ANALYSIS: All bets are off in midterms as turnout surges

As American voters head to the polls, who has the edge in the battle for Congress - and just how much faith can we put in polls?

David Rosenberg ,

Trump addresses supporters during a Make America Great Again rally in Tennessee
Trump addresses supporters during a Make America Great Again rally in Tennessee

While the polls are set to open in the US Tuesday morning, well over 38 million voters have already cast their ballots for this year’s midterm election, leading to predictions that this year’s turnout could be the highest for a midterm in over a century.

All 435 seats of the House of Representatives are up for election Tuesday, along with 35 of the Senate’s 100 seats, 36 state governorships, and 87 of the 99 state legislative bodies across the 50 states.

The GOP currently has a majority in both chambers of Congress (51 of 100 seats in the Senate, 241 of 435 in the House), as well as a majority of governorships (33 of 50) and state legislatures (66 of 99).

Democrats, who have been in the minority in the Senate since 2014 and in the House since 2010, hope to recapture both chambers Tuesday in a “Blue Wave” fueled by a popular backlash against President Trump in the Democratic base.

Polling for much of the summer seemed to corroborate the Blue Wave theory, showing relatively low voter enthusiasm – a standard predictor of likely turnout – among Republicans and Republican-leaning voters, compared to high enthusiasm amongst traditionally Democratic-leaning demographic groups including younger voters, unmarried women, and Hispanics.

Polling in late September suggested that the election could deliver as many as 40 to 45 net-pickups in the House for Democrats, nearly double the 23 needed to flip the lower chamber.

In the Senate, despite a map highly unfavorable to the Democratic Senate caucus, it appeared that Democrats had a viable path to a 51-seat majority, requiring two net pickups. Surprisingly competitive races in red states like Texas and Tennessee fueled Democratic hopes for a wave election even in the Senate.

Republicans appeared equally vulnerable in gubernatorial races, with predictions of 8 to 10 net pickups for Democrats, including states won by Donald Trump in 2016 like Florida, Michigan, Ohio, Iowa, Wisconsin, Georgia, and even deep-red Kansas.

Following the confirmation hearings for President Trump’s Supreme Court nominee Brett Kavanaugh, however, in late September, the vaunted ‘Blue Wave’ began to appear less of an inevitability, with surges in voter enthusiasm amongst Republicans – particularly Republican-leaning male voters.

This shift has continued to be reflected in the polls, with a rise in President Trump’s net approval rating, as well as improvements for Republicans in gubernatorial and congressional races across the country.

According to the RealClearPolitics average of polls on the eve of the election, Republicans are favored to retain control of the Senate and even pick up several seats, while the Democratic lead in the House has fallen to an average of 26 to 27 pickups, just slightly over the 23 needed to flip the lower chamber. Democrats’ biggest opportunities now appear to be in the 36 governor races, where the RCP average still shows Democrats favored to flip seven to eight governorships.

FiveThirtyEight’s analysis of polling data shows Republicans favorites to hold the Senate with an 81% chance of keeping the chamber, while Democrats are given an 88% chance of flipping the House.

Slickly-produced analyses of polling data did not fare well in 2016, with virtually every model making Hillary Clinton the clear favorite to win. The New York Times gave Clinton an 85% chance of winning on the eve of the election, while CNN’s model put Clinton’s chances at 91%. These models appeared conservative, however, compared to the Huffington Post’s whopper, which showed Clinton a near-lock with a 98% chance of beating Trump.

But there are some indications this year’s midterm elections could be even more difficult to predict than past votes.

While turnout typically is significantly lower during midterm elections than in presidential votes – only 36% of eligible voters cast ballots in 2014 – this year turnout is likely to be high – perhaps record-setting high for the post-war era, with some predictions that it could reach 45% - the highest in 50 years of midterm elections – or potentially even 50%, the highest since 1914 when 51% of voters turned out.

University of Florida professor of political science Michael McDonald, who has been tracking early voting to predict likely voter turnout, noted on the eve of the election Monday night that over 38.1 million votes had already been cast before the general voting even began. By comparison, just 27.5 million early votes were cast in 2014.

With this surge in turnout, many House, Senate and gubernatorial races which appear close in polling could be swamped by an unexpectedly high turnout by either Republicans or Democrats.

Last-minute polling gives no clear indication of who is likely to benefit most from the rise in turnout. John James, the Republican candidate for a US Senate seat in Michigan, suddenly pulled to within the margin of error with incumbent Debbie Stabenow in what was previously believed to be a safe race. According to a Mitchell Research poll released Tuesday morning shows James trailing by just three points – 46 to 49 – after trailing Stabenow by seven points in a poll by the same firm s a day earlier.

A series of polls in Arizona show a similar bump for Republican Martha McSally, though her race against Kyrsten Sinema was already very competitive.

In Florida, on the other hand, incumbent Democratic Senator Bill Nelson has surged in recent polls, after trailing Governor Rick Scott for much of the summer.

A break-down of the early vote on Monday showed a dead-heat in terms of who was casting ballots, with 42% of votes cast by registered Republicans, versus 41% by registered Democrats.

In key states, according to McDonald’s United States Elections Project, early votes show no nation-wide trend, with early voting in some states showing improvements for Republicans over 2016 in terms of turnout, while other states see the makings of an apparent ‘Blue Wave’.

In Arizona, a Republican-leaning state which Trump won by 3.5 points in 2016, registered Republicans lead registered Democrats in early voting 7.1 points. That points to a slight increase over 2016 when Republicans led by just 6.7 points, but a major dip from the 11.5-point lead in 2014. In 2014, Republican Senate candidate Doug Ducey won by a margin closely mirroring that early vote lead – 11.9 points. In 2016, Trump under-performed in comparison to the lead registered Republicans had in early voting – with Trump winning by 3.5 points versus a 6.7 point lead for the GOP in early voting.

In Florida, Governor Rick Scott won by 1.0 point in 2014, when registered Republicans outvoted Democrats in early voting by 2.8 points. But two years later, Trump beat Hillary Clinton in Florida by 1.2 points, despite Democrats outvoting registered Republicans there by 1.5 points. This year, registered Republicans narrowly outvoted registered Democrats, by six-tenths of a point.

While early voting in both states hints at close races, in Iowa, Democrats appear better positioned for pickups. While Democrats narrowly outvoted Republicans in Iowa in 2014 by 1.9 points, this year that lead has grown to 7.9 points.

If early voting and polling have failed to provide grounds for clear predictions regarding the outcome of Tuesday’s election one thing appears clear: both Republicans and Democrats see the vote as a referendum on Trump’s presidency – and both sides appear motivated by that fact to turnout in larger-than-usual numbers, a clear change from 2010 and 2014, when Obama voters disproportionately remained at home leading to outsized Republican wins.