Presidential Metric Refined

In my earlier diary, I first explained that the collapse of one of the major parties marked the end of a party system, with no need to rely on more technical metrics. (Indeed, such metrics might be misleading, since a good metric would depend on continuous data, while party disintegration could make such data rather anomalous.) Regarding the Presidential metric, I wrote:

For the presidency, we take the last three elections before the potential realigning election and compare them to the three elections after the re-election of the President elected in the potential realigning election.  We ask how many presidential elections were won by candidates of the same party--before and after....

Using this metric we get the following:

President/        Same Party President in
Election          3 Prior Elections   Next 3 Elections
                                    After Re-election
Jefferson 1800	           0                  3
Lincoln 1860               0                  3
McKinnley 1896             1                  3
FDR 1932                   0                  3
Nixon 1968                 1                  2
Reagan 1980                2                  1

All the realigning elections have 3 presidents of the same party elected after the first president's re-election.  If we included Jackson, however (for whom the first total is ill-defined), we would have one realigning election with only two presidents of the same party elected afterwards--no better than Nixon.

If we use the above metric, and apply it to all 2-term Presidents in historical order (aside from Washington, who was elected in the pre-party era), we come up with the following chart:

Note: One could very well argue that Monroe and Jackson should only be counted as half in the same party, given that both Jackson and Adams were Monroe's party mates when he was elected (and, indeed, Adams was more highly placed.)  That would make the figure for "before two elections" 1 (counting "1/2" twice) rather than 2--making Jackson's tally the same as McKinley's. I actually think this would be more accurate, but it makes my schema neater, so to avoid the appearance of being arbitrary and self-serving, I go with 2 instead.

The "strong" realignments are characterized by two things: (1) No president of the same party was elected in the 3 elections prior to the realigning election. (2) Three presidents of the same party were elected in the 3 elections after the reelection of the president elected in the realigning election.

(In addition, though not in the chart, the strong religniments represent a shift from a party system dominated by one party to a party system dominated by the other--except for Jefferson (1800), which represented a sort of mongrel shift from the pre-partisan era (Washington) and a shift within the first party system, from the Federalist's lone victory (1796) to the Democratic-Republican's complete dominance thereafter.   However, this presumes the structure of party systems.  That's fine with me, because I actually consider them a more suitable basic unit of analyis.  But here I'm trying to show how a reasonable set of metrics can be used to "build up" the notion of realignments and party systems from data from small sets of elections.)

The "weak" realignments are characterized by two  things: (1) One or two presidents of the same party were elected in the 3 elections prior to the realigning election. (2) Two presidents of the same party were elected in the 3 elections after the reelection of the president elected in the realigning election.

(They also represent a shift from one party system dominated by a given party to another party system dominated by the same party.)

The "dealignment" is similar to a weak realignment.  However, it is characterized by a shift from a party system dominated by one party to a system dominated by neither party. The House metric will provide a proxy for this distinction, without assuming the entire party system structure in advance.

All other first elections of two-term presidents are neither realigning or dealigning elections.  

There is just one problem with this analysis: Grant. According to the above criteria his first election should qualify as a weak realignment.  But this clearly defies common sense. No president who's simply continuing his party's control of the presidency can be considered to have won a realigning election, any more than a 1-term president could.  Consequently, we add one more requirement--that for a realignment, the previous election had to be won by a president of the other party.  This added condition also serves to further distinguish Madison and Monroe, as it should.  The resulting table now looks like this:

And resorting it by strength of realignment, we have the following:

House Patterns

I had originally planned on discussing the wave patterns of House balances, but changed my mind to focus more narrowly on the intersection of House balances and the presidential elections already identified, supplemented only by the peaks and troughs of 6-election House cycles, which I'll explain shortly.  First the reasoning, then the definition. The most basic results of any House election as a whole is the balance--how many seats each party has--and the shift from the previous balance.  Because elections can swing back and forth a lot (they don't nowadays so much, but historically, they can), if you really want a good feel for a political era, it makes sense to take multi-election averages.  And so that's what I've done--take averages over 2, 4 and 6 house elections--and then take the shifts from one cycle to the next--one 2-cycle average to the next one (1980 & 1982 compared to 1984 and 1986, for example), one 4-cycle average to the next (1980-1986 compared to 1988-1994), and one 6-cycle to the next (1980-1990 compared to 1992-2002).  The longest one is the one I'm most interested in, but the shorter ones provide a feel for how sharply things are changing.

The following table shows that the strong realignments all correspond with 6-cycle shifts of 20 or more (absolute value), and 4-cycle shifts over 20 as well, always swinging toward the party of the president elected--25.8% more Democratic for Jefferson, 20.1% more Republican for Lincoln, and 24.0% more Democratic for FDR:

The weak realignments are both problematic, but for different reasons.  McKinley's 6-cycle shift is barely half what the strong realignments are, while Jackson's is healthier, but in the wrong direction!  There are simple explanations for both anomalies.  Before Jackson's election the opposition Whig Party had dissolved, leaving the Democratic-Republicans unchallenged, and fighting amongst themselves.  It's really impossible to clearly assign party loyalties in a consistent way through the elections of 1824-1828, so the figures I use are arbitrary in a sense, although they have a logic to them.  It's just that the logic shifts, because the old frames of reference no longer hold.  But whatever metric one uses, the fact that the Whigs left a power vacuum, which the Democratic-Republicans briefly filled,  means that even a very popular president like Jackson would preside over a decline in ruling party power.  The situation with McKinley is even simpler: the Democrats lost badly in two consecutive elections, 1892 and 1894--so badly that the GOP picked up a lot of seats it could not hold in 1896, even though McKinley won the election handily.  So the 6-cycle difference reached its peak in 1894--a peak just under 20 points in the right direction.  Of the rest of the two-term presidents, Grant and Clinton are both notable for relatively high values in the opposite direction.  Nixon, who won a dealigning election shows the smallest shift of all.

We can get a clearer picture of the ebb and flow of party power balance in the House by charting the years that Democratic Party power hit its highs and lows:

And by combining the two charts we can see how closely the realigning elections converged with those highs and lows:

Note in particular the gray cells at the bottom of the chart.  These are projections based on assuming that the Democrats record of gains and losses in the years ahead are the mirror image of the Republicans following 1994. If the GOP does pull of a wave election in 2010, then the 2000 6-cycle would decrease, and the other two might also.  But if the Democrats weather 2010 with minor loses, or even gains, their fortunes could pick up in 2012, and increase the 6-cycle figure for 2006.  It's almost impossible for it to reach the "strong realignment" range. But there is also no guarantee that future years might yield even higher 6-cycle averages, especially if the GOP really does fragment even further. It is simply too early to say.  What we can say is that House swings have become quite muted by historical standards in the past four decades, and so we should not expect further dramatic swing elections--but if they do occur, they could herald a truly dramatic turn in American political history.