As I have done every two years for  quite some time, here are my predictions for the midterms.

House of Representatives

There are several dozen House races and five Senate races where one candidate or the other has a lead of 2 points or less in the most recent survey.  Picking individual winners in these House races is hazardous since there are few public polls of these districts .  Another few dozen House seats show a candidate leading by five points or less.  The great majority of the close House races are Republican-held seats, hence their vulnerability to Democrats taking over the House, where a net pickup of 23 seats is required.

Most every analyst -- Larry Sabato, Stuart Rothenberg, the Cook Political Report, the Real Clear Politics average, Nate Silver’s fivethirtyeight.com, Nate Cohn in the upshot in the New York Times --  is predicting a pickup for the Democrats of more than the 23 seats needed.  Several have increased their range of the likely pickup in the last week to the 30 to 40 seat range.

Polling is becoming more difficult with ever lower response rates to telephone surveys with live callers.  .  There are sometimes higher response rates to automated surveys, whether by phone or online (though not much higher), and there are issues related to whether such surveys reach a representative sample of likely voters. 

Rasmussen, an automated survey,  shows Trump with a positive approval score, and most other surveys show him 8-12 or more points net negative.  The generic ballot for which party voters prefer control Congress has consistently shown large margins for the Democrats, currently over 7 points on average, though a few surveys have shown smaller margins this week. 

The Democrats seem poised for significant gains in the House delegation in Pennsylvania , due to court-ordered redistricting.  Democrats should also net big gains in California, New Jersey, and possibly Florida and Virginia, where a weak (or in California, nonexistent) GOP Senate candidate could drag down GOP House candidates.  There are many other states where there are 1 or 2 endangered GOP held seats.

RCP has placed 15 GOP-held House seats in the lean or likely Democratic column, and two Democratic held seats in the lean or likely Republican column.  That accounts for a net 13 of the 23 needed.  Add to that 31 GOP-held seats regarded as tossups, and 26 more as Lean GOP.  That makes 57 GOP held seats that are tossups or only slight margins for Republicans  in which Democrats need to pick up a net 10.  There are another 18 Likely GOP seats, which are not really safe this year.  Only 6 Democrat-held seats are regarded as tossups.  Given these daunting numbers, it is very difficult to see how Republicans hold the House and easier to see a possibility of a significant Democratic gain or 30 or more.

There is of course, the possibility of systematic polling error, in which one party runs several points higher than poll averages in many races.  In 2016, Trump ran 2 points better than polling averages nationwide, but  4 or 5 points better than polling averages throughout the Midwest, enabling  his narrow win in a few rustbelt states no Republican had won since 1988 -- Michigan, Wisconsin and Pennsylvania.  The biggest obstacle facing Republicans in the contest for House control is how many of their incumbents are stuck in the mid-40s in recent polls, even if only a few points down.  In general, late-breakers do not decide to go with an incumbent if they have found no reason to do so during a long campaign.

Normally, the party that has control of  a seat, especially one where an incumbent is running, has a strong financial advantage over the challenger candidate.  This year, the Democrats have out-raised their opponents who are defending seats all over the country.

Turnout

Turnout is an issue, too, in midterms.  Normally, some of the strongest groups for Democrats in Presidential years, African Americans and Hispanics, vote in greatly reduced numbers in the midterms.  The drop-off is not usually as steep for GOP-voting groups.  This year, turnout -- based on early voting, and survey results showing intent to vote -- looks higher across the board than prior midterm races.  Turnout in suburban areas appears to be particularly high.  So, too, in two Southern states where African American candidates are running for Governor, Florida and Georgia, turnout among this group seems to be much higher than normal for a midterm.  When Barack Obama ran for President, not only did black turnout increase substantially (by 4 million from 2004 to 2008), but the share of the black vote the Democrats won grew as well.  An increased margin for Democrats among African Americans  is likely in these two states as well, which should help down-ballot Democrats running for the House in both states.

So far, early voting numbers have shown strength for Republicans in several states including Arizona,, Indiana, Michigan, Texas, Missouri, North Dakota, Tennessee, Wisconsin, West Virginia, Ohio and even Pennsylvania.  These numbers could change with weekend voting, as they have in Florida where Democrats had a big day on Saturday.  The dynamic of who votes early versus on Election Day is not stable in individual states from one election cycle to another, so one should not read too much into the early voting totals. 

In some states, registered Democrat often vote Republican, particularly in the Southern states.  In suburban districts in many states, early voting by registered Republicans could mean women who normally vote Republican are showing up to vote against Trump and Republicans this cycle.  In most polling to date, Democrats are holding a higher percentage of their voters than republicans.

Overall, my forecast is for a pickup for Democrats in the 30-35 seat range, slightly more likely above the range than below..  This range is not that far above 23, so it is not impossible for Republicans to hang on, but I think Nate Silver’s odds are about right -- maybe a 15% probability they can do it.

Senate

The Senate is a different story.  Currently, the GOP has a 51-49 edge after throwing away the Alabama seat .  The GOP could lose a seat and wind up at 50, or have a great night and end up as high as 55.  There are 4 somewhat or very vulnerable GOP held Senate seats: Texas, Tennessee, Arizona and Nevada.  Arizona is an open seat (Jeff Flake retiring), and polls in the last week have shown the Democrat Kyrsten Sinema up 6, and the Republican  Martha McSally up 7.  Other polls have shown a tie.  I think McSally will win a very close race.  Arizona is changing -- it is becoming more of  a purple state than a red state, but the Republicans should win the Governor’s race handily, and some slight spillover from that one will aid McSally.

Nevada is very tough to call for a Republican -- incumbent Dean Heller over Democrat Jacky Rosen.  Rosen is ahead in the last two public polls, after Heller had led most of the campaign season.  This is nothing new for Nevada, where Harry Reid somehow always pulled close races to his side in the final weeks.  Even Michael Madigan could get lessons from Reid.  Heller won by 1 point in 2012, a Presidential election year when Obama won the state by 6%.  This year the Governor’s race is as close as the Senate race, and they will probably both go in the same direction -- my guess for the Democrats.

Republicans Ted Cruz and Marsha Blackburn  have held mid-single digit  leads in Texas and Tennessee for the last month.  In both states, however, the Democrat -- Phil Bredesen in Tennessee, and Beto O’Rourke in Texas --  are still in the race, and an upset is not impossible.  The fact that President Trump is holding a rally in Tennessee in the last few days is indicative of the tightening in the Tennessee Senate race.  Texas is  less solidly Republican than in prior years (Romney won the state by 16%, Trump by 9%).  I expect both Republicans to win in part because Republican governor candidates at the top of the ticket in each state will win easily, with some bump for Republican Senate candidates.

Republicans are expected to pick up a Democratic held seat in North Dakota, where Kevin Cramer has a solid lead over incumbent Heidi Heitkamp.  If Republicans only lose Nevada among their seats up in 2018, then their floor is 51 with the North Dakota pickup.  If Republicans lose both Arizona and Nevada,  by no means a shock if it happens, then  North Dakota would preserve their Senate control at 50 with Pence the tie-breaker.  That is truly a worst case scenario. 

Two races, in Missouri and Indiana, are both true toss-ups, with Josh Hawley, the Republican, maybe a slight favorite in Missouri over Claire McCaskill, and Joe Donnelly, the Democrat, slightly ahead of Republican challenger Mike Braun in Indiana.  Republicans also have a shot at a pickup in Florida where Democrat Bill Nelson holds a narrow lead over Governor Rick Scott, who has made a habit of eking out narrow victories in two prior statewide races.  Scott has taken the lead in a few recent polls, and seems to be running stronger than the Republican candidate for Governor in Florida .If Republicans sweep the three close races -- Missouri, Indiana and Florida --  they would be at 55, if they also hold Arizona and Nevada.

I am not that optimistic. I think Republicans will net at least 1-- most likely Missouri to go along with North Dakota and a loss of Nevada and possibly 2.  But their chances of gaining a net 2 or 3 seats are much better than the likelihood of losing control (meaning a loss of Texas or Tennessee plus Nevada and Arizona, and  a pickup of North Dakota.

The race in West Virginia has been closing with Joe Manchins’s lead over Republican Patrick Morrissey shrinking this week.  A late visit by President Trump suggests this race is still winnable.  Manchin voted for Kavanaugh, after it became clear his vote did not matter in the outcome, and this probably helped his chances to save  his seat.  The other possible GOP pickup is Montana, where Democrat Jon Tester has led in every poll, but the libertarian candidate withdrew this week and endorsed Matt Rosendale the Republican.  The problem is that many Montanans had already voted before this occurred.  Tester’s lead shrank in recent weeks to the 2-3 point range, so a Rosendale win would be a surprise, but not a complete shock.

In New Jersey, Bob Menendez , despite his ethics issues, will likely survive in a very blue state, but by smaller margins that normal for an incumbent Democrat.

Governors

Governors’ races are likely to be a very bad story for Republicans this Tuesday.  Democrats are poised to pick up Michigan and Illinois easily, and hold Pennsylvania.  Democrats are narrowly ahead in Wisconsin and finally may turn out Scott Walker.  In the two marquee races, Republican Brian Kemp seems very narrowly ahead of  Democrat Stacy Abrams in Georgia, and in Florida, Anders Gillum is a few points ahead of Republican Ron DeSantis.  DeSantis seems to have closed the gap a bit in the last few days, but is still an underdog. 

Georgia is another state that is less red than in the past, and heavy black turnout is keeping this close.  There are also surprisingly close Governors races in South Dakota, Iowa, Oklahoma  and Kansas.  Republicans should not lose statewide races in Kansas, Oklahoma  or South Dakota, but might this year due to the specific candidates on the ballot.  Ohio is another close governor’s race, where Democrat Richard Cordray appears to be a few points ahead of Republican Mike DeWine, another uninspiring choice for Republicans. 

If Democrats elect governors in Wisconsin, Ohio, Michigan and hold Pennsylvania, they will have a say in state legislative and US House redistricting for the 2022 midterms.  Republicans have held  the House since 2010 in part due to favorable redistricting in North Carolina, Ohio, Michigan and Pennsylvania after the 2010 census, at one point a 44-17 House split favoring Republicans in these four states.  Some of these states will lose seats after the census, but in any case, GOP margins will be greatly reduced in all of them, starting this year in Pennsylvania with the court-ordered map favoring Democrats.  Though Democrats could win both Florida and Georgia,  I think Kemp wins narrowly in Georgia and Gillum wins in Florida.

When black candidates are on the top line of the ballot, there are possible polling errors due to voters not wanting to appear racist to a pollster by supporting the white candidate (a kind of fake virtue signaling known as the Bradley effect).  This embarrassment factor may have been one reason why Trump surpassed his poll numbers in 2016.

The so-called blue wave will occur if Democratic turnout is much higher than Republican turnout, as it appeared it might be until September.  The Senate Democrats’ ham-handed behavior in the Kavanaugh nomination for the Supreme Court seemed to help lift Republican enthusiasm for voting in the midterms and may be responsible for a good outcome for the party in a few Senate races in heavily red states.

The major takeaway from looking  at the key races is how close they are in many states.  Normal margins of error in most surveys are plus or minus 3 or 4 points, and there are many statewide (Senate or governor) and House races where one candidate’s lead is in the 1-2 point range.  These are hardly secure margins.

Image credit: Pixabay