How to make election predictions in sea of American political discontent

How to make election predictions in sea of American political discontent
© Getty Images

Listen to the news these days and one will quickly realize the 2018 election is about many things. It is a referendum on the president, it is the year of the woman, it is Donald TrumpDonald John TrumpA better VA, with mental health services, is essential for America's veterans Pelosi, Nadler tangle on impeachment, contempt vote Trump arrives in Japan to kick off 4-day state visit MORE taking over the Republican establishment, it is the Supreme Court nomination, it is the resistance. In truth, 2018 is all these things and more. When close to 100 million people go out to vote, there is plenty of space for different storylines.

It is tempting to emphasize one single factor to predict an election, particularly if it is what someone happens to be an expert on. However, the best professional forecasters understand that no single approach leads to knowing the future. Good prediction uses independent indicators and looks for common conclusions or divergent directions.

Accurately being able to handicap elections calls for willingness to consider information that challenges preconceptions. A partnership we are forming between the deep knowledge of Crystal Ball at the University of Virginia Center for Politics and the opinion research of Ipsos Public Affairs is one such effort. Our new Political Atlas combines different indicators to look at all the races for the House, Senate and governors.


So what do the data tell us? First of all, a historical note. It is normal in American politics for the party outside the White House to pick up House seats in midterm elections, and this year looks to fit that broader pattern. The party of the president in office has lost ground in the House in 36 of 39 midterms since the Civil War, with a loss of 33 seats on average. Democrats need to net 23 seats to win the House this year, which is hardly guaranteed but also would not be an ahistorical outcome.

Public opinion research tells us that the 2018 election will be marked by the same discontent and desire for change that helped propel Trump to the White House. Despite a strong economy and no major foreign wars, which two factors that when present can hurt the incumbent party, the president is unpopular and, while his popularity is not really dropping, his approval rating is stuck roughly in the low to middle 40s.

A Democratic lead on House generic ballot polling suggests that many seats held by Republicans should be in play this fall, an assessment that is confirmed by a seat by seat analysis based on both polling models as well as other indicators, such as recent special election results, fundraising, district level polling, and a high number of open seats controlled by Republicans. Social media indicators also point to an enthusiasm advantage for the Democrats, as do surveys that ask members of each party about their level of excitement about the election.

Put it all together, and the Democrats appear to be favored in the race for the House. However, there is still enough uncertainty that Republicans could hold on to the majority, albeit with a reduced size. The Senate is a different story because the roster of seats in play is so unfavorable to Democrats, who are defending 26 of the 35 seats this year. Trump carried five of these states by 18 points or more. To put that in context, only two House Democrats represent districts that Trump won by such a margin. Democrats benefit from the usual midterm trend and the specifics of this election cycle in the Senate, too, but likely not in a way in which they could pick up the two seats they need to win the majority.

We would be remiss if we did not mention some contradicting indicators. Fundamentals models that rely more heavily on economic conditions and incumbency point to Republicans having stronger chances. These same models proved to be more accurate at predicting the ultimate success of Trump in 2016 and should be considered. However, given the pronounced level of public dissatisfaction, as the president is unpopular and fewer than one in three Americans believes the country is on the right track, these structural models should be somewhat discounted.

This multidisciplinary approach to election forecasting will be central to our efforts through the rest of midterms. However, it is important to remember that circumstances change and forecasts need to change along with it. The Political Atlas will give everyone the ability to triangulate different data sources and reach their own conclusions.

Larry Sabato is author of the Crystal Ball newsletter and director of the University of Virginia Center for Politics. Clifford Young is president of Ipsos Public Affairs. They publish the Political Atlas election model.