RCP’s Sean Trende performs a public service today by comparing political science models to the standard journalistic treatment of national elections and demonstrating that they don’t differ nearly as much as you might think. To my delight he uses a 1906 take on the 1874 midterms by James Ford Rhodes to illustrate the multi-faceted journalistic take:
“The political revolution from 1872 to 1874 was due to the failure of the Southern policy of the Republican party, to the Credit Mobilier and Sanborn contract scandals, to corrupt and inefficient administration in many departments and to the persistent advocacy of Grant by some close friends and hangers-on for a third presidential term. Some among the opposition were influenced by the President’s backsliding in the cause of civil service reform, and others by the financial question. The depression, following the financial panic of 1873, and the number of men consequently out of employment weighed in the scale against the party in power. In Ohio, the result was affected by the temperance crusade in the early part of the year. … Since Republicans were in the main the instigators of the movement, it alienated from their party a large portion of the German vote.”
This analysis is over 100 years old, but the factors he lists should be pretty familiar to a modern reader. Rhodes mentions the economy, scandals, policy failures, radicalism within the Republican Party, and the incumbency of the president. We might fit the reference to the “Southern policy” under the rubric of “war.” A modern writer would probably also look at polling on presidential job approval, gaffes made by the candidates, the amount raised and spent by the candidates, as well as endorsements by newspapers and other prominent people.
Trende depicts the journalistic formula as:
Popular Vote = Economy + Job Approval + R Gaffes – D Gaffes + R Spending – D Spending + R Endorsements – D Endorsements + War + Policy Failures + Radicalism + Incumbency.
Political scientists, he explains, tend to boil off a lot of these factors as canceling each other out, and thus wind up with models like Alan Abramowitz’s “Time for Change” that focus strictly on the Economy and Presidential Job Approval and Incumbency (a positive after one term but neutral after that).
I would note that Abramowitz himself modified his model in 2012, suggesting that partisan polarization has for some time reduced the degree of “swing” from election to election by about half.
But in any event, concedes Trende, no model has a perfect record, noting Al Gore’s significant underachievement in 2000; according to nearly all poli sci models, Gore should have won in 2000 by a popular vote margin that would have put Florida out of the Supreme Court’s reach. And so he concludes:
There are many explanations for why Gore underperformed, but one plausible read is that the conventional wisdom about the campaigns at the time was correct: Gore ran a really lousy campaign, and George W. Bush ran a great campaign. In this situation, things did not cancel out, and it mattered.
That’s certainly the way I felt at the time as I watched the Gore campaign lurch from outsider “People Versus the Powerful” messages to credit-taking for the Clinton-Gore administration’s economic record. But perhaps other factors didn’t cancel out, either. And perhaps they won’t in 2016.
Certainly Democrats think GOP “Radicalism” could put a thumb on the scales in their favor. I tend to think it will have the slightly different effect of making 2016 less a referendum on the Obama administration than a straight-up “two futures” election where the GOP agenda matters as much as retroactive judgments on Obama’s record. Any way you look at it, though, this should be a close election where any number of variables could wind up mattering. And if this or that model misses the outcome by a crucial percentage point or so, you can expect the political science community to more or less shrug and say, “another election, another sample.”