Extreme Referendum: Donald Trump and the 2018 Midterm Elections

LONG-TERM POLITICAL TRENDS and the public’s reaction to Donald Trump’s surreal presidency converged to make the 2018 midterm elections the most sweeping and discordant national referendum on any administration at least since the Great Depression. Midterms have grown increasingly nationalized, partisan, and president centered over recent decades, but Trump’s persona, rhetoric, and policies extended all of these trends into uncharted territory. Trump’s polarizing performance and electoral salience cost Republicans control of the House of Representatives but strengthened their grip on the Senate (Table 1), setting the stage for fierce partisan battles within and between the branches during the 116th Congress. In this article, I place the 2018 referendum in historical context; examine the extraordinary influence Trump exerted on the electoral environment, strategic political choices, and voting decisions; and consider the referendum’s implications for the new Congress and the future of the two parties. In its basic outline, the 2018 midterm referendum replicated past patterns quite faithfully. Since the pioneering work of Gerald Kramer and Edward Tufte, the idea that voters in aggregate treat the midterm elections as a referendum on the administration, rewarding or punishing its party’s congressional candidates for its perceived successes or failures, has been well documented and widely accepted by the research community.1 In typical referendum models, the president’s current approval ratings and economic conditions are the prime movers, although the number of House seats held by the president’s party also matters (all else being equal, the more seats it holds, the more it has to lose). Recent iterations of a basic model in which the independent variables predicting midterm seat swings are the percentage approving of the president’s performance in the Gallup Poll closest to Election Day, the change in real disposable income per capita over the election year, and the deficit or surplus of House seats currently held by the president’s party compared with its eight-election moving average (its “exposure”) have been quite accurate.2 Updated and applied to 2018, with Trump’s approval in the last Gallup Poll before the election at 40 percent and real income growth at 2.1 percent, this model predicted that Republicans would end up with 41 fewer House seats than they held after the 2016 election, 40 seats fewer than they held on Election Day—improbably, the precise outcome in 2018 (Table 1).3