Midterm elections are important harbingers of political futures, except when they are not. Parties in control of the presidency lose big in midterm elections, especially in the second term — except when they do not. Midterm elections almost tempt observers to wonder whether politics can indeed be reduced to a science.
TBR Research presents insights and excerpts from peer-reviewed scholarship.
As the United States gears up for the midterm election of President Barack Obama’s second term, we can gain perspective from a hundred years or so of midterm elections. Abraham Lincoln famously argued for re-election in 1864 with a plea that the country should not “change horses in midstream.” Midterm elections are opportunities to harness presidential horses in midstream. The party that does not control the White House usually does well in midterm elections, gaining, since 1894, an average of approximately 35 seats in the House of Representatives and four Senate seats, according to Andrew Busch’s Horses in Midstream, a fine study of midterm elections since 1894. For another, cartoon-filled take on the midterm penalty, see Shallat’s “Changing Horses, by Rule.”
Let us define some terms. Midterm elections take place in the second year of a president’s term—midway between the president’s election and his effort for re-election or the end of his presidency. All members of the House of Representatives are up for re-election during mid-term elections as well as one-third of Senators. Thirty-eight states have also now aligned their gubernatorial elections with the midterm, a decisive change since 1940. Before 1940 almost all states had their gubernatorial elections during presidential years. States changed because they did not want choices for governor to be overly affected by the national issues that often dominate the public debate during presidential years.
Presidents have a more difficult time after midterms than before them, more so in the House than the Senate. The president’s party has gained seats in the House during three midterm elections (nine in 1934, five in 1998 and eight in 2002), while the president’s party has gained seats in the Senate nearly one-third of the time (nine out of 30).
Partly, this is the luck of the draw, as only a portion of the Senate is up for re-election each national election and some Senate seats are more vulnerable than others. Partly it is the effect of our former method of selecting senators, when state legislatures would choose the upper chamber (four out of the nine times when the president’s party gained seats in the Senate came before the passage of the 17th Amendment in 1913, which provided for the popular election of senators).
There is also a theory, not without its advocates, that midterm elections have some sort of predictive value when it comes to the succeeding presidential election.
“Wave elections” (which I tentatively define as elections where one party gains more than 40 seats in the House with a simultaneous move toward that party in the Senate, or where smaller changes lead to a change in party control in a chamber) in the midterm would, on this theory, spell trouble for the side that is shellacked in the midterm election. Yet this theory is true only in a highly qualified way, depending on the type of midterm seen.
Busch helpfully places midterm elections into four categories: vanguards of realigning changes, vanguards of presidential change, calibrating elections and catch-all—“dogs that did not bark”—elections.
Examples of “vanguards of realigning changes” elections include the decisive Republican victory of 1894, in which the GOP gained 116 seats in the House and five in the Senate, and tested out issues such as the gold standard and protective tariff; and the Democratic sweep of 1930, in which Democrats began a commitment to a larger federal role in relief and public works that would continue through the New Deal.
Midterms that are “vanguards of presidential change” do not foster such a decisive change in government policies or long-term trends of election results, but they are midterms that prepare the way for a change in presidential party in the next election. Presidential control has changed 12 times since 1894 (1896, 1912, 1920, 1932, 1952, 1960, 1968, 1976, 1980, 1992, 2000, 2008) and in almost every case (1950, 1976, 1990, 1998 being the exceptions) the party that would take over the presidency won a “wave election” in the previous midterm.
Eisenhower’s Republicans lost pretty big in 1958, preparing the way of John F. Kennedy’s election, but JFK was not the great policy innovator that Franklin D. Roosevelt was. Nor did Richard Nixon’s victory in the wake of the Republicans’ excellent midterm election of 1966 presage much of a change in governmental philosophy.
“Calibrating” midterms are victories that check a president’s course of action, slow down the pace of change or lessen the depth of innovation. These need not be large losses for a president’s party and they do not presage a defeat for the president’s party in the subsequent election. Examples of such elections include Roosevelt’s Democratic defeats of 1938 and 1942; Ronald Reagan’s Republican losses in 1982 and 1986; and decisive Republican victories in 1946, 1994 and 2010. Sometimes such victories for the out-party are large (1946 and 2010); sometimes nominal or relatively minor (1982).
Then there are the elections where the “dog doesn’t bark or,” that defy these patterns. Democrats did great in the 1934 elections, allowing for a deepening and extension of the New Deal. The Democratic loss in 1962 was not too bad and it brought in a decidedly more liberal Congress, though party composition did not change overmuch. Presidential parties did well in 1998 and 2002, mostly due, as was noted at the time, to contingent circumstances of passing significance.
The great difficulty for students of American politics lies in distinguishing, during the stream of events, these kinds of elections. How could we know, for example, in 1994 or 1995 that the 1994 election would be a calibrating election (checking President Bill Clinton’s perceived liberalism) rather than a vanguard election? Do we even know today whether the Democratic wave of 2006 was a vanguard of realignment or a mere vanguard of presidential change (or even what the Republican wave of 2010 was)? The 2010 midterm, like 1946, seemed to be a vanguard election, but, like 1946 it ended up as a calibrating election, preventing further progressive legislation from passing Congress, but not preventing Obama’s re-election.
Explaining the past — even the immediate past — proves as difficult as predicting the future. So much depends on how statesmen and politicians react to losses or gains — reactions both in policy and in rhetoric. Much also depends on the success or failure of policy innovations made during a president’s pre-midterm years.
All of this is a short way of saying that I, as a political scientist, have no idea what the results of the 2014 elections will be, nor do I know whether those results will prove important for the long term. Our very best pundits seem to think that Republicans are poised for a four- to eight-seat pick-up in the Senate and a single-digit pick-up in the House.
What really matters, however, is what each of the parties does with the new configuration — and how the electorate greets those actions. This “known-unknown” is what makes politics fascinating and unscientific.
The views and opinions expressed here are those of the writer and do not necessarily reflect those of Boise State University or the School of Public Service.