Is the Six-Year Itch Just a Senate Thing?

This piece was originally posted in Mischiefs of Faction on January 13, 2014.

Midterm elections are bad news for presidents or their parties, period. No question about it. But are midterms in the sixth year of power—in the middle of a second presidential term—worse news than those in the second year of power? Are the seat losses greater? The term “six-year itch” seems to imply that distinction.

The answer is yes for Senate elections. But, surprisingly, it seems to be no for House elections.

Here are the specifics. Take all the relevant midterms from 1914 through 2008. Beginning at 1914 makes sense since the size of the House stabilized at that point, and senators started getting directly elected then.

The Senate midterms line up nicely for the special itch. New presidents who lasted through six years have actually gained an average of one seat in their second year, but lost an average of six seats in their sixth year. (The gain of one become a loss of one if Truman is included in the list—about which more below.) Looking at it another way, parties fresh to the presidency have also gained one seat and then lost six. This grouping accommodates all parties that made it through the sixth year whether or not their presidents did. An instance of a same-party sequence is Kennedy and Johnson in the 1960s. All these numbers are arithmetic means.

There is it: a special six-year itch on the Senate side.

But the House is another story. It needs details. Being unconventional, it can use a sense of historical terrain to back it up.

Let’s consider the individual presidents first, then the parties.

New presidents in their second and sixth year. There are seven cases of presidents who lasted through a sixth-year midterm election—Wilson, FDR, Truman (who needs an asterisk), Eisenhower, Reagan, Clinton, and Bush 43. Truman, reaching the White House through FDR’s death, missed the first month of a full eight-year presidency. He is normally thought of as having a first and a second midterm. But he didn’t reach the office through getting elected on his own, and, as it happens, his first midterm was pretty bad, so he should get special treatment in the statistics. Let’s run the averages with and without him.

Here are the average House seat losses.

•28 – first midterm (including Truman)
•29 – second midterm (including Truman)
•24 – first midterm (without Truman)
•29 – second midterm (without Truman)

There it is, you might say. Leaving aside Truman, the contrast of 24 versus 29 shows exactly a special six-year itch. But not so fast! All these numbers are small. They can easily be veered. What about Obama? We don’t have a sixth-year result for Obama, but we certainly do have a second-year result—the Democrats lost 63 House seats in 2010. Why lose this information? Plugging it in moves the above first-midterm figure of 24 to a higher 29. As for the second midterm, does anybody expect a Democratic loss of House seats in November 2014 that exceeds 29? Probably not. When the dust settles next November, it is an excellent bet that adding Obama’s two midterms to the comparison will kill off entirely any idea of a special sixth-year penalty on the House side. Also, note that five of the relevant presidents have lost more seats, or in Obama’s case will have done so, in their first midterm than in their second—Wilson, the asterisked Truman, Reagan, Clinton, and Obama. Only three have lost more in their second midterm—FDR, Eisenhower, and Bush 43. How about losing party control of the House? Truman, Eisenhower, Clinton, and Obama suffered that fate in their first midterm. Wilson and Bush 43 suffered it in their second.

But perhaps we should be looking at parties rather than presidents?

Fresh parties in their second and sixth year. This analysis offers a somewhat different dynamic. Voters might get fatigued with parties rather than, or besides, individual presidents. Perhaps it is an incumbent presidential party that falls off especially in the sixth year. That’s what to scout for. A tweaked dataset is in order. There are nine sequences since 1912 (not including Obama) where a new party’s control of the White House lasted through a sixth year. The names of the presidents can serve as labels. The instances are Wilson, Harding/Coolidge, FDR, Eisenhower, Kennedy/Johnson, Nixon/Ford (Nixon didn’t quite make it to a second midterm), Reagan, Clinton, and Bush 43. Truman skirts this list since the Democrats were not fresh in power in 1945. Here are the average House seat losses:

•26 – first midterm
•31 – second midterm

There it is again! A special sixth-year penalty! But again, consider Obama. Why lose the information? Add his election disaster of 2010 and the above first-midterm figure of 26 rises to 30. And November 2014 isn’t likely to increase the second-midterm average of 31. Soon, in this measure also, we will be looking back at a century-long statistical wash.

Put it this way. The Democrats need to lose at least 30 House seats or so next November to stave off a reverse century-long itch pattern—that is, one showing first-midterm losses to be (at least slightly) more ample than second-midterm losses.

What is going on here? Why the itch gap between Senate and House? One good bet for clues is the Senate’s pattern of staggered six-year terms. For any selection of midterms, peer back at the political context six years earlier. Any class of senators up for reelection this time won their seats back then. In this kind of lookback, party logics play out differently for the two brands of midterms under the microscope here. Take for openers the first midterms of the new presidents or parties. In all such cases here (except Truman), the classes of senators up for reelection this time had won their seats last time when presidents of the opposite party were winning the White House. But this time, any presidential coattails that thus favored that opposite party are gone. Bingo, an exposure problem for any old coattail enjoyers. Now, Senate candidates of the newly incumbent presidential party have a decent chance to fare well. The result: on average, on the Senate side, a calm first midterm for a new party in the White House.

On the other hand, a party’s Senate candidates running for election in the sixth year of a White House governing span lose the presidential coattails they enjoyed six years earlier. The result: carnage in the second midterm. In 2014, a good many Democratic senators elected in the lush context of 2008 are understandably jittery.

Other things were going on in all these cases, and of course any average does accommodate cases that bounce around. But the general argument is intuitive as well as familiar.

On the House side, the calculus is different. No such props for a special six-year itch. This reasoning about the itch gap is at least plausible speculation. Bernard Grofman and Robert Erikson have pursued logics like this that dwell on resonances from previous elections.

All the above analysis, let it be said, sidesteps the question of whether presidents or parties end up worse off in total seat holdings just after a second midterm, as opposed to a just after a first midterm. The answer is: yes, on average they do—in both House and Senate. Here are the figures for the nine same-party spans (leaving out Truman) through 2008. On average, the parties have enjoyed five fewer Senate seats, and 22 fewer House seats, in their fourth Congresses just after a second midterm, as opposed to their second Congresses just after a first midterm. That is about 5% of the total membership of each body. Seats lost in a first midterm are hard to win back. And then comes a second midterm. It is a picture of attrition. Presidents don’t have an easy time of it in their fourth Congresses. But the incidence of “itch” is a separate question.

David Mayhew is Sterling Professor of Political Science and a Resident Faculty Fellow at ISPS.