On Monday, Daily Kos Elections rolled out its new 2018 elections portal. It’s a place with one-stop shopping for elections geeks, with interactive maps showing our race ratings and with polling data from our partners at Civiqs. Maybe most significantly, we have separate Senate, House, and gubernatorial pages, where you can look in more detail at each of those three battlegrounds. Click on any state on those pages and you can get a full list of polls released in those races, along with a trendline that charts the current polling average in each state.
Each week, we’re going to be talking in more detail about each of these battlegrounds, with separate posts about how each of the Senate, House, and governors’ situations are shaping up. Today, we’re going to start with the Senate. The House, oddly enough, has seemed to get the most attention in the runup to the 2018 election, probably because it always seemed a better shot for the Democrats to flip one chamber and restore divided government. (Democrats need to flip 24 seats to take control of the House, which seems like a lot, but midterm waves quite often reach that size for the party out of the White House. And a phenomenal pace of House Republican retirements, often in already vulnerable seats, certainly greased the skids toward that goal.)
The Senate, by contrast, was always kind of the ugly duckling of the duo. Because 2018 is the boomerang for the very successful Democratic Senate campaigns of the 2012 election, there are only nine Republican-held seats up for grab this year (including a special election in Mississippi), and that’s compounded by the problem of the Democrats having to defend a number of seats in dark-red states. Most prognosticators have stayed skeptical about Democratic odds of flipping the Senate, even while considering a House takeover likelier than not.
In recent weeks, though, the Senate has started to move a bit more toward center stage — potentially very important, because, as nice as control of the House would be from an oversight perspective, it’s only the Senate that gives you control over the power of confirmation for the federal courts and executive branch positions. That shift in focus may partly be because, as things increasingly fall apart for House Republicans, many prognosticators have started treating a flip in the House as almost a given (note: it’s not a given, and please don’t treat it that way) … but also because several key Senate races have either solidified a bit for the Democrats (i.e. Florida) or moved more visibly into contention (i.e. Texas), making the math more plausible. The path to a majority still requires almost everything to go right for them, but it’s also becoming more clearly visible.
The key thing to remember about the Senate is that we’re currently at 49 Democrats (including Dem-caucusing indies) and 51 Republicans, so the Democrats need to pick up a net two seats to get to 51 seats and take control. (A one-seat gain for 50-50 tie isn’t sufficient with a Republican vice-president acting as tiebreaker.) There are multiple ways to accomplish this: the shortest route, of course, would be to pick up two seats while losing no incumbents. But it can also be accomplished by picking up three seats while losing one incumbent — and that’s increasingly looking like what may happen. (A really crazy outcome could be picking up four seats while losing two incumbents, but that’s kind of at the outer bounds of what’s possible.)
To help visualize this, we’re going to arrange the competitive Senate races in something of a “totem pole,” with the likeliest Democratic wins at the top and the likeliest Republican wins at the bottom. (To keep the scope manageable, we aren’t going to include the majority of the races: the ones we rate as Safe for one party or the other. It won’t increase your situational awareness to know that we’re on track to win the Maryland Senate race by 40 points, for instance.) Slicing through the middle of the totem pole is the Red Line of Death, at the 51 D/49 R seat mark. To win the Senate, there need to be either two Democratic pickups above the line and no losses below it, or three Democratic pickups above the line and one loss below it.
| STATE |
D CAND. |
D AVG. |
R CAND. |
R. AVG. |
DIFF. |
FLIP? |
| OHIO |
Brown (inc) |
50 |
Renacci |
37 |
+13 |
|
| WISCONSIN |
Baldwin (inc) |
52 |
Vukmir |
40 |
+12 |
|
| NEW JERSEY |
Menendez (inc) |
42 |
Hugin |
34 |
+8 |
|
| MINNESOTA (sp.) |
Smith (inc) |
44 |
Housley |
37 |
+7 |
|
| WEST VIRGINIA |
Manchin (inc) |
46 |
Morrisey |
39 |
+7 |
|
| ARIZONA |
Sinema |
47 |
McSally |
43 |
+4 |
D FLIP |
| FLORIDA |
Nelson (inc) |
48 |
Scott |
44 |
+4 |
|
| MONTANA |
Tester (inc) |
48 |
Rosendale |
45 |
+3 |
|
| NEVADA |
Rosen |
46 |
Heller (inc) |
43 |
+3 |
D FLIP |
| INDIANA |
Donnelly (inc) |
45 |
Braun |
43 |
+2 |
|
| MISSOURI |
McCaskill (inc) |
46 |
Hawley |
45 |
+1 |
|
| TENNESSEE |
Bredesen |
47 |
Blackburn |
46 |
+1 |
D FLIP |
| RED LINE |
RED LINE |
|
RED LINE |
|
|
|
| NORTH DAKOTA |
Heitkamp (inc) |
45 |
Cramer |
48 |
-3 |
R FLIP |
| TEXAS |
O’Rourke |
45 |
Cruz (inc) |
48 |
-3 |
|
| MISSISSIPPI (SP.) |
Espy |
28 |
Hyde-Smith + McDaniel |
48 |
-20 |
|
Lo and behold, today’s averages actually show the Democrats poised to take control of the Senate! (I say “today,” because one bad poll in any of half a dozen states could tip the average in that state and move it below the red line tomorrow, meaning Democrats don’t take control of the Senate. That’s the kind of knife-edge we’re on, right now.)
And it’s not happening the ways you might be expecting, which may be either picking up Arizona and Nevada — states where the Democratic candidates have been narrowly leading the polling averages for most of the cycle — while saving all our incumbents … or losing one incumbent while picking up Arizona, Nevada, and Texas, where Beto O’Rourke has certainly exceeded expectations and captured a lot of people’s imaginations (even while not, at any point, having captured a lead in the polling average).
Instead, the third pickup may be happening in the somewhat below-the-radar open seat race in Tennessee, where the Democratic former Governor Phil Bredesen has led a majority of polls against hard-right Republican Rep. Marsha Blackburn, who gave more establishment-flavored incumbent Bob Corker a hard nudge out of the GOP primary and into retirement. The septuagenarian and decidedly unhip Bredesen — who won re-election in 2006 in his dark-red state by a whopping 69-30 margin — may in fact be the Democrats’ biggest recruiting score of the entire cycle, because it’s hard to imagine anyone else putting Tennessee into play like this.
Unfortunately, the tradeoff may be the loss of Heidi Heitkamp, the Democratic incumbent in North Dakota — a state that, despite its occasional willingness to elect Democrats downballot, swung even harder in Donald Trump’s direction (63-27 in 2016) than Tennessee (61-35). The polling average puts Heitkamp down by low single digits to Republican Rep. Kevin Cramer, which seems to match general rumors about what unreleased internal polls in that race are showing.
Polls aside, though, there’s historical precedent for thinking that Heitkamp might survive — not just that Heitkamp trailed in the poll averages in the runup to the 2012 election as well (North Dakota is a poorly, infrequently polled state, compounded by its unusual prohibition against automated dialing), but also because it’s just extremely unusual for incumbent Senators from the party out of the White House to lose in midterms, even in theoretically difficult states (the last time that happened was in 2002).
In fact, if you’re looking closely at the qualitative ratings on our Senate page, you might be noticing that we still rate North Dakota’s race as a Tossup (along with the other Democratic incumbents in tight races) while we rate Tennessee’s race as Lean Republican (along with Texas). You might be saying “What gives? That doesn’t jibe with the polling.”
Well, that’s sort of the point of the qualitative ratings that we and other prognosticators do: they can take more concepts into consideration than just the polling averages, like candidate quality or a state’s quirky historic preferences, or similar factors where it’s hard to assign a number. Also, if you look at charts comparing all qualitative prognosticators' picks, you’ll see that slightly more race raters view North Dakota as a “Tossup” race than Tennessee, so we’re hardly alone in that stance. (And in fact, if the polls continue their current trajectory, we may well soon be swapping Tennessee and North Dakota’s places on the chart anyway; none of these ratings, of course, are set in stone.)
Heitkamp, for instance, benefits from not just her incumbency status, but also from a reputation as a strong retail campaigner, in a state that’s sparsely-populated enough that retail politicking in a Senate race still works. And the experience of Evan Bayh in the 2016 Indiana Senate race can be considered informative when thinking about this year’s Tennessee race; Bayh, like Bredesen, was another moderate blast-from-the-past who surprisingly re-emerged and started strong, though Bayh lost momentum at the end. Bredesen, though, unlike Bayh, never “went Washington” or had exploitable residency issues, so it’s not clear that the undecided voters will break as badly against him as they did with Bayh in 2016.
One development you might be noticing is that we aren’t running a full “model” this year; in other words, we aren’t running thousands of Monte Carlo simulations of each race and then assigning a specific percentage to the Democrats’ odds of flipping the Senate. One thing that 2016 reminded us is that a predictive model like that is only as good as the polls that are fed into it, but it also reminded us that people’s tendency is to grasp onto one topline number and then take that number way too literally.
A “40 percent chance” of something happening, for instance, shouldn’t be taken to mean that it’s likely that thing won’t happen, no more than a “60 percent chance” of something happening should be taken to mean that it’s likely it will happen. And studies show that the potential harm of probabilistic forecasting goes beyond just users misunderstanding the range of possibilities, to actual harmful voter behavior that we don't want to encourage, i.e. discouraging turnout, out of either despair or overconfidence.
If I needed to boil it down to what I think would happen if we were truly running a model, though, I suspect the odds of picking up the Senate would still be slightly weighted against the Democrats (i.e. under “50 percent”), even though the “totem pole” formulation suggests that the modal outcome is one where the Democrats would hit 51 seats. That’s because so many races are only narrowly tipping in the Democrats’ favor, by one or two points — so there are many more chances for something to go wrong on the Democratic side, if the prevailing breeze shifts a little or simply because of the nature of random error. In other words, there are going to be more possible permutations where the Democrats lose, for instance, North Dakota and Indiana, or North Dakota and Missouri, or North Dakota and Nevada, compared with permutations where the Republicans, say, lose North Dakota and Texas.
But the fact that the most direct route still does, in fact, lead us to a 51-49 Democratic majority, should be very heartening (especially considering where our expectations about the Senate were a year or even six months ago).
