One way to delegitimize your Democratic opponent? Refuse to debate them.

The Debate over Debates

I recently noted a proliferation of stories about endangered Republican House incumbents avoiding debates. In places like OH-12, CA-45, CA-48, NY-27, and IL-14, Democratic House candidates have been hitting their opponents on their unwillingness to publicly discuss the issues important to voters. I began to wonder: how widespread is this problem?

To answer this question, I compiled a dataset of the 68 races currently rated Lean D, Toss-Up, or Lean R by Cook Political Report. I searched online for evidence that a debate had yet occurred or was scheduled for the future. I was startled to find that about one third of the races — fully 22 (32%) of the 68 — do not yet appear to have a debate scheduled.

The seats, all currently held by Republicans, are listed below:

Race	Rating
CA-50	Lean Rep
FL-15	Lean Rep
GA-6	Lean Rep
IL-14	Lean Rep
MO-2	Lean Rep
NC-2	Lean Rep
NY-27	Lean Rep
OH-12	Lean Rep
SC-1	Lean Rep
TX-31	Lean Rep
WV-3	Lean Rep
CA-45	Toss-Up
CA-48	Toss-Up
FL-26	Toss-Up
FL-27	Toss-Up
NC-13	Toss-Up
NM-2	Toss-Up
PA-1	Toss-Up
TX-7	Toss-Up
TX-32	Toss-Up
CO-6	Lean D
MI-11	Lean D

It’s important to note that not all of these races fit the “Republican incumbent ducking debates” narrative. Some, like MI-11 and NM-2, are open seats where I can’t find evidence for either candidate publicly seeking a debate. And even some Dems facing incumbents aren’t prioritizing debates in the closing weeks of their campaigns.

Nonetheless, many races fit a pattern of Republicans seeking to delegitimize opponents by refusing to share a debate stage. And, as I’ll discuss below, the problem is particularly acute for younger Democratic candidates.

Comparison to 2016

Is the paucity of debates this year notable? At least relative to the last election, the answer appears to be yes.

I analyzed the final Cook ratings from 2016. The playing field was narrower that year, with only 37 races rated Lean D, Toss-Up, or Lean R. But of these races, I only found six (16%) where a debate wasn’t ultimately scheduled: CA-10, CA-49, KS-03, MI-08, NV-03, and NV-04.

With four weeks until the election, there are certainly still opportunities to schedule debates. Some race ratings will also change. But as it stands, the proportion of competitive House races without a debate may double relative to the last election.

What correlates with a debate?

Seeing these data, I had two hypotheses that were immediately proven false. First, I supposed that incumbent Republicans would be less likely to debate than non-incumbent Republicans. But I found that the proportion was virtually identical between the two groups: 32% for non-incumbents vs. 33% for incumbents.

I also thought Republican males might be hesitant to debate female Democratic challengers. But no — it turns out that races between Republican men and Democratic women actually have the best debate coverage. Races with female Republicans are less likely to have debates, with all-female races the least likely of all. Yet the overall number of such races is small, so we should be cautious in over-interpreting these results.

The percentage of un-debated races of each type is provided below, with the total number of such races in parentheses:

		Dem Gender
		Male	Female
Rep Gender	Male	28% (25)	26% (27)
	Female	43% (7)	56% (9)

A somewhat stronger relationship is found when looking at the race ratings. The more a competitive race favors the Republican, the less likely there is to be a debate scheduled:

This makes intuitive sense: if the Republican candidate senses he or she is favored, then a debate has little upside and much potential risk. But if he or she is behind, then it’s worth publicly sparring with the Democratic opponent.

A final, surprisingly salient effect is that of the Democratic candidate’s age. Specifically, the youngest crop of nominees — those 35 and under — have been denied debates at high rates. Again, the total number of such nominees is small (only a dozen), so we should exercise some caution. The percent of races without a debate by Dem nominee age is given below:

Dem Nominee 35 and Under	Dem Nominee 36+
58% (12)	27% (56)

Further analysis showed that neither the Republican’s age nor the candidate age gap seemed to be a strong predictor. And the relationship isn’t linear: Democrats ages 36 to 45 secured debates just as often as Democrats over 45.

Again, there seems to be a fairly intuitive explanation. The youngest Democratic candidates will often be those with the shortest resumes. Republicans, particularly if they are incumbents, can assail such opponents as inexperienced and unprepared for the rigors of the job. Sharing a debate stage with a young opponent would implicitly lend them legitimacy as seekers of higher office. And it would force Republicans into a face-off with young and frequently telegenic candidates.

So it’s clear why Republicans House candidates are well-served to avoid debating their opponents. Less well-served, of course, is democracy itself.

 

Donating to “Robustify” the Blue Wave

As we enter the final sprint of the 2018 midterms, I’ve been thinking about how things could fall apart. Democrats are by no means guaranteed a House takeover. The current 538 forecast gives them a 77% chance of winning. That’s only about five percent higher than Hillary Clinton’s final probability of winning in 2016.

I’ve also been thinking about how to strategically donate to House candidates. As Democrats, I’ll argue that the smartest way to make the Blue Wave “robust” is to donate to underfunded candidates in races slightly favoring the Republicans. Folks like Linda Coleman, Lauren Underwood, and George Scott may well be the key to a Blue Wave.

The Nightmare Scenario

To see how Republicans could narrowly hold onto the House, we need only look at the results from the last election.

Around this time in 2016, Cook rated 8 seats as Lean D, 17 as Toss-Ups, and 12 as Lean R. All but one of the Lean D seats ultimately went to Dems; all of the Lean R seats ultimately went to Republicans. But the Toss-Up seats were a disaster for Team Blue: only 5/17 (29%) were won by Democrats in November.

This raises an interesting point about the Cook ratings. Over a horizon of many elections, the ratings are spot-on. A review of their pre-Labor Day ratings between 1984 and 2008 found that “of the 130 Democratic-held seats rated as Toss Up, 49.2 percent went for Democrats, and 55.0 percent of the 160 Republican held seats rated as Toss Up were won by the GOP.”

Yet within an individual election, Toss-Ups can break disproportionately for one party or the other. These effects just tend to cancel out over time as some races break for Team Blue and others for Team Red. And 2016 — in which state-level polling failed to account for stark educational differences in party support and Democratic enthusiasm collapsed following the Comey letter — was a dramatic example of Toss-Ups breaking for one party.

Now, the nightmare: suppose 2018 is also suffering systematic polling error, or an October surprise weakens Democrats late in the race. If the same win probabilities were to be applied to the current crop of 11 Lean Democrat and 31 Toss-Up seats, Dems would pick up 18 seats: five short of the number needed for the majority.

And that would be the worst.

Expanding the Map

In this model, Democrats have two options for a robust Blue Wave: either they push more Toss-Up seats to Lean Dem, or they expand the pool of Toss-Up seats.

The first option is intriguing: if nine currently rated Toss-Up seats shifted to the Lean D column (all else being equal), then Dems would win a 218 seat majority even under 2016 win probabilities. This may well be a path to victory. But a lot of smart election folks have argued this year and last that Democrats are better off expanding the map than they are investing narrowly in promising candidates.

Part of the argument comes down to money, and this is particularly true if you’re a small-dollar donor considering whom to support. Many Toss-Up races are already attracting blockbuster spending from outside groups. Sixteen such races already have more than $10MM committed from outside entities, and the candidates themselves are raising crazy amounts. With parties pouring in money, the spending required to get one of these races from Toss-Up to Lean Dem may well be astronomical.

An alternative path for the civic-minded Democrat would be to invest in underfunded Dem candidates who just might make a sleeper race competitive. We’d need 17 additional Toss-Up seats to get to a victory under the 2016 probabilities — a heavy lift. But even moving a handful of races leftward would provide some “insurance” for the Democrats.

The Data

Unfortunately, as of this writing, Q3 fundraising numbers aren’t available for most candidates. To give some early suggestions, I’m going to use the Q2 numbers available from Daily Kos. I’ll aim to update this post when newer fundraising numbers are released, as I anticipate substantial changes in some of the races.

I’m also lising outside spending numbers from Open Secrets. In the results below, the amount of variability is really quite striking.

CD	Rep Candidate	R Cash on Hand ($1000s)	Dem Candidate	R Cash on Hand ($1000s)	Ratio	Outside Spending ($1000s)
NY-27	Chris Collins (I)	1,342	Nathan McMurray	82	6%	6
GA-06	Karen Handel (I)	1,003	Lucy Mcbath	151	15%	?
MO-02	Ann Wagner (I)	3,388	Cort VanOstran	541	16%	30
GA-07	Rob Woodall (I)	529	Carolyn Bordeaux	98	19%	0
PA-16	Mike Kelly (I)	1,700	Ron DiNicola	343	20%	31
AR-02	French Hill (I)	1,645	Clarke Tucker	386	23%	469
NE-02	Don Bacon (I)	1,067	Kara Eastman	258	24%	1,025
FL-16	Vern Buchanan (I)	2,492	David Shapiro	785	32%	653
MT-AL	Greg Gianforte (I)	1,373	Kathleen Williams	462	34%	?
PA-10	Scott Perry (I)	554	George Scott	208	38%	67
IL-13	Rodney Davis (I)	1,592	Betsy Londrigan	659	41%	408
FL-26	Carlos Curbelo (I)	2,601	Debbie Mucarsel-Powell	1,264	49%	3,681
OH-12	Troy Balderson (I)	275	Daniel O’Connor	148	54%	?
TX-23	Will Hurd (I)	2,019	Gina Ortiz Jones	1,151	57%	2,359
WA-05	Cathy McMorris Rodgers (I)	1,711	Lisa Brown	1,053	62%	74
NC-02	George Holding (I)	253	Linda Coleman	157	62%	4
IL-14	Randy Hultgren (I)	1,020	Lauren Underwood	652	64%	62
CA-50	Duncan Hunter (I)	352	Ammar Campa-Najjar	280	80%	3
WV-03	Carol Miller	189	Richard Ojeda	163	86%	161
UT-04	Mia Love (I)	1,233	Ben McAdams	1,249	101%	323
TX-31	John Carter (I)	538	M.J. Hegar	867	161%	27
FL-15	Ross Spano	108	Kristen Carlson	193	179%	56
VA-05	Denver Riggleman	208	Leslie Cockburn	483	232%	94
WI-01	Bryan Steil	619	Randy Bryce	2,154	348%	620
SC-01	Katie Arrington	62	Joe Cunningham	319	515%	21

Open Secrets doesn’t appear to separate spending between special elections and regular elections each cycle, so correct outside spending numbers for GA-6 and MT-AL aren’t readily available.

Some Semi-Scientific Suggestions

There are a lot of rational ways to pick candidates from the above list. Here is my thinking:

• I’d generally advocate picking candidates who aren’t already way ahead in cash (since you might get diminishing returns) and who aren’t being swamped by outside spending (since your money might not go very far). Cheaper races, in terms of total dollars spent, are also attractive.
• Given that we’re working with out-of-date data, I’m not suggesting OH-12, CA-50, or NY-27 because these races’ dynamics have shifted so much since Q2 (In OH-12, the Republican won the special election; in CA-50 and NY-27, the Republican incumbent was indicted). Q3 fundraising numbers will strongly influence my decision to donate.
• Polling data never hurts, and here is a great repository of available polls maintained by a Redditor. Though all the races are rated Lean R, candidates who have a few polls showing them within striking distance might be the closest to becoming true Toss-Ups.

With this in mind, here are six candidates I’m considering supporting financially:

1. Linda Coleman (NC-2): As of Q2, little money had flowed into the race, and Dem Coleman was about $100K behind incumbent Republican George Holding in cash on hand. Yet something odd is happening here. Coleman has had strangely good polling, especially given that NC-2 voted for Trump by 9 points and Romney by 13. Maybe it’s a fluke, but giving Coleman the resources to mount an aggressive final month of campaigning might just be worth it.
2. Kristen Carlson (FL-15): For this race, we actually do have more up-to-date fundraising numbers, showing Carlson in a tough spot before her primary. The CD is Trump +10, but Carlson has put up some strong poll numbers and she is not facing an incumbent. Some promising Florida districts, like the 26th, are getting a ton of outside spending, but this one hasn’t drawn attention yet.
3. Lauren Underwood (IL-14): Underwood’s race is quite a bit pricier, but the contest has attracted very limited outside spending so far. The only available poll showed her four points behind, but Underwood’s CD is one of the less Trump-y on this list (he only won by four points). And the extremely low media profile of this race could offer an opening.
4. Carolyn Bourdeaux (GA-7): Bourdeaux had actually out-raised her opponent as of Q2, but spent down her cash to win a contested primary. It’s plausible that she used Q3 to replenish her cash reserves, but she may be worth a donation anyway. She was ahead in one (very biased) poll, and GA-7 swung from Romney +22 to Trump +6.
5. George Scott (PA-10): This is a race on which Nate Silver is decidedly not bullish, But with both candidates holding relatively little cash, and the district redrawn from Trump +21 to Trump +9, it’s hard not to wonder if money could be beneficial here.
6. Richard Ojeda (WV-3): This one comes with a caveat that Ojeda’s media savvy might have financially shifted the race in Q3 — but as of Q2, both candidates had very little money. Polls of this race are all over the place, with Silver giving him very low odds. This race seems the definition of “high-risk, high return.” And who wouldn’t want to flip a Trump +50 CD?

In the 116th House, Expect a Narrower Age Gap — But Dems Will Still be Older on Average

Introduction

A strange political trend is found in the fact that Democrats — who are overwhelmingly favored by younger voters — tend to elect older Congresspeople. At the start of the 115th Congress in 2017, the average Democratic Congressperson was 60, while the average Republican Congressperson was 55.7. The gap between the medians was even larger: 62 for Democrats vs. 56 for Republicans.

Much of this has to do with the recent electoral success of Republicans. Bringing more new faces to Congress tends to yield a younger caucus overall. Matthew McConaughey put it best: incumbents get older, but freshmen stay the same age.

Republicans had a wildly successful election in 2010 and another successful one in 2014, while their losses in the subsequent presidential years were limited. Of 217 Congresspeople, serving at the start of the Trump presidency, who were first elected in one of the 2010-2016 regular elections, 140 (65%) were Republicans. Many of these are young conservatives swept into power by the backlash against President Obama — folks like Justin Amash (38), Carlos Curbelo (38), and Elise Stefanik (34).

Democrats now stand to make their biggest gains in Congress since at least 2008. I was curious what this would mean for the age distribution of Democrats in the House. The answer: the Congressional age gap is set to narrow, but unlikely to disappear.

Dataset

An existing dataset from Daily Kos provides birth years for all incumbent Congresspeople. To compute expectations for the next Congress, I supplemented this dataset with data from the Prognosticator Tracker, which aggregates election ratings across the three big election handicappers (Cook Political Report, Inside Elections,  Sabato’s Crystal Ball). Probabilities of Democratic or Republican victory in each seat are estimated by computing historical averages and then averaging across the three raters.

Candidate birth years had to be manually sourced. Much of this work was done using ages listed in news articles, meaning there are likely many off-by-one errors. The data can be found here — let me know if you spot any issues!

Results

Expectation

In the next Congress, the average Democrat is expected to be about 2.5 years older than the average Republican, a decline from the 4.3 year gap at the start of 2017.

The drop is mostly driven by a crop of younger Democratic candidates in competitive races: the average Democrat is expected to be 58.7 years old at the start of 2019, 1.3 years younger than the caucus average at the start of the current Congress. The average Republican will be 56.2, a slight uptick relative to the current Congress.

As always, I should emphasize that this is an expectation, not a projection. There is a wide range of ages among Democratic candidates in Toss-Up races, from 77-year-old Donna Shalala in FL-27 to 31-year-old Katie Hill in CA-25. If things break disproportionately toward the older or the younger crop, the average Democrat’s age could shift around.

Uncertainty

This brings me to the next point: what we’d really like is a confidence interval. This would be straightforward to compute if the race outcomes were independent, but as I talked about in the last post, they’re not. Covariance between the races can have a big effect on the overall uncertainty. And the covariance matrix is tough to estimate empirically (though FiveThirtyEight seems to have found a way to estimate it).

Even absent a confidence interval, I’ll argue that it’s quite unlikely for the age gap to go away. To see why, note that only 107 races are rated competitive by any of Cook, Gonzales, or Sabato. That means that three out of four races are more or less decided, barring an unexpected event. And here is the breakdown of those seats:

	Number of Seats	Average Age (as of 1/1/2019)
Safe Democratic	183	61.0 (D)
Safe Republican	143	56.7 (R)
Competitive	107	47.9 (D) 54.7 (R)

 

Among the safe seats, future Democratic Congresspeople are a full half-decade older than their Republican counterparts. So to overcome the age gap, Democrats have to both win a lot of these competitive seats (where their average win probability is 36%) and have their wins favor younger Democrats (even though there is zero correlation between Dem win probability and candidate age).

There are configurations where it is possible, but they are exceedingly unlikely.

Next Time

I hope to explore why Democrats aren’t yet electing lawmakers younger than the Republican average.

One intriguing hypothesis is that gender may play a role. In a wide variety of professional settings, biases lead women to be judged as less competent than men. As a result, many often wait to garner more experience before seeking elected office. A consequence is that female lawmakers tend to be older than male lawmakers: the average Congresswoman is about three years older than the average Congressman, for example.

Now, women are on the cusp of major gains in the House. And 78% of the 233 women nominated for Congress are Democrats. So it’s possible that part of the story is about a cohort of mid- and late-career professional women, galvanized by Trump to cast aside doubts and finally throw their hats in the ring.

I’ll aim to explore this theory, and others, in a follow-up piece.

It’s Beginning to Look a Lot Like 1992

With the conclusion of primary season, we can now consider how the November elections will alter women’s representation in government.

The House

Women stand to make their greatest gains in the House. As always, I’m continually updating http://www.morewomeninthehouse.com/ to reflect the current ratings from Cook Political Report.

Historically, the largest gain from one Congress to the next came between the 102nd and 103rd Congress, when the count of Congresswomen jumped from 30 to 48 due to broad victories in 1992’s Year of the Woman. No other election has produced more than a single digit gain in the number of Congresswomen.

As it stands, the expected number of women in the next House is 99.34. There were 83 women in the House at the start of the current (115th) Congress, though 84 are now serving. So, to make an apples to apples comparison: the expected gain from start of the 115th to the start of the 116th Congress is currently 16.34 Congresswomen, just shy of the record.

But of course, this is an expected value, not a projection. If we were to make the (wrong) assumption that the races were independent, then the distribution of the number of women in the next Congress would follow my favorite distribution, the Poisson Binomial. If we then approximate via a normal distribution, a 95% (Wald) confidence interval would encompass outcomes from 94 women in the next Congress to 105 women.

In reality, the race outcomes are mostly positively correlated with one another, because the vast majority of the female candidates are Democrats. If existing measures are generally over-estimating Democrats’ chances, then many fewer women will land in Congress; if they’re under-estimating them, then many more women will be elected. These (mostly) positive correlations induce larger variance in the total number of female-held seats, so a true 95% interval is probably quite a bit wider than (94, 105).

A final point: the expected number of newly elected Congresswomen in the next House is 31. This is well above the record of 24 set in 1992. The reason next year won’t easily set the record for the greatest gain in Congresswomen is the high number of incumbent Congresswomen (13) not seeking reelection. But this “cannibalization” has some benefits for women’s representation: six of these women are now nominees for governorships or Senate seats.

The Senate

Things look less rosy in the Senate. There are currently a record 23 female Senators, but thirteen of them (57%) are up for reelection this year, including two highly vulnerable red state Democrats: Claire McCaskill (MO) and Heidi Heitkamp (ND). Early handicapping also identified Democrats Debbie Stabenow (MI), Tammy Baldwin (WI), and appointed Senator Tina Smith (MN) as potentially vulnerable, but recent polling has seemed to favor all three women. Ditto for appointed Republican Senator Cindy Hyde-Smith (MS), who could theoretically fail to advance to a runoff from Mississippi’s nonpartisan blanket primary, or even lose a head-to-head match-up against Democrat Mike Espy, but I wouldn’t bet on it.

Women do stand to pick up a few seats. Arizona will gain a female Senator in 2019, as Democrat Kyrsten Sinema is facing Republican Martha McSally in a toss-up race for retiring Senator Jeff Flake’s seat. And women could conceivably pick up seats in Nevada — where Congresswoman Jacky Rosen is taking on incumbent Republican Dean Heller — and Tennessee, where Congresswoman Marsha Blackburn is taking on Phil Bredesen, a moderate Democrat and former Governor. All three races, along with McCaskill’s and Heitkamp’s, are rated as Toss-Ups by Cook.

Again, women’s gains are closely tied to outcomes for Democrats. If they have a great night on November 6, then women could conceivably hold 25 seats in January (the incumbent Senators, plus Rosen and Sinema). If Democrats underperform, we might see Heitkamp and McCaskill lose but McSally and Blackburn win, yielding no change in the total number of female Senators. And of course, weirder outcomes are possible that might yield a net loss in female Senators.

Governor’s Mansions

The greatest uncertainty lies in the number of female Governors. As I’ve discussed in previous posts, governorships are unique in that they are the one office where women have actually lost ground in recent years (there are currently six female governors, as opposed to nine at their peak). Governorships are also much more idiosyncratic than federal elections. Case in point: Democrats are almost certain to lose the Massachusetts Governor race, but are competitive in gubernatorial races in Oklahoma and Kansas.

As it stands, six women are favored to win by Cook: incumbents Kay Ivey (R-AL), Gina Raimondo (D-RI), and Kate Brown (D-OR), as well as Kristi Noem (R-SD), Michelle Lujan Grisham (D-NM) and Gretchen Whitmer (D-MI). Polling has continued to be a mixed bag for both Raimondo and Brown, who are the only incumbent Democratic female governors and have middling approval ratings. These races could conceivably move into Toss-Up territory. Things seem to be moving in the other direction for Whitmer, who faced early doubters, but has led all polls since winning the nomination.

Four women are also in Toss-Up races. The buzziest candidate is Democrat Stacey Abrams in Georgia, who is tied in the polls as she vies to become the first black female governor. Democrats also have competitive female candidates in Janet Mills of Maine and Laura Kelly of Kansas. And appointed incumbent Republican Kim Reynolds of Iowa is also in a competitive race against Democrat Fred Hubbell. Five additional Democratic women are gubernatorial candidates, but none are currently considered competitive.

More so than in the House and Senate, it’s plausible that competitive races won’t disproportionately break for one party or the other. In more statistical terms: the covariance between the race outcomes is probably lower. So it’s possible we could end up with a record ten female governors next year, if all the Toss-Up races go to the female candidates. More likely is that we see an increase in the number of female governors, but fall short of or tie the record of nine.

The Ballot Theorem and the NYT/Siena Polls

Introduction

The New York Times and Siena College have partnered to create some really cool “live polls” of key races in the midterm elections. Results are updated in real time as polling participants are dialed, allowing viewers to see just how much the results move around during the polling process.

Early results have shown that races rated as Toss-Ups by Cook Political Report are mostly polling as almost exact ties between the candidates. I recently found myself shouting irrationally at my phone as I watched veteran, viral video star, and all-around baller Democratic Amy McGrath lose her lead to incumbent Republican Andy Barr in the KY-6 poll.

And then I remembered something interesting from my first-year Stats classes: can the Ballot Theorem be applied to this problem?

Ballot Theorem

Bertrand’s Ballot Theorem applies to a simplified election scenario when we only have two candidates (and ignore things like blank ballots, overvotes, etc.). We suppose ballots are counted in a random order, one at a time. We keep a running tally of the margin of votes between the candidates.

The Theorem tells us that the probability that the winning candidate leads throughout the entirety of the counting process is precisely given the winning candidate’s final margin. So if a Democrat beats a Republican by 5% in the final tally, there is a 5% chance the Democrat will lead throughout the entirety of the ballot counting process.

A really nice proof can be found in Chapter 4 of Rick Durrett’s Probability: Theory and Examples. The basic idea builds on the theory of simple random walks: random processes defined by a sequence of variables such that . If we define , then will trace out a path starting from such as the one below:

Here is a sketch of the proof:

• Denote the candidates as A and B, and define as A’s final vote margin over B and as the total number of votes. Any ballot counting order in which A leads throughout can be represented as a path from to which never touches 0 again. The path must go through , so we can equivalently state this quantity as the number of paths from  to without ever touching 0.
• We make use of the Reflection Principle, which tells us that if , then the number of paths from to that are 0 at some time is equal to the number of paths from to . This implies the number of paths from  to which are 0 at some point equals the number of paths from to .
• Some combinatorics tells us that the number of paths from to is given by where .
• Putting this together, we see that the number of paths from to that are never 0 is given by . Some algebra shows us that:
  
  where the final term is the total number of paths from to . Thus, the proof is complete.

Modifying the Ballot Theorem

Now, in the context of the NYT/Siena Polls, we make a few observations and simplifications:

• The random ordering assumption inherent in the Ballot Theorem is more appropriate in this context. In an election, votes are tallied by precinct, inducing a very non-random ordering to the reporting. But in this case, because the polls are conducted by dialing voters at random, the assumption is essentially correct.
• To get to basic results, we’ll ignore several complexities:
  • Respondents are allowed to state that they are undecided, but we ignore this and pretend each response is either for the Democratic or the Republican candidate.
  • Poll results are weighted so as to match the demographic profiles of the expected voter population in November. We ignore this and assume the reported results are simply based on proportions of respondents who say they will vote for each candidate.
  • We ignore any issues induced by non-response and time series dependence (highlighted by Professor Gary King on Twitter).

With these simplifications, we can adapt the Ballot Theorem to these polls. In particular, results are first reported when 150 responses are tallied, and the polls terminate at roughly 500 responses. We want to answer the question: if the Dem is leading by responses with votes tallied, and ultimately leads by responses with votes tallied, what is the probability that she remains in the lead for the full tally between and responses?

• Again, we can count paths. We are interested in the number of paths from to that are never 0.
• Intuitively, this is equal to the number of paths from to that never hit . By the Reflection Principle, the number of paths from to that do hit is equal to the number of paths from to .
• So our total count of relevant paths is given by:
  
• Dividing this by gives us the proportion of paths (and thus the probability of occurrence) in which the Democrat remains in the lead from to responses. This simplifies a bit to:
  Prob(Dem always leads after votes) = 

Results

We can now apply these results with and to see the effect on the NYT/Siena polls. Below, we plot the probability that the poll shows the Democrat consistently in the lead for a variety of potential values at 150 and 500 responses. Note that we’re only looking at positive values on both axes, since if the winner flipped between 150 and 500 responses, the result would obviously have to cross 0.

A key takeaway from these results: even if the Dem has a pretty big lead at both 150 responses and 500 responses, it’s still plausible that she will lose the lead at some point in the counting. If she leads by just over 5% at both times, for example, then there is only a 56% chance she’ll continually hold the lead throughout (and thus a 44% chance she’ll lose it at some point). And with most of the races polling within a point or two in the end, it’s exceedingly likely that the leader will change throughout the counting.

Conclusions

There are a lot of simplifications here, so the above plot is emphatically not a perfect representation of these probabilities. But even in a simplified setting, we see that the polling leader is very likely to change even after 150 responses — especially in the tight races being polled. So if your preferred candidate is losing her lead, take a deep breath and check back in when the poll is completed.

A Blue Wave is a Pink Wave

Tuesday’s primaries more than doubled the expected gains by women in the next Congress. Women are now expected to gain 5.9 seats next January. As always, a full database of candidates and ratings can be found at More Women in the House.

Does the Blue Wave Mean More Congresswomen?

One question that has interested me lately: how intertwined are the Blue Wave, of Democrats overperforming in seats all over the country; and the “Pink Wave,” of women seeking Congressional seats in record numbers? After all, a few things are simultaneously true:

• More than 78% of the female candidates currently set to be nominated to Congress are Democrats. This exceeds the already very high proportion of Congresswomen who are Democrats: 73% (61 out of 84).
• Of the 104 non-incumbent Democratic women nominated, more than half (61) are in races not currently considered competitive by Cook Political Report.
• The Blue Wave represents a threat to incumbent Republican Congresswomen.
  • Of the 17 Republican Congresswomen, nine (53%) are rated by Cook Political as at least somewhat competitive, including two Toss-Up seats (Barbara Comstock in VA-10, Claudia Tenney in NY-22).
  • This is a bit of a “glass cliff” phenomenon, as Republican women disproportionately represent swing seats vulnerable to a wave. About 65 (36%) male Republicans out of roughly 180 seeking reelection are facing competitive races.

So, I was curious: if the Blue Wave endangers female Republican incumbents, and many nominated female Democrats aren’t likely to win seats, does a Blue Wave mean a substantial uptick in women in the House?

The short answer? Yes. 

The Chessboard

As a precursor, let’s look at the current arrangement of female nominees and their districts. A caveat to all that follows: fifteen states have yet to hold primaries, and while my data includes incumbent Congresswomen in those states, it does not include the many races that may have a female nominee (CT-5, MN-5, MN-2, AZ-2, FL-27, MA-3, …). Nonetheless, we can get a pretty good picture from results so far:

• Of the nine Republican incumbents mentioned above, five are facing female Democrats in the general election. The exceptions are Tenney, Mia Love in UT-04, Ann Wagner in MO-02 and Jackie Walorski in IN-02. So even if Democrats “run the table,” only four Republican Congresswomen could plausibly lose their seats to male Democrats (and both Wagner and Walorski are in reach districts).
• Republicans have nominated only five women to competitive open seats: Diane Harkey in CA-49 (Lean Democrat), Young Kim in CA-39 (Toss-Up), Yvette Herrell in NM-2 (Lean Republican), Carol Miller in WV-3 (Lean Republican) and Lena Epstein in MI-11 (Toss-Up). Herrell and Epstein face female challengers — so again, a Democratic wave only potentially blocks three seats from going to freshman  Congresswoman.
• By contrast, Democrats have been consistently filling competitive slots with female nominees. I count 35 non-incumbent Democratic women nominated to races rated as competitive by Cook. Twenty-eight of these women are facing male challengers, and most (19) are in the highly competitive categories of Lean R/Toss-Up/Lean D.

These facts point to a simple conclusion: many female nominees will benefit from Democratic over-performance, and comparatively few will be endangered by it

To visualize these results, we estimate the probability of a Democratic victory by the average of the current ratings from the three big election handicappers (Cook, as well as Inside Elections and Sabato’s Crystal Ball), making use of the great work by the folks at the Prognosticator Tracker. Let’s start by plotting all the current nominees (where we assume all female incumbents will secure their party’s nomination): 

Most seats strongly favor one party or the other. The big checkered blue bar at the left of the plot is the crop of Democratic women vying for extremely red seats. These women are unlikely to make it to Congress. The smaller red bars above it represent likely Republican Congresswomen, most of whom are incumbents. At the right of the plot, we see many more Democratic women (mostly incumbents, but some newcomers as well) who are in safe Democratic seats, along with some female Republican nominees likely to lose their races.

In the smaller pool of competitive races (those toward the middle of the plot), we see a small number of incumbent Republican Congresswomen and Republican nominees, and a larger crop of non-incumbent female Democratic contenders.

What happens in a Blue Wave?

The current ratings reflect a toss-up race for Congress, which is already a big improvement for Democrats relative to recent midterms. We consider two “Blue Wave” scenarios in which Democrats overperform the expected seat count given by the current ratings.

The “Toss Ups to the Left” Scenario

In this scenario, the list of competitive seats remains roughly the same — but on election night, these seats break dramatically toward the Democrats.

To see the effect, we filter to all races where women face male challengers (dropping the 22 races where both contenders are female), and zoom in on the competitive races. We can see immediately why this scenario benefits female representation: 

Suppose Dems sweep every race rated from a 40% rating upward, while the remaining races follow their assigned probabilities (a scenario where Dems would be expected to win about 233 House seats overall). Eleven Democratic women with 40%+ probability ratings would get to Congress by defeating men, whereas only two Republican women will be defeated by Democratic men. If Dems sweep from 20% upwards, the corresponding numbers are 13 and three.

It’s simple: any such scenario helps more women than it hurts, because Dems have more nominees in competitive races against male opponents.

The “Rising Tide” Scenario

An alternative Blue Wave scenario involves an expanding map, yielding more competitive districts before Election Day. We can roughly model this scenario by supposing there is a uniform shift in probabilities toward the Democrats (up to a maximum of 100%). Below, we list the expected gains for women in Congress under a few scenarios:

Probability Shift	Expected Total Dem Seats	Expected # of Women in Congress (84 today)
+5% R	213.1	84.7
+0%	218.3	90.3
+5% D	223.7	93.9
+10% D	228.7	97.5
+15% D	233.4	100.9

 

Every shift toward the Democrats yields more women in Congress — again, by the simple reason that Democrats have more women nominated and many of them are in competitive seats.

And this scenario is somewhat more “efficient,” meaning it yields more women in Congress per seat flipped to Democrats, because there are also so many Democratic women nominated in seats that aren’t currently competitive.

Conclusion

As we have seen, women’s representation in Congress stands to benefit from a Blue Wave. Democrats have nominated many more women to competitive seats, and in 85% of races where a man and a woman will be on the ballot, the woman is a Democrat and the man is a Republican.

The result is that even if the Blue Wave washes some female Republicans from office, they will be replaced with many more Democratic women. We have seen that in two different Blue Wave scenarios — one in which Toss-Ups break for Democrats, another in which the map expands for Democrats — women will still be expected to gain quite a few seats. The latter scenario is especially likely to yield large Democratic gains.

Constituents Prefer Their R Governors and D Senators. The Latter Preference is Growing.

Morning Consult provides quarterly approval ratings for every sitting governor and senator, and their Q2 2018 ratings were just released.

The Latest Numbers

For the sixth consecutive quarter, Morning Consult finds that Republicans hold every slot among the ten most popular governors. This fact is partially driven by overrepresentation: Republican governors outnumber Democratic governors 2 to 1. Another contributing factor is the outsize popularity of moderate Republican Governors in blue and purple states: Charlie Baker of Massachusetts (number 1), Larry Hogan of Maryland (number 2), Chris Sununu of New Hampshire (number 4) and Brian Sandoval of Nevada (number 7).

The newest rankings show stablemate Phil Scott of Vermont suffering a tumble in popularity after signing gun control legislation, ousting him from the top ten. But the overall news still isn’t good for Democrats: while they have some popular incumbents, none achieve the sky-high approvals of the most beloved Republicans.

The Senate rankings paint a rosier picture for Team Blue. Democrats and the Independents who caucus with them are slightly overrepresented at the top: they comprise six of the ten most popular Senators vs. 49% of the overall chamber. Bernie Sanders (I-VT) again registers the highest approval rating. But some some relatively low-key Democrats without large national profiles (Amy Klobuchar of Minnesota, Jack Reed of Rhode Island) also make the list.

I was curious to examine the distribution of approval ratings for Democrats and Republicans in governors’ mansions and the Senate. Above, I’ve plotted the empirical densities of the net approval ratings (percent approving minus percent disapproving) of each party’s governors and Senators over the past six quarters. The median of each density is plotted with a vertical dashed line.

Two important caveats should be noted. First, Morning Consult is perceived by some to have a Republican-leaning House Effect, so the ratings may paint Democrats as somewhat less popular than they truly are.

Second, the politicians featured in these polls have not been static over the past six quarters: Dems picked up one governorship (Phil Murphy in NJ) and one Senate seat (Doug Jones in Alabama), while five governors and two Senators have been replaced by members of their own party. But most of the individuals have remained the same, making comparisons across time fairly reasonable.

Dominant Trends

In both the Senate and Governor populations, approval ratings have tended to fall over time. This is partially the end of the “honeymoon phase” for those elected in 2016, though it could also reflect general displeasure with elected leaders in the Trump era.

An interesting partisan divide exists in how approval ratings are evolving:

• Republican governors are typically somewhat more popular with their constituents than Democratic governors, and this gap has stayed consistent throughout the Trump presidency. As of the most recent survey, the median Democratic governor had a net approval of +12.5 points vs. +19 points for the median Republican.
• Republican Senators are losing ground faster than Democratic Senators. Both parties’ Senators were about equally popular with their constituents at the start of Trump’s presidency. As of July, however, the median Democratic Senator has a net approval of +16 points vs. +8 points for the median Republican.

A few other points are evident from the data:

• People typically like their governors a bit more than they like their Senators.
• There is much greater variability in Governor approvals, with higher highs for popular governors and lower lows for despised ones. This makes some intuitive sense: a governor’s policies are likely to have a more immediate impact on constituents’ daily lives. A good one might bring fiscal balance and improvements to services; a bad one might bring earthquakes.
• Among governor approvals, the variance among Republicans is about twice as high as among Democrats (evidenced by the flatter red density curve). While the typical Republican governor is more popular than the typical Democratic governor, Dems only have one governor with a net approval below -10%: Daniel Malloy of Connecticut. Republicans have six governors in this category.

Discussion

An odd wrinkle to these results is that Democrats are considerably more “overextended” in the Senate than in governors’ mansions. Dems hold 11 Senate seats in states that were lost by Clinton, and 38 of the 40 Senate seats in states that she won. But the inverse is true of governor’s mansions: Republicans hold governorships in eight of the 20 Clinton states and 25 of the 30 Trump states (Alaska’s governor, Bill Walker, is an independent).

With Democratic governors mostly confined to blue states, we might assume their popularity would be higher. But as Nathaniel Rakich and Dhrumil Mehta point out at FiveThirtyEight: “There is basically no correlation between each governor’s net approval rating in the latest Morning Consult poll and his or her state’s partisan lean.”

So perhaps a better way to understand these data is to think about how connected a politician is to the national political environment. Trump’s approvals have been fairly stable for the past three months, but he remains mired in the low 40s. Republican Senators, who share the national political stage with the president, may be suffering by association. But Republican governors — whose policies and leadership are fundamentally local — can better craft their own narrative.

For Democrats, this poses a few problems. Most obviously: they’re very unlikely to take back the governorships in MA, MD, and NH, because the Republican incumbents are so popular going into reelection campaigns. I also wonder if voters see the fundamental character of Democrats as more amenable to legislating — where compromise and consensus-building are key — than to the executive functions of a governorship.

Perhaps an even simpler explanation applies: voters associate Democratic governors with higher taxes, but they don’t always see the benefits of those taxes in the form of improved services. Republicans offer the promise of lower taxes, which is a popular approach — at least until it leads to fiscal ruin and draconian cuts to popular programs. This could explain why Republican governors see greater variation in approval. And it implies that Democrats have to get better at making the case that expanded government can improve people’s lives.

The Blue Wave will make landfall on the Jersey Shore

As the landscape of competitive House races takes shape, I’ve been interested in the geography of the Blue Wave. Which states will be most important for taking back the House this November?

It’s hard to answer this question precisely, so I decided to focus on a more straightforward one: which states are projected to have the largest change in the proportional makeup of their Congressional delegations in the midterm elections?

It turns out my home state of New Jersey is poised for the biggest change in its Congressional composition, while other big moves are possible in Obama-Trump states (Iowa, Pennsylvania) as well as rapidly Dem-trending states (Arizona, Virginia). Read on for the full breakdown.

Methodology

I’m making use of the fantastic work of the folks maintaining the Prognosticator Tracking spreadsheets. Using historical performance as a baseline, they provide probabilities of each seat flipping based on its rating from the three big election handicappers: Cook Political Report, Sabato’s Crystal Ball, and Inside Elections with Nathan Gonzales.

I use these probabilities to compute an expected value for the number of Democratic-held Congressional seats in each state after November’s elections. These are expected values, not projections, so fractional seats are possible. I then compare these expected values against the current makeup of the Congressional delegation.

Note that the current ratings indicate Democrats are poised to gain about 20 seats — just shy of the 23 seats needed to flip Congress. Ratings have been trending toward Dems over the past two years and will likely continue to do so, so these estimates are not final. But they start to give us an understanding of where dense opportunities exist for Democrats.

Results

Results for each state are summarized below.

State	# Seats	Current Democratic Proportion	Expected Democratic Proportion	Expected Change
New Jersey	12	58%	77%	19%
Iowa	4	25%	43%	18%
Montana	1	0%	15%	15%
Pennsylvania	18	33%	48%	14%
Maine	2	50%	64%	14%
Virginia	11	36%	50%	14%
Arizona	9	44%	54%	9%
Kansas	4	0%	7%	7%
Colorado	7	43%	50%	7%
Nebraska	3	0%	7%	7%
Michigan	14	36%	42%	6%
California	53	74%	79%	5%
Florida	27	41%	46%	5%
Washington	10	60%	65%	5%
Illinois	18	61%	66%	5%
Kentucky	6	17%	21%	5%
New York	27	67%	71%	4%
Utah	4	0%	4%	4%
Ohio	16	25%	29%	4%
North Carolina	13	23%	26%	3%
Texas	36	31%	33%	3%
Arizona	4	0%	3%	3%
Wisconsin	8	38%	40%	2%
Minnesota	8	63%	65%	2%
Georgia	14	29%	30%	2%
West Virginia	3	0%	2%	2%
New Mexico	3	67%	68%	1%
Oklahoma	5	0%	1%	1%
Missouri	8	25%	25%	0%
Indiana	9	22%	23%	0%
Arkansas	1	0%	0%	0%
Alabama	7	14%	14%	0%
Connecticut	5	100%	100%	0%
Delaware	1	100%	100%	0%
Hawaii	2	100%	100%	0%
Idaho	2	0%	0%	0%
Louisiana	6	17%	17%	0%
Massachusetts	9	100%	100%	0%
Maryland	8	88%	88%	0%
Mississippi	4	25%	25%	0%
North Dakota	1	0%	0%	0%
Oregon	5	80%	80%	0%
Rhode Island	2	100%	100%	0%
South Carolina	7	14%	14%	0%
South Dakota	1	0%	0%	0%
Tennessee	9	22%	22%	0%
Vermont	1	100%	100%	0%
Wyoming	1	0%	0%	0%
Nevada	4	75%	72%	-3%
New Hampshire	2	100%	94%	-6%

 

Highlights

The Blue Garden State

New Jersey tops the list — and for good reason: of its five Republican-held CDs, Trump lost one (NJ-7) and won three by seven or fewer points (NJ-2, NJ-3, NJ-11).  The incumbent Congressman is retiring in both NJ-2 and NJ-11, and the former seat has essentially been ceded to the Democrats after it was revealed that the Republican nominee had made racist comments.

Meanwhile, Democrats have attracted strong recruits in Mikie Sherill for the 11th and Tom Malinowski for the 7th, putting these seats in contention for the first time in years. And even in the less heralded NJ-3 race, Obama staffer Andy Kim is polling competitively with incumbent Tom MacArthur. Democrats currently hold seven of the state’s 12 seats, but could conceivably pick up as many as four more in November.

The Big Eastern Movers

Fellow populous eastern states Pennsylvania and Virginia are also near the top of the list, but for different reasons. Pennsylvania was redistricted under an order from the State Supreme Court, unwinding the extraordinary gerrymandering that had yielded a 5R-13D delegation from a state with a nearly even partisan divide. The new map gives Democrats a number of terrific opportunities. Pennsylvania would, in fact, have come in above NJ had Conor Lamb not won his special election in PA-18 in March, handing Democrats one more seat ahead of the midterms.

Virginia’s districts are intact, but the state’s Dems are hoping to harness the energy that powered them to a near-takeover of the State House of Delegates last fall. All eyes are on ousting moderate Republican Barbara Comstock in VA-10, who has held on through challenging elections but is facing her toughest race yet against Dem State Senator Jennifer Wexton. VA Dems have recruited strong women candidates for several other competitive races, including former CIA Officer Abigail Spanberger in VA-7 and former Navy Commander Elaine Luria in VA-2.

Smaller State Opportunities

A few smaller states, such as Iowa, Montana, and Maine, also rank high by virtue of having one or two competitive districts.

Iowa is the most interesting case, as three of its four Congressional districts swung from Obama in 2012 to Trump in 2016. One of those districts is D-held and is considered safe, while the other two — IA-01 and IA-03 — are R-held and hotly contested. Just as in Virginia, the competitive seats are being contested by female Democrats, including yet another badass Abigail (state legislator Abby Finkenauer) as well as businesswoman Cindy Axne.

In Montana, incumbent carpetbagging rich guy/bodyslammer in chief Greg Gianforte could be vulnerable in this otherwise very red state. Democrat Kathleen Williams is behind in cash but could conceivably pull off an upset. In Maine, it all comes down to whether Dems’ strong nominee, Jared Golden, can unseat Bruce Poliquin in ME-02.

Big Blue States Have Pockets of Competition

Notably, the three biggest solid blue states — California, New York, and Illinois — do not rank in the top ten. Opportunities exist in these states, but they’re not particularly concentrated.

Part of this has to do with Dems’ existing strength. Team Blue already comprises at least 60% of each state’s delegation, so there is limited room to grow. Many of the opportunities in November have emerged due to the realignment of college-educated white voters (putting districts like CA-45 and IL-6 in play) or due to weak or junior Republican incumbents (IL-12, NY-19, CA-10).

These states are certainly important, and Dems must focus on maximizing their contributions to a potential Blue Wave. But the delegations of big, solidly Democratic states won’t be as profoundly altered in November as those of some smaller and swingier states.

Non-Competitive States and Republican Opportunities

Only about one in four Congressional seats are considered competitive. So it’s not surprising that 20 states have no expected change in their Congressional representation, including some relatively large states like Massachusetts and Tennessee. Some races in these states may become more competitive before Election Day. But this is a disturbing demonstration of the effects of gerrymandering, ideological sorting, and partisan polarization: many states simply don’t have competitive districts.

The only states with a (very slight) expected shift toward Republicans are New Hampshire and Nevada. Retirements are the driving factor: Democrats Carol Shea-Porter (NH-1), Ruben Kihuen (NV-4) and Jacky Rosen (NV-3) all occupy swing seats and are not seeking reelection (Rosen to run for Senate). Since these seats are now somewhat competitive, Dems’ expected representation in the delegation has fallen slightly. But Dems have recruited strong candidates for all three seats, and I expect them to stay in Democratic hands.

Will 2018 be the Year of the Female Democratic Governor?

Much recent punditry has focused on the potential for women’s gains in the House. And with good reason: women have been winning Democratic House primaries at unprecedented rates.

But with Janet Mills now the official Democratic nominee for governor in Maine, the stars may also be aligning for women to head to governor’s mansions in large numbers. Given the diminished slate of Democratic female governors (there are only two nationwide), female Democrats may have the most to gain.

Women in the House Update

First, a brief update: with the results of recent primaries, women are finally expected to make gains in the next House of Representatives. I’ve migrated the full database and ratings to http://www.morewomeninthehouse.com/ where I’ll make continual updates until the general election.

The current projected gains are extremely modest (1/2 a seat, in expectation), though this number should rise as primaries continue. The expected number of Democratic Congresswomen is 65.85 (a gain of about five seats from the current 61), while the expected number of Republican Congresswomen is 18.65 (a loss of 4 or 5 seats from the current 23).

As I mentioned in my last post, one thing holding women back is the high number of female incumbents vacating their House seats. Fourteen of the 84 women currently serving in the House will not seek reelection.

While those retirements limit female gains in the House, they are also part of the gubernatorial story. Two Congresswomen have already won their party’s gubernatorial nomination, Kristi Noem (R-SD) and Michelle Lujan Grisham (D-NM), while two more are leading contenders in upcoming primaries: Colleen Hanabusa (D-HI) and Diane Black (R-TN).

Governorships are Different

When considering gender, governorships are different from representative positions in two notable ways: Democratic women do not outnumber Republican women, and women’s representation has declined in recent years.

Despite the fact that Republicans control both chambers of Congress, 73% of women in Congress are Democrats. Women comprise about a third of the Democratic caucus in the Senate (16 out of 49 Senators) and the House (61 out of 193 seats). But they only comprise about 10% of Republican Senators (6 out of 51 seats) and Republican House Reps (23 out of 235).*

The same pattern doesn’t hold for governorships, where Republicans hold twice as many offices as Democrats — and about 12% of each party’s governors are women. The four female Republican Governors are Kay Ivey of Alabama, Kim Reynolds of Iowa, Mary Fallin of Oklahoma, and Susanna Martinez of New Mexico. The two Democrats are Gina Raimondo of Rhode Island and Kate Brown of Oregon. This imbalance isn’t entirely a fluke of Democrats’ ebbing political fortunes. While many of the first female governors were Democrats, more Republican women than Democratic women have served as governors in the 21st century.

The more troubling reality is that women haven’t made consistent gubernatorial gains. The plot below shows women’s representation in both Congress and Governorships over the past century.**

Female Congressional representation has tended to alternately rise and plateau since 1980. Governorships, by contrast, have fallen since a peak in the mid-2000s. At the start of Trump’s term, there were just four sitting female governors. A decade prior, there had been nine.

From a statistical perspective, this is partly a consequence of the fact that there are 10 times more Congressional seats than governorships. Hence, variation in the female proportion is likely to be larger even as women generally make gains. But there is also evidence that women tend to pursue legislative positions more frequently than executive ones, and are seen as more natural fits for the legislative role. So it’s plausible that, even smoothing out variation over time, women’s gains will be more muted.

The Current Roster

Based on the current slate of nominees, we’d expect the count of female Governors to remain roughly unchanged at six, but for the partisan composition to move toward the Democrats.

Raimondo, Ivey, and Brown are all seeking reelection and are favored to win (though Raimondo could conceivably lose the nomination to a primary challenger). Noem and Lujan Grisham are both the favorites in their open races. Reynolds and Mills are both Toss-Ups.

The three other current female nominees — Democrats Stacey Abrams (GA), Paulette Jordan (ID), and Lupe Valdez (TX) — are not considered competitive by Cook Political Report. But I’d keep on the lookout in Georgia: Abrams ran a smart primary campaign, has captured the imagination of the left, and may face a Republican damaged by scandal.

The full slate of current female nominees and ratings is given below.

Name	State	Party	Status	Cook Race Rating
Kate Brown	OR	Democrat	Incumbent	Likely Democrat
Gina Raimondo	RI	Democrat	Incumbent	Likely Democrat
Michelle Lujan Grisham	NM	Democrat	Open Seat, R-held	Lean Democrat
Kim Reynolds	IA	Republican	Incumbent	Toss-Up
Kay Ivey	AL	Republican	Incumbent	Solid Republican
Stacey Abrams	GA	Democrat	Open Seat, R-held	Solid Republican
Kristi Noem	SD	Republican	Open Seat, R-held	Solid Republican
Lupe Valdez	TX	Democrat	Incumbent Challenger	Solid Republican
Janet Mills	ME	Democrat	Open Seat, R-held	Toss-Up
Paulette Jordan	ID	Democrat	Open Seat, R-held	Solid Republican

 

Will the Stars Align?

The key to large female gains – particularly female Democratic gains – will be upcoming primaries. And unlike races for the House, women have not been dominating Democratic gubernatorial primaries. Competitive women have lost the Democratic nomination to male candidates in states including Iowa, Alabama, and Nevada. There is no clear signal that women will triumph in upcoming primaries either.

Among Democrats, Hanabusa is the likeliest to win the general election, but only if she successfully primaries incumbent Democratic governor David Ige. Some early polls showed her ahead, but polls appear to be tightening as the scandal over Ige’s botched handling of an erroneous missile alert fades. The smart money still appears to be on Hanabusa, but it’s no sure thing.

Former Senate Minority Leader Gretchen Whitmer in Michigan and former Florida Congresswoman Gwen Graham are also in competitive primaries for governor. Both find themselves in the unenviable position of the “establishment” candidate. Whitmer is probably likelier to prevail, though she is in a tight race with well-funded businessman Shri Thanedar (who has led some polls) and progressive favorite Abdul El-Sayed.  Graham is running near the head of the pack in Florida, but most recent polls have shown Miami Beach mayor Philip Levine in the lead. If either Whitmer or Graham gets through the primary, she will face a Toss-Up race.

In Minnesota and Colorado, women are also competitive in Democratic primaries. The Minnesota race was thrown into chaos by the late entry of MN Attorney General Lori Swanson, who declared her candidacy only three weeks ago. State Representative Erin Murphy is also running, and won the endorsement of the DFL (Minnesota’s version of the Democratic party). Before Swanson entered, Congressman Tim Walz was considered the frontrunner, and there hasn’t been any subsequent polling. But I’d still bet on Walz to win the nod in the end, and advance to a Toss-Up race. Colorado State Treasurer Cary Kennedy is facing off with Congressman Jared Polis for the nod to succeed termed-out Democratic Governor John Hickenlooper. But Polis, among the wealthiest members of Congress, is spending heavily, and has led in the available polls. If Kennedy somehow triumphs, she’ll be favored to win in November, but I wouldn’t bet on it.

Democratic women are competing elsewhere, but appear highly unlikely to make it through both the primary and the general. Polling data indicates that female candidates are far behind in primaries in Wisconsin and Oklahoma. And while Cynthia Nixon’s challenge to Andrew Cuomo in New York may excite the base, she still isn’t drawing much support. Female candidates have good chances of winning the Democratic nods in Kansas, Vermont, Wyoming, and New Hampshire — but all these races are rated either Likely or Solid Republican by Cook Political Report.

Among Republicans, the field of potential nominees is much narrower. Black is the most significant contender, as she is leading most polls for the Tennessee Republican nomination, and would be heavily favored to win the general. Rhode Island House Minority Leader Patricia Morgan is also seeking the Republican nomination to take on Raimondo, although 2014 nominee Allan Fung is generally seen as the frontrunner.

Takeaways 

The number of female-held governorships has declined since the mid-2000s, with the ranks of Democratic women particularly decimated. However, those trends may reverse in 2018.

Female Democrats are likely to expand the ranks of their governorships, with incumbents Kate Brown (OR) and Gina Raimondo (RI) likely to win reelection, and Michelle Lujan Grisham favored in New Mexico. Large numbers of Democratic women are competing in upcoming gubernatorial primaries. They have a good shot at winning the nomination in Hawaii, Michigan, and Florida as well as an outside shot in Colorado and Minnesota. Along with Maine nominee Janet Mills, this could yield up to six additional female Democratic nominees in competitive governor’s races in 2018.

Republican women are less likely to increase their count of governorships from the current four. Incumbent Kay Ivey (AL) is favored to win reelection, and current nominee Kristi Noem (SD) and probable nominee Diane Black (TN) will likely win as well. Incumbent Kim Reynolds (IA) is in a Toss-Up race, and few other Republican women are likely to receive nominations.

* There are currently seven vacant seats in the House, so these numbers add up to 428 instead of 435. I also do not count nonvoting delegates. 

** To smooth the governor’s line, I show the maximum number of women who concurrently served as governor in any given year. The Congress line includes only the number of women elected at the start of each Congress, and ignores special elections. Data was sourced here. 

How many women will be in the House of Representatives in 2019?

The New York Times recently put out a great report highlighting the challenges of electing more women to the House of Representatives. Their thesis boiled down to a few key points:

• Many Democratic women are running for the House this year, and they appear to be winning primaries at disproportionate rates.
• A lot of these women are challenging entrenched incumbents in heavily Republican districts, and are likely to lose the general election.
• An abnormally high number of women in the current House are not seeking reelection: 14 out of 84. As a result, even if women triumph in quite a few races this fall, their numbers in the House may barely budge.

The Times put together some nice infographics about the state of the primaries as of May 12. I’ve put together a simple tracker to keep updating the projection of the expected number of women to serve in the next House. To compute expected values, I’m using the current Cook Political Report Rating of the seat along with the average win probabilities associated with those ratings, computed here.

Based on these ratings and nominee selections from primaries that have already occurred, we would expect 77 women to serve in the next House, seven fewer than today. However, there are two processes that will increase this expected value:

• More women will be selected in upcoming primaries. I am not including women who are heavily favored to win nominations in competitive seats, like Mikie Sherill in NJ-11 or Lisa Brown in WA-5, because those women have not been formally nominated yet.
• Ratings will shift as we head into November. If ratings continue to shift toward Democrats, this will typically increase the expected number of women, as Democrats have many more female nominees.

It is noteworthy that despite all the recent women triumphing in primaries, we’re nowhere near an expected increase in the number of Congresswomen. To see why, let’s look at each of the four types of candidates: incumbents, incumbent challengers, those vying for open seats currently held by their own parties, and those vying for open seats held by the other party.

Incumbents

Of the 84 women currently serving in the House, 70 are seeking reelection: 53 Democrats and 17 Republicans. Among the Democrats, virtually everyone is safe (the only two marginally contested seats are Stephanie Murphy in FL-7 and Annie Kuster in NH-2).

But among the Republicans, more than half are in competitive races. Six are rated either Toss-Up (Claudia Tenney in NY-22, Barbara Comstock in VA-10) or Leans Republican (Mimi Walters in CA-45, Karen Handel in GA-6, Cathy McMorris Rodgers in WA-5, and Mia Love in UT-4).

While some of these women will face female Democratic challengers, it still yields an interesting dynamic: Democrats could triumph in the fall partially by decimating the ranks of the (typically moderate) female Republicans in swing districts. There are currently 2.65 times as many Democratic women serving in the House as Republican women, and this ratio may grow even more lopsided in 2019.

Party	Seat Rating	# of Women	Win Prob	Expected Value
Dem	Safe Democrat	51	100.00%	51
Dem	Likely Democrat	2	99.00%	1.98
Dem	Lean Democrat	0	89.20%	0
Dem	Toss-Up Dem	0	48.50%	0
Rep	Toss-Up Rep	2	48.50%	0.97
Rep	Lean Republican	4	89.20%	3.57
Rep	Likely Republican	3	99.00%	2.97
Rep	Safe Republican	8	100.00%	8
			Total:	68.49

Incumbent Challengers

Here, we see the most dramatic reason that women aren’t currently favored to pick up seats: there are 41 women currently nominated by their party to face an incumbent Congressperson, but 31 of the challengers (28 D, 3R) are in races currently rated Safe for the incumbent party. It’s exceedingly unlikely that these women will ultimately triumph in November.

Only a small handful of women, all Democrats, have been nominated in races where they have a good chance of defeating an incumbent. The only seat rated as a toss-up is TX-7, where Lizzie Pannill Fletcher will be the Democratic nominee. Five Democratic women are challenging incumbents in races rated Lean Republican: Lauren Underwood in IL-14, Kathy Manning in NC-13, Kara Eastman in NE-2, Amy McGrath in KY-6, and Gina Ortiz Jones in TX-23.

The result: of these 41 women, the current expected number to serve in Congress next January is just one.

Party	Seat Rating	# of Women	Win Probability	Expected Value
Rep	Safe Democrat	3	0.00%	0
Rep	Likely Democrat	0	0.99%	0
Rep	Lean Democrat	0	10.81%	0
Rep	Toss-Up Dem	0	51.49%	0
Dem	Toss-Up Rep	1	51.49%	0.51
Dem	Lean Republican	5	10.81%	0.54
Dem	Likely Republican	4	0.99%	0.04
Dem	Safe Republican	28	0.00%	0
			Total:	1.10

Open Seats: Home Challengers

The news is a little better among women nominated for seats held by their own party, where the incumbent is not seeking reelection. Of these seven women (6D, 1R), most are expected to win. Pennsylvania’s newly redrawn Congressional maps — and the fact that a number of women triumphed in competitive Pennsylvania primaries — has been particularly beneficial.

There are three female Democrats in seats rated Safe D: Madeline Dean (PA-4), Veronica Escobar (TX-16), and Sylvia Garcia (TX-29). Three other Democratic women, all in Pennsylvania, are in either Likely or Lean seats: Mary Gay Scanlon (PA-5), Susan Wild (PA-7), and Chrissy Houlahan (PA-6).

The lone Republican woman likely to pick up a seat is Carol Miller in WV-3, but she is facing a very spirited Democratic challenger in Richard Ojeda. This race might be one to watch.

Party	Seat Rating	# of Women	Win Probability	Expected Value
Dem	Safe Democrat	3	100.00%	3.00
Dem	Likely Democrat	2	100.00%	2.00
Dem	Lean Democrat	1	94.44%	0.94
Dem	Toss-Up Dem	0	62.50%	0
Rep	Toss-Up Rep	0	62.50%	0
Rep	Lean Republican	0	94.44%	0
Rep	Likely Republican	1	100.00%	1.00
Rep	Safe Republican	0	100.00%	0
			Total:	6.94

Open Seats: Away Challengers

Among the women challenging for open seats currently held by the other party (6D, 1R), virtually none are expected to win. The lone Republican is Pearl Kim in PA-5, who will likely fall to Mary Gay Scanlon. The only Democrat in a remotely competitive seat is Bibiana Boerio in the newly redrawn PA-14, but this is still rated Likely Republican.

The other five Democratic women — Cristina McNeil (ID-1), Tobi Beck (IN-4), Jeannine Lake (IN-6), Susan Palmer (OH-16), and Jana Lynn Sanchez (TX-6) — are all competing in races rated Safe R.

Party	Seat Rating	# of Women	Win Probability	Expected Value
Rep	Safe Democrat	0	0.00%	0
Rep	Likely Democrat	1	0.00%	0
Rep	Lean Democrat	0	5.56%	0
Rep	Toss-Up Dem	0	37.50%	0
Dem	Toss-Up Rep	0	37.50%	0
Dem	Lean Republican	0	5.56%	0
Dem	Likely Republican	1	0.00%	0
Dem	Safe Republican	5	0.00%	0
			Total:	0