As midterm election results roll in Tuesday night, University of Virginia Center for Politics Director Larry Sabato predicts Democrats will regain control of the House with a “very slight edge,” while Republicans will likely retain a narrow majority in the Senate.

He’ll be playing close attention to gubernatorial elections, particularly in the Midwest, which he calls “a bright spot” for Democrats.

And he’ll be looking to New Jersey, where a tight Senate race and several House races could serve as a bellwether, indicating how the night will go for both parties.

We sat down with Sabato and Kyle Kondik, managing editor of the Sabato’s Crystal Ball political analysis and election-handicapping newsletter, to talk through Tuesday’s elections and their impact on the Senate, the House and governor’s offices.

All 435 House seats are up for election, along with 35 seats in the Senate. Republicans are hoping to retain their majority in both chambers, while Democrats are hoping to retake control by gaining at least two seats in the Senate and 23 seats in the House.

Here’s what Sabato and Kondik had to say.

Q. The 2016 presidential election defied pre-election polls across the country. How did that election affect your polling and analysis of the midterms?

Sabato: It certainly did change our approach, and I will let Kyle describe several things we have done differently.

Kondik: When the 2016 general election began, there was a tendency nationally to position Clinton as the favorite, with many not considering several historical factors – a party seeking a third straight term in the White House, the state of the economy, etc. – that could have pointed to an upset. We have been very cognizant of the history of midterm elections, which strongly favors the party out of power, while also paying attention to factors like the strong economy, generally a good thing for the party in power.

In terms of polling, we have partnered with Ipsos and Reuters to bring in several new sources of information, such as social media analysis and very comprehensive state-level polling. We have used that to create a new 2018 Political Atlas that offers an interactive breakdown of races in each state.

Q. What does your analysis indicate about results in the House and the Senate?

Kondik: This election has developed decently for Democrats, but it is not overwhelmingly in their favor. I don’t think we are looking at the same level of sea change we saw in 2010, where Republicans retook more than 60 seats in the House. At this point, some of the models we rely on suggest a close national result with Democrats favored to win the House, but not overwhelmingly. Our own seat-by-seat analysis suggests a Democratic pickup of somewhere in the neighborhood of 30 seats.

Sabato: On the Senate side, we have been very consistent in our projections for about a year now, projecting that Republicans will hold the Senate because of the territory involved. The third of the Senate that is up for election disproportionately includes smaller, more rural states that typically go Republican and features several at-risk Democrats in states that have heavily favored Trump. Those include Democratic senators like Claire McCaskill in Missouri, Joe Donnelly in Indiana, Jon Tester in Montana, Heidi Heitkamp in North Dakota and Joe Manchin in West Virginia. Some of those will survive, but they are certainly being hard-pressed.

Q. What gubernatorial races will you be paying close attention to?

Sabato: The headline for Democrats, and likely the brightest part of the night for them, will be governorships in the Midwest. Michigan will almost certainly switch parties, with Democratic candidate Gretchen Whitmer defeating Republican nominee Bill Schuette. Illinois and Iowa are likely to switch as well, and two tight races in Ohio and Wisconsin could flip for Democrats.

Gubernatorial races in Florida and Georgia, with African-American candidates Stacey Abrams and Andrew Gillum, have gotten a lot of national press. I will be surprised if at least one of those candidates does not win. Gillum currently leads in almost all polls in Florida. It’s a bit closer in Georgia, where we might have to wait until a runoff on Dec. 4.

Kondik: Gillum is a notably more liberal gubernatorial nominee than Democrats would typically put up in Florida, so it will be interesting to see how he does. He could present a different kind of model for Democrats in that state.

Q. Are there any other bellwether races that could indicate how the night will go for both parties?

Kondik: There are competitive House races all over the country, but California is one key state to watch, which could make for a late night for viewers on the East Coast. Democrats need to pick up three to five seats in California to have a good shot at retaking the House.

Virginia could also tell us a lot. A win for Republican Rep. Barbara Comstock in Northern Virginia would be a good sign for Republicans, while Democratic wins in areas around Richmond and Charlottesville would be a positive sign for Democrats

Sabato: Early in the night, you’ll also want to take a look at New Jersey. It’s a blue state in a blue year, so you would think it would be an easy win for Democrats. However, Democratic Sen. Bob Menendez has had problems with corruption charges, and he is being challenged by Republican nominee Bob Hugin, who, incidentally, is a graduate of UVA’s Darden School of Business. It has been a remarkably close race, and it would not be a tremendous shock if Menendez was upset.

Kondik: At the same time, in New Jersey, there are four House seats that Democrats could credibly win. Watch how those seats go, because that could indicate how the House will go.

Q. Do you expect voter turnout to diverge from past averages?

Kondik: Turnout in 2014 was very low, with about 36 percent of the eligible voting population showing up, the lowest since 1942, when there was a war on. We expect this election to be around 40 percent or higher, a bit higher than average for midterm elections.

It’s important to note that high turnout does not necessarily favor one party or the other. Often, the assumption is that it favors Democrats, because their voter base is younger, more diverse and generally less reliable in the midterms. However, we will likely see irregular midterm voters showing up for both parties because there is a lot of interest in politics right now.

Q. The Center for Politics has a long history of getting young people interested in politics. How could youth voter turnout affect these elections?

Sabato: We believe youth voter turnout will be up, among those aged 18 to 29 or even 18 to 24. It will certainly increase relative to the last midterm, which saw spectacularly low youth turnout.

We have been working on increasing youth participation for 20 years, and we have made progress in some areas, particularly among college-enrolled youth, who are much more likely to vote than their working counterparts.

Higher youth turnout will not necessarily make youth a larger percentage of the electorate, because we believe turnout will be higher among every other age cohort as well.

The increases, quite simply, are due to Donald Trump. Voters are either for him or against him, and they are passionate either way. He is the center of attention for both parties.

Q. In general, what trends have characterized this election season?

Sabato: The level of interest is extraordinary. Both parties, particularly Democrats, believe the stakes are incredibly high. Democrats are still furious about losing in 2016, and there have been so many events – things in tweets, rallies and policies – that have stoked outrage since then. Democrats feel they have something to prove and to this point they seem determined to prove it. We’ll see if that sentiment is reflected in the turnout and the results.

Kondik: It’s also interesting to note the rise of the potential next generation of prominent politicians. Democrats in particular really had their “bench” knocked out during the Obama-era midterm elections. I would assume that several young candidates elected in both parties this year will become prominent on a national level.

Q. Does your analysis indicate any changes that could play a role in the 2020 presidential election?

Sabato: Right after the election, everyone will start claiming how the results will affect 2020. The real answer is: surprisingly little.

I’ll give two examples. I can still remember the headlines after the 1994 midterms, when Clinton and the Democrats were waxed. Bill Clinton was nick-named OTB, or “One-Term Bill” and there was consensus in both parties that he could not possibly win a second term. Two years later, of course, he won a second term.

We saw the same thing with Barack Obama, when Democrats lost badly in the 2010 midterms and then won reelection in 2012.

I am not saying that Donald Trump will be reelected, but I am saying that, because so much will happen in two years and because there are so many unknowns, speculation is pointless right now.