Sabato's Crystal Ball

Exit Stage Left or Right: Midterm Retirements and Open Seats in the House From 1974 to 2018

Republicans already have many abandoned districts to defend this November. Are there more to come?

Geoffrey Skelley, Associate Editor, Sabato's Crystal Ball March 22nd, 2018



— By multiple measures — such as retirements, “pure” retirements, and open seats that must be defended — this cycle’s GOP has one of the highest levels of exposure in the U.S. House of any presidential party dating back to 1974.

— Republicans already have more retirements than any presidential party in a midterm cycle from 1974 to 2018, and they are not far behind Democrats in 1978 in terms of the open seats they hold. Because seats lacking an incumbent are more difficult for the incumbent party to retain, this situation should deeply worry the GOP.

— Earlier filing deadlines relative to past cycles mean that fewer GOP House members still have to decide if they are running for reelection relative to past presidential party House conferences. While additional retirements or surprise open seats might occur after filing deadlines have expired, the range of possible retirements might run from just a handful to the double digits.

Midterm retirements and open seats in U.S. House elections from 1974 to 2018

In the aftermath of now-Rep. Conor Lamb’s (D) special election victory on March 13, a constant refrain has been the stated fear among Republicans that the result would precipitate more retirements among GOP members in the U.S. House. As the Crystal Ball has noted in the past, open seats held by the president’s party in midterm elections have typically seen large average swings toward the opposition, making retirements a serious concern for the party in the White House. Because of the strength of incumbency, political parties have a more difficult time retaining a seat it controls when its incumbent does not seek reelection. That is, “seat maintenance” becomes harder for the incumbent party, in part because it now has to defend an exposed seat. Looking ahead to this November, the number of additional Republican retirements could be a critical factor in determining whether the GOP maintains its majority in the House. As a result, we wondered the following: How bad is the GOP retirement picture compared to past midterm cycles going back to 1974, and how many additional Republican retirements might occur this cycle?

This article seeks to explore those questions using data on House exits from 1974 to 2018 (through March 21). I gathered information about members who left the House, why they exited, and when they left. I used a number of sources, primarily Roll Call’s “Casualty Lists” and old newspapers, though I utilized other sources such as CQ Weekly, too.

There are two aspects to consider when trying to account for House members and their choice to leave Congress’ lower chamber. From an electoral standpoint, the most important measurement is the number of open seats in November, which will have no incumbent defending them. For our purposes, the most vital figure is the number of open seats held by the president’s party. There can be some confusion in counting such seats, however. For example, sometimes members have announced their retirement, only to resign or die prior to the regular November election, leading their state to call a special election to fill the seat, which usually produces a new incumbent who seeks election in November. However, in other cases, that seat might remain vacant until November, maintaining its status as an open seat. Additionally, an open seat could occur when an incumbent loses a primary, which one cannot account for by just looking at retirements. So not every open seat is the result of a retirement. In this article, the definition of an “open seat” is any House seat lacking an incumbent in the regular federal November election.[1]

Another consideration is the nature of what “retirement” means. Some House members really do ride off into the sunset, uninterested in seeking election to another office. However, others are retiring to run for another elected position, often a higher office such as U.S. senator or governor. In those cases, members running for another office are not truly retiring, per se. For this article, a “pure” retirement refers to someone who retired and did not seek another office; this also includes those who withdrew from their House reelection bids prior to the general election but after their filing deadline. Otherwise, “retirement” or “overall retirement” will generally refer to sitting House members who decided to exit the House, whether to run for another office or not. Any member who retired but then officially left office prior to the election (e.g. because of death, resignation due to scandal or to focus on a campaign for higher office, etc.) is included in the retirement total even if a special election occurred in the interim between the member’s exit and the November election. So not every retirement results in an open seat. I will examine both open seats and retirements to see how the GOP’s 2018 situation compares to past midterms and to speculate how many additional retirements might occur based on those data.

The timing of retirements and open seats

A good starting point for judging how the 2018 cycle compares to past midterm cycles and getting some idea of how many more members will retire (particularly Republicans as the presidential party) is to examine when retirements occurred in the election calendar. These data show that 2018 is already set to be a cycle with one of the highest rates of House turnover among recent midterm cycles.

Chart 1 below shows the number of House members who retired, starting 600 days before Election Day up to 75 days prior.[2] The dates are based on official retirement announcements, as best could be determined, or the date a member officially announced a run for another office. Some complications — such as incumbents with self-imposed term limits, a lack of a firm retirement announcement, or seemingly no official campaign announcement — do exist in the data, but the dates chosen reflect my best judgment as to what was appropriate.

Chart 1: Overall retirements over time in midterm cycles, 1974-2018

Notes: *Indicates a redistricting cycle. Data as of March 21, 2018. These data include all House members who did not seek reelection to the House by choice. It also includes some retirees who officially left office before Election Day but well after their retirement announcements. These members typically left early because of death or resignation, with the latter often occurring because of health problems, scandal, or a desire to concentrate on an ongoing election campaign for another office. Click on the chart to enlarge the image.

Sources: Roll Call’s “Casualty Lists”; assorted newspaper articles from LexisNexis, NewsBank,, and ProQuest Historical Newspapers; CQ Weekly.

In terms of retirements — “pure” or running for other offices — 2018 is already a record-setter for midterms dating back to 1974. As of this writing (Wednesday, March 21), 52 members from both parties have opted against reelection bids, surpassing the previous midterm record of 49 in 1978.[3] March 21 marked the 230th day prior to Election Day 2018, and the 2018 cycle’s 52 retirements thus far dwarf the 42 that had occurred in 1978 by the same point in the cycle. However, a lurking variable in this comparison is state filing deadlines, which in recent years have collectively become earlier (more on that below).

Electorally speaking, open seats are probably the more important count than merely retirements. After all, seats can be open for other reasons besides retirement (and as mentioned above, not all retirements end up leading to open seats). Moreover, the incumbent party of an open seat matters. Chart 2 repeats the same design as Chart 1 but with open seats instead of retirements and running from 600 days to 25 days before Election Day.[4]

Chart 2: Open seats over time in midterm cycles, 1974 to 2018

Notes: *Indicates a redistricting cycle. Data as of March 21, 2018. These data include seats that had no incumbent running in them on Election Day. The open seat count also includes a few instances in which an incumbent’s name was still on the ballot in November despite the fact that the incumbent was not actually running. Example: In 2006, Rep. Mark Foley (R, FL-16) resigned in scandal on Sept. 29, but his name remained on the November ballot, with votes cast for him going to a replacement GOP candidate. The 2018 data do not include the former seat of Rep. Louise Slaughter (D, NY-25), who died recently. It is unclear if Gov. Andrew Cuomo (D-NY) will call a special election to fill her old seat before November. Click on the chart to enlarge the image.

Sources: Roll Call’s “Casualty Lists”; assorted newspaper articles from LexisNexis, NewsBank,, and ProQuest Historical Newspapers; CQ Weekly.

Looking at open seats, the 2018 cycle does not have the highest total among midterm cycles going back to 1974 — at least not yet. Both 1978 and the redistricting cycle of 1982 had more open seats (58).[5] At this point, 54 seats will be open in November 2018. However, that number is likely to grow. Only two states, Illinois (March 20) and Texas (March 6) have held primaries, so there will be additional chances for incumbents to lose renomination. Additionally, many states’ filing deadlines are months away, leaving time for incumbents to opt out of reelection bids. Still, 2018 is close to the midterm high of 58, and the 2018 total does not include a possible open seat in New York: Rep. Louise Slaughter (D, NY-25) died on March 16 and Gov. Andrew Cuomo (D-NY) has yet to determine if he will call a special election to fill Slaughter’s old seat or leave it vacant until November.

One other thing to note is that the open seat count is complicated in redistricting cycles (and in some mid-cycle redistrictings). The timing of “new” open seats (i.e. truly new districts or reworked ones that had no sitting incumbent who might have run in them) in Chart 2 is based on the date redistricting plans in states with new districts became law. This includes two seats that I count as new in Pennsylvania in the aftermath of its court-ordered redistricting in February.[6]

The president’s party, retirements, and open seats

Having looked at the timing and total number of retirements and open seats, we turn to figures for the president’s party. In a midterm environment, open seats held by the incumbent presidential party tend to be the most vulnerable to flipping, and retirements tend to create most open seats (though not all). So how do the retirement and open seat numbers in 2018 for the GOP compare to the presidential party in past midterms?

Chart 3: Overall retirements by the president’s party in midterms, 1974 to 2018

Notes: *Indicates a redistricting cycle. Data as of March 21, 2018. Click on the chart to enlarge the image.

Chart 3 above shows the overall retirement figures for U.S. House members of the presidential party in midterm elections since 1974. Out of 52 retirements in 2018 so far, 36 retirees have been Republican. This figure is the highest for any presidential party, topping the 31 retirements that Democrats had in 1978 during Jimmy Carter’s presidency as well as the 28 for Democrats in 1994 during Bill Clinton’s White House tenure. Many of these Republican retirements have opened up seats that Democrats are already at least slightly favored to win (e.g. CA-49, FL-27, and NJ-02), have a reasonable chance of winning (e.g. CA-39, NJ-11, and WA-08), or could potentially put in play in a wave environment (e.g. FL-06, NM-02, and TX-21).

Chart 4: Open seats held by the president’s party in midterms, 1974 to 2018

Notes: *Indicates a redistricting cycle. Open seats in redistricting cycles do not include newly-created seats. Data as of March 21, 2018. Click on the chart to enlarge the image.

Chart 4 tracks the open seats held by the president’s party from 1974 to 2018. So far, all 36 GOP-held open seats in 2018 have come via retirements (either “pure” or running for another office), which differs from the 1978 cycle for Democrats. In that election, Democrats had to defend 39 open seats, 31 of which came by retirements, five that came via incumbents losing renomination, and three because of incumbent deaths. If one discounts death — a mostly unpredictable fate — Republicans in the 2018 cycle already have to defend as many open seats as Democrats had to in 1978 without any primary losses for GOP House members taken into account (or any other unforeseen circumstances).

In Chart 4, one complication is the redistricting cycles of 1982 and 2002. While I could account for newly-created open seats to some extent in Chart 2, assigning a party to those seats becomes more difficult. Thus, many open seats are not assigned to either the presidential or the non-presidential party in 1982 or 2002, so the numbers for those cycles would be higher if not for redistricting.

Notably, the 2018 GOP has the highest share of overall retirements and open seats of any presidential party in a midterm going back to 1974. Table 1 presents data on the presidential party’s share of retirements and open seats. More often than not, the president’s party has more retirements and open seats to defend in midterm cycles than the opposition.

Table 1: Presidential party’s share of retirements and open seats in midterm cycles, 1974-2018

Notes: *Indicates a redistricting cycle. Open seats in redistricting cycles do not include newly-created seats. Data as of March 21, 2018.

The purpose of all these data about retirements and open seats is to emphasize the fact that the GOP has a very high number of retirements and open seats relative to previous midterms. These openings represent a challenge for Republicans in their quest to retain a House majority, and time remains for their exposure to grow.

Considering filing deadlines and how many more GOP retirements there might be

Importantly, the deadlines for House incumbents (and potential candidates) have changed over the years. With the November 2018 election 230 days away as of Wednesday, 24 states’ filing deadlines have already come and gone. At the same point in 1974, only 11 states’ filing periods had concluded. Table 2 displays data on how many states had passed their filing deadlines by the 230 days-to-go mark in midterms from 1974 to 2018.

Table 2: Filing deadlines come earlier than they used to

Notes: Data as of March 21, 2018.

Sources: Federal Election Commission documents; CQ Weekly; assorted newspaper articles from LexisNexis, NewsBank,, and ProQuest Historical Newspapers.

As Table 2 shows, the number of states with expired filing deadlines at 230 days until Election Day has essentially doubled in the past 44 years. In other words, a lurking variable in the data on retirements and open seats is increasingly earlier filing deadlines: More states require earlier filing decisions, so that means that members have to make decisions about their electoral futures earlier than they once did.

Before we speculate about how many more retirements might happen this year, we need to consider retirements as they relate to filing deadlines, not just Election Day. This becomes a little complicated because some members left races after their state’s filing deadline (e.g. withdrawing from a race, resigning, dying, etc.), but we can graph retirements that occurred prior to deadlines and leave post-deadline exits out. Chart 5 is a histogram of all midterm retirements by days before filing deadline from 1974-2014, based on 10-day periods. The 2018 cycle is not included because only about half of all states’ filing deadlines have expired.

Chart 5: Histogram of retirements by time before filing deadline in midterm cycles, 1974-2014

Notes: Grouped by 10-day bins. These data include all House members who did not seek reelection to the House by choice, but only those who announced their retirements prior to their state’s filing deadline. It also includes some retirees who officially left office before Election Day but well after their retirement announcements. These members typically left early because of death or resignation, with the latter often occurring because of health problems, scandal, or a desire to concentrate on an ongoing election campaign for another office. Click on the chart to enlarge the image.

Unsurprisingly, most retirements come closer to the filing deadline as incumbents have to decide if they are going to pay filing fees and/or gather signatures to qualify for office as well as determine if they are personally up to another campaign. Incumbents also may be weighing runs for another office. In the 1974 to 2014 era, the most common time for midterm retirement announcements came in the 29- to 59-day period, with over 20 retirements in each 10-day period within that range.[7] Retirements inside the 60-day period made up 30% of the total pre-filing deadline retirements, and those inside of the 120-day period made up 49%. Using the 60- and 120-day marks makes sense because Louisiana’s candidate filing deadline is July 20 — 121 days from this writing, and the last state on the 2018 filing calendar. A substantial share of the retirements inside of 60 and 120 days came from the president’s party because — as Table 1 shows above — at least a slight majority of total retirements in a cycle have usually come from that party.

With 24 states’ filing periods concluded, there have been 52 retirements in the 2018 cycle. Of those, 51 came before the filing deadline — Rep. Blake Farenthold (R, TX-27) withdrew his reelection bid three days after Texas’ deadline because of a sexual harassment scandal[8] — so the 51 pre-filing deadline retirements are useful for comparing 2018 to the data for 1974 to 2014 in Chart 5. So far in 2018, 11 retirements have come inside of the 60-day period (22%) and 18 inside the 120-day period (35%). Those numbers represent a smaller share for those time periods than the data in Chart 5 as a whole. But as Table 3 shows, there is a lot of variation from cycle to cycle as to how early or late retirements occur.

Table 3: Retirements relative to filing deadlines in midterm cycles, 1974-2014

Notes: *Indicates a redistricting cycle. Data include only retirements that occurred prior to filing deadlines.

On average between 1974 and 2014, roughly half of all retirements in midterms came before or after the 120-day mark, or roughly four months before filing deadlines. But in some cycles the share was relatively lopsided, particularly 2006, when three-fourths of all retirement announcements (“pure” or running for another office) occurred at least 120 days before filing deadlines.

What does all this mean for estimating additional retirements, especially Republican ones? If the 2018 cycle approximates retirements in midterms dating back to 1974, we should expect many more retirements. To get to the average 48% share for inside of 120 days, 18 more retirements prior to filing deadlines would have to occur, shifting the total to 69 total retirements, with 36 coming inside of 120 days (practically all 18 hypothetical additional retirements would fall within the 120-day period because Louisiana’s filing deadline is 121 days away). As 69% of retirements in 2018 have come from the GOP, that would mean about 12 more Republican retirements if the party retirement split continued at the same rate. However, 18 additional retirements, with roughly two-thirds being Republican, surely sounds like a large number, especially because most retirements beyond this point would likely be “pure” retirements. While it is not too late to become a candidate for higher office (e.g. the U.S. Senate or governor) in more than 20 states, it seems unlikely that many (if any) House members would make bids for other offices this late in the cycle. For example, Sen. Cory Gardner (R-CO) formally announced his Senate bid in 2014 on March 1, which most observers considered to be quite late. Still, there may be a House member pondering a bid for another office who has not officially announced it, so I would not entirely write off that possibility.

Uncertainty about such a high number of retirements means that we need to explore “pure” retirements, i.e. those where a member retires without seeking another office. With our interest in Republican retirements, it’s worth noting how many “pure” retirements tend to come from the president’s party and how many from the other party. Table 4 presents data on “pure” retirements by party for the 1974 to 2018 period.

Table 4: “Pure” retirements by party in midterm cycles, 1974-2018

Notes: *Indicates a redistricting cycle. Data as of March 21, 2018.

Whereas the presidential party has had, on average, 56% of overall retirements in midterms from 1974 to 2018, it has had an average of 62% of the “pure” retirements, with 73% for the GOP thus far in 2018. The current cycle has almost equaled the record total for “pure” exits — 31 in 1978 — and the current presidential party (the GOP) has almost as many as the presidential party did that cycle (the Democrats). As most retirements from this point forward are likely to be “pure” ones, it is quite possible that 2018 will set a new modern midterm record for both the total number of “pure” retirements and those by just the presidential party.

Table 4 also includes data for “pure” retirements prior to filing deadlines. This allows us to consider such retirements in the context of the timing data presented above in conjunction with Table 5, presented below, which lays out data on the timing of “pure” retirements using the same 60 and 120 days intervals employed above.

Table 5:  “Pure” retirements by presidential party relative to filing deadlines in midterm cycles, 1974-2014

Notes: *Indicates a redistricting cycle. Data include only retirements involving incumbents who did not seek election to another office that occurred prior to filing deadlines.

Table 5 shows that, on average, 64% of presidential party “pure” retirements came within 120 days of filing deadlines between 1974 and 2014. Another 39% came inside of 60 days. So for the most part, “pure” retirements tend to come closer to the filing deadline than overall retirements. The average number of “pure” exits to occur with 120 days left before filing deadlines is about eight for the president’s party. At this point in 2018, 11 of the GOP’s 21 pure retirements have come within 120 days of the state’s filing deadline (though some of those are in states where the filing period has not yet expired), a 52% rate. If the 2018 cycle were to run close to the 40-year midterm average for “pure” retirements announced inside of 120 days, we would expect roughly another seven such retirements by GOP incumbents prior to filing deadlines. If that were to happen, the GOP would have 18 “pure” retirements (64%) within 120 days of filing deadlines out of 28 total. Obviously, 28 total “pure” retirements would be a record for the president’s party since 1974.

However, another consideration is the fact that a large share of the Republican conference in the U.S. House already has decided whether to file for reelection: Only 100 of the 240 seats (42%) that the GOP controls (or most recently controlled, as in the case of vacant AZ-08 and OH-12) are in states where filing deadlines have not expired. So if the GOP is to have a number of additional “pure” retirements, they will likely come from states with sizable Republican delegations and many potentially competitive seats, laid out in Table 6:

Table 6: States with outstanding candidate filing deadlines and Republican-held seats

Notes: *“Competitive” indicates that a seat is rated as Toss-up, Leans Republican, or Likely Republican in the Crystal Ball’s U.S. House ratings.

Table 6 notes the number of Republican incumbents who have not announced their retirements, the number of seats held by these incumbents that the Crystal Ball rates as competitive (Toss-up, Leans Republican, or Likely Republican) as of March 21, and the share of the remaining GOP-held seats that are competitive. Among the states to keep an eye on are Florida, Michigan, Minnesota, New Jersey, New York, Virginia, and Wisconsin. At least one of two (if not both) things are true for these states: They have many Republicans in their House delegations and/or a large number of competitive seats featuring Republican incumbents.

While past midterm exits by members of the president’s party prior to filing deadlines might suggest a sizable number of additional Republican retirements could occur, the fact that much of the GOP conference has already decided its fate may limit additional retirements (though some could change their minds post-deadline). At this point, it seems reasonable to expect a few more Republican retirements. However, as laid out in this piece, different data points suggest varying potential retirement numbers, and the fact that so many states’ filing deadlines have already passed may reduce any estimate. Of course, this is all speculation: Although retirements involve the consideration of factors such as electoral competitiveness, one’s political future, and the strength of the opposition, potential retirees also may weigh more idiosyncratic factors such as family and health. However, while it’s hard to predict what incumbents might do before the filing deadline, it’s even harder to predict retirements that come after filing deadlines via candidate withdrawal. Many midterm cycles have had at least one such withdrawal. One possibility in 2018 might be Rep. Ryan Costello (R, PA-06), who has been pondering retirement in the wake of court-ordered redistricting that made his seat more Democratic. Costello filed his petitions on Tuesday for reelection but, as the Philadelphia Inquirer notes, he “has not committed to actually running for reelection.” We’ll just have to wait and see.


By multiple measures — such as retirements, “pure” retirements, and open seats that must be defended — this cycle’s GOP has one of the highest levels of exposure in the U.S. House of any presidential party in a midterm dating back to 1974. Republicans already have more retirements than any presidential party in a midterm cycle from 1974 to 2018, and they are not far behind Democrats in 1978 in terms of the open seats they hold. Seats lacking an incumbent are more difficult for the incumbent party to retain, so these data points should worry the GOP. Past midterm data suggest that additional Republican retirements probably will occur between now and Election Day. However, earlier filing deadlines relative to past cycles mean that fewer GOP House members still have to decide if they are running for reelection. While additional retirements or surprise open seats might occur after filing deadlines have expired, the range of possible retirements might run from just a handful to the double digits. As Republican incumbents weigh their political futures, the large number of GOP retirements and open seats are additional evidence — on top of President Trump’s poor approval rating, the Democratic lead in the generic ballot, and the Democratic overperformance in state and congressional special elections — that Democrats may be able to win back the House in November.


1. The open seat count does not include concurrent special elections in November to fill vacant House seats for the remainder of the expiring term.

2. The latest retirement in a midterm cycle occurred in 1994, when Rep. Dean Gallo (R, NJ-11) withdrew from his reelection bid 77 days prior to Election Day due to his struggle with prostate cancer. The GOP replaced him on the ballot with then-state Assemblyman Rodney Frelinghuysen (R), who won the 1994 general election. Coincidentally, Frelinghuysen announced his retirement from the U.S. House in January 2018.

3. As defined here, the retirement record in a presidential cycle occurred in 1992, when 66 members did not seek reelection. Redistricting combined with the House banking scandal to boost voluntary exits that cycle. The 1992 cycle was also the last time eligible House incumbents could keep campaign money for personal use following retirement. That remarkable benefit almost certainly encouraged some retirements, whether members admitted it or not.

4. The latest a seat opened up in a midterm cycle occurred in 1978, when 49-year-old Rep. Goodloe Byron (D, MD-06) tragically died while out on an evening run on Oct. 11, 1978, 27 days before the general election. His wife, Beverly, replaced him on the ballot. She won in November and held MD-06 until losing renomination in her 1992 primary. Oddly enough, Byron’s death came only one day after Rep. Ralph Metcalfe (D, IL-01) passed away from a heart attack at the age of 68. Metcalfe was a four-time Olympic medalist in sprinting, including at the 1936 Olympics in Berlin as a part of the United States’ 4×100 meter relay team, which included the famed Jesse Owens.

5. As defined here, the open seat record in a presidential cycle occurred in 1992 with 91 seats, or about one-fifth of the House. Redistricting and retirements combined to generate this very high number of open seats. See Footnote 3.

6. To remain consistent with how I handled data for other cycles, Rep. Bob Brady (D, PA-01) is counted as a retirement but his seat is not counted as an open seat because the redistricting plan dismantled the old PA-01. In another universe where Brady did seek reelection, he likely would have run in the new PA-02, which Rep. Brendan Boyle (D) is contesting. In states that underwent redistricting in a given cycle, I checked local news reports, hometowns, and the old and new district maps to attempt to situate incumbents who did not seek reelection in the appropriate district.

7. The first 10-day period is Day 0 to Day -9, the second Day -10 to Day -19, and so on.

8. Farenthold reportedly may resign his seat before November to avoid an ethics investigation. If he does, there might be a special election to replace him prior to the regular general election. However, Texas already held its primary on March 6, with both parties’ primaries going to primary runoffs on May 22. Presumably, the eventual major-party nominees would be interested in running in a special election. This scenario might also create the circumstances whereby TX-27 would no longer be an open seat because a new incumbent would be running in November having been elected in a special beforehand. But Farenthold would still count as a retirement as defined in this article.

Illinois Primary: Rauner Is Reeling

The Land of Lincoln’s gubernatorial contest now Leans Democratic

Geoffrey Skelley, Associate Editor, Sabato's Crystal Ball March 22nd, 2018


Gov. Bruce Rauner (R-IL) endured a difficult night on Tuesday. Although he won his party’s primary to earn a reelection shot in November, the contest in some ways confirmed his overall weakness as the most endangered incumbent Republican governor facing the voters in 2018. As such, the Crystal Ball is moving the Illinois gubernatorial contest from Toss-up to Leans Democratic, the first time in the 2018 cycle that we have rated an incumbent U.S. senator, U.S. House member, or governor as an underdog for reelection. Rauner’s vulnerabilities are two-fold: his party base is not solidly behind him — the GOP primary made this abundantly clear — and he faces an energized Democratic Party in what is typically a blue state.

Table 1: Crystal Ball ratings change

Map 1: Crystal Ball gubernatorial ratings

In the Republican primary, Rauner narrowly fended off a challenge from state Rep. Jeanne Ives, who few initially expected to endanger the incumbent’s renomination chances. However, by primary day stories and rumors abounded that Rauner might actually lose — and he very nearly did. Rauner escaped with a three-point win, 51.4% to 48.6%, over Ives, who gained support during her campaign from GOP megadonor Richard Uihlein, among others. The incumbent governor infuriated many conservatives last year when he signed into law legislation passed by the Democratic-controlled state legislature that increased the range of public funding for abortion services in Illinois to include individuals on Medicaid and participants in the state employee insurance program. Ives attacked Rauner from the right, running now-infamous ads that many called bigoted and transphobic. Even the state Republican Party chairman called for her to stop running them. Yet Ives clearly found some success as she nearly defeated an incumbent governor whose personal wealth provides a bottomless campaign war chest, some of which Rauner used to attack Ives by portraying her as a lemming of state House Speaker Michael Madigan (D).

Meanwhile, the Democratic primary electorate nominated its own billionaire businessman, J.B. Pritzker, to be the party’s standard-bearer in November. In some ways, Pritzker (D) might actually be the Democrat that Rauner would most prefer to face. Pritzker showed up on old wiretap recordings talking to defrocked ex-Gov. Rod Blagojevich (D), who the state legislature impeached and removed from office in 2009 following federal corruption charges. Rauner has attempted to push the Pritzker-Blagojevich connection in ads. On top of this, some Democrats do not exactly get warm and fuzzy feelings when thinking about billionaires, and now their party has one as its nominee. Nonetheless, Pritzker won a solid plurality of the vote over state Sen. Daniel Biss and businessman Chris Kennedy, son of former Sen. Robert Kennedy (D-NY).

With our ratings change, Pritzker starts the general election campaign with an edge over Rauner. On top of Rauner’s intraparty problems, primary turnout also augured poorly for the incumbent. As many analysts noted, Democrats showed their energy and engagement on Tuesday. With competitive gubernatorial primaries on both sides and an open primary system that permits Illinoisans to vote in either party primary, Democrats accounted for 64% of the gubernatorial primary vote and Republicans just 36%. While primary turnout is not in and of itself a great predictor of general election results, the impressive Democratic turnout is notable because of the state’s open primary system and because in the last competitive gubernatorial primary in 2010 Democrats had a much smaller 54%-46% turnout edge.

With two extremely wealthy major-party nominees and few campaign finance limitations for self-funders, the Illinois contest could end up being the most expensive gubernatorial contest in U.S. history. At the starting gun, Pritzker holds an early advantage.

Are Republicans in More Special Trouble?

Assessing upcoming House specials in Arizona and Ohio after Lamb’s upset in Pennsylvania

Kyle Kondik, Managing Editor, Sabato's Crystal Ball March 15th, 2018



— An apparent Democratic takeover in a western Pennsylvania House district that President Trump won by 20 points is an embarrassing setback for Republicans.

— The Republicans’ poor special election performances in general, combined with other factors such as the president’s low approval rating and a Democratic lead on House generic ballot polling, suggest the GOP House majority is in considerable danger.

— Republicans remain favored in two pending special elections, AZ-8 and OH-12, but we are downgrading their chances in both. We also are moving three other races in Ohio from Safe Republican to Likely Republican.

— That said, the AZ-8 and OH-12 specials have key differences from PA-18 that may make them easier for Republicans to defend.

Table 1: Crystal Ball House ratings changes

Conor Lamb’s (D, PA-18) apparent victory in a special House election in western Pennsylvania isn’t as much of a disaster for Republicans as Sen. Doug Jones’ (D-AL) win in a special election in Alabama was in December. But both results, combined with bigger-picture factors, suggest widespread Republican weakness as the midterm draws closer.

Unlike Jones’ Senate seat, which Democrats will hold through at least 2020, Lamb’s newly-won House district is disappearing. A new Pennsylvania map reconfigured western Pennsylvania, and Lamb is likely to run as an incumbent against another incumbent, Rep. Keith Rothfus (R, PA-17), in a suburban Pittsburgh district that President Trump only won by less than three points. In contrast, Trump won the old PA-18, the seat Lamb won on Tuesday night, by 20 points. Only one House Democrat, Rep. Collin Peterson (D, MN-7), holds a seat that voted more strongly for Trump than PA-18. The Rothfus-Lamb battle — if that in fact is the matchup — should be very competitive, although while we are keeping it as a Toss-up, Lamb eventually might have an advantage given his strong performance Tuesday and the overall environment. It is also worth mentioning that Lamb himself will have to get through a primary first in PA-17: contenders include attorney Beth Tarasi (D), who plans to run to Lamb’s left on guns and other issues, as well as Erin McClelland (D), who was the Democratic nominee against Rothfus in 2014 and 2016.

State Rep. Rick Saccone (R), Tuesday night’s narrow loser, seems likely to run in the new PA-14, a district that is even more Republican than the one he lost. It will be interesting to see if he draws credible primary or general election opposition: There is a Democratic bench even in the new district, although we rate it Safe Republican to start. Saccone, who was badly outraised by Lamb, was certainly not a great candidate, but he also was no Roy Moore (R), the disastrous Republican nominee in Alabama who was a fringe candidate even before credible accusations emerged of his disturbing behavior toward teen girls.

We’ll know more about the Pennsylvania House races after the filing deadline passes next Tuesday.

Had Saccone won, there’s little doubt that Trump would have taken credit for boosting him with a late weekend rally, and some Republicans, including now-dutiful Trump supporter Paul Ryan (R, WI-1), the speaker of the House of Representatives, credit the president for closing the gap at the end. It is possible that Trump’s late appearance helped Saccone make up some late ground. On the other hand, if Trump’s approval rating wasn’t at just a little over 40% nationally, Saccone wouldn’t have been in such trouble to begin with. A candidate less skilled than Lamb may not have won, but Trump’s unpopularity opened the door to this district being competitive. Now, does that mean that every single Republican-held district Trump won by 20 or less is now in play? Of course not. For one thing, many of those districts will have incumbents running in the fall, and many of them will have Democratic challengers less skilled and less well-funded than Lamb. That said, the House playing field is large and getting larger: With this week’s ratings changes, we now have 75 GOP-held House seats rated in competitive categories (Likely, Leans, or Toss-up).

Democrats have cut the number of House seats they need to flip in November to take control from 24 to 23, although that assumes Lamb is renominated and wins reelection this fall in a Toss-up race. We now have 214 seats rated at least Safe/Likely/Leaning Republican, 197 Safe/Likely/Leaning Democratic, and 24 Toss-ups.

We’ll now wait to see if Saccone’s loss has any ripple effects. We suspect national Republican groups spent so heavily there so they could avoid a public relations blow that might push even more GOP incumbents to decide against reelection. The Republicans already have a sizable number of vulnerable open seats, and that list may grow.

The other specials: Looking ahead

With the PA-18 special now in the books, the House now only has three vacancies, seats previously held by ex-Reps. Trent Franks (R, AZ-8), John Conyers (D, MI-13), and Pat Tiberi (R, OH-12).

All will hold special elections later this year, although MI-13’s will be held concurrently with the November election, which means Conyers’ former constituents will have gone almost a year without a congressman (Conyers resigned under a cloud of sexual harassment allegations in December, and the remaining weeks of his term will be filled in the November “special”). And about 55,000 MI-13 residents will also go months without a state senator as well because of a similarly-delayed special election for a vacant state Senate seat. There’s no doubt MI-13 will elect a Democrat: Hillary Clinton won the majority-black seat 79%-18% in 2016.

The other two House vacancies, GOP-held seats AZ-8 and OH-12, will be filled on April 24 and Aug. 7, respectively. Given some of the big Republican underperformances in House specials this year, both could feature an unusual (for them, anyway) level of two-party competition, although there are also reasons to believe that Republicans may have an easier time holding these seats than they did PA-18.

Part of the reason is that these two districts have different political lineages from PA-18. Before the PA-18 election, Republican Party of Pennsylvania Chairman Val DiGiorgio argued that PA-18 is a “Democrat district,” even though Donald Trump won it by 20 points. DiGiorgio’s comment inspired groans from many observers, but his observation is true “from a certain point of view” (to borrow a line from Star Wars’ Obi-Wan Kenobi).

Democrats do retain a voter registration edge in PA-18 even though the previous incumbent, Rep. Tim Murphy (R), had a history of big victories (his lowest vote share was 58% in 2006). But as recently as 2000, southwestern Pennsylvania was exclusively represented by Democrats in the House, according to RealClearPolitics’ Sean Trende, and a different district in the region elected Mark Critz (D) to replace longtime Rep. John Murtha (D) in a closely-watched special in 2010. That held no predictive value for November 2010, when Republicans netted 63 House seats and the majority.

To get a sense of the Democratic lineage of PA-18, consider Washington County. About three-quarters of Washington is in PA-18, and the county voted Democratic for president in every election from 1932 through 2004 (with the exception of Richard Nixon’s 1972 national blowout). But it flipped from John Kerry in 2004 to John McCain in 2008, and by 2016 it gave Trump a nearly 25-point countywide victory. Washington County’s story is similar in both Appalachian western Pennsylvania and across the border in eastern Ohio, and the Republican shift in these places (which began before Trump) accelerated during Barack Obama’s presidency, helping Trump capture both Ohio and Pennsylvania. But just because this region has become Republican at the presidential level does not mean it is uncompetitive for Democrats with the right candidate in a good environment, as Lamb’s victory showed (he lost the Washington County portion of the district by just seven to Saccone).

The history in AZ-8 and OH-12 is different: Neither district has much recent Democratic lineage.

AZ-8 is based entirely within Phoenix’s Maricopa County, the fastest-growing county in the whole country (its population is more than 4 million total and has roughly doubled over the last three decades). Maricopa remains one of the only big counties that still votes Republican at the presidential level, but that edge may be eroding: Trump won Maricopa by just under three points, although Republicans have generally won it by double digits since 1952 (though Bill Clinton only lost it by about three points in 1996). AZ-8 covers Maricopa’s West Valley, including Glendale, where the home of the Arizona Cardinals and college football’s Fiesta Bowl, University of Phoenix Stadium, is located. Trump won AZ-8 by 21 points, down a little from Romney’s 25-point win.

Franks’ direct predecessor in the House was long-serving Rep. Bob Stump (R), who actually was a Democrat when first elected in 1976, although he was a conservative Democrat (otherwise known as a “pinto” Democrat in local parlance). Stump became a Republican in 1981 and kept easily winning reelection. Meanwhile, Stump’s district — which back at the time of his party switch covered much more territory than just Glendale and other western Phoenix suburbs — gave Ronald Reagan about 73% of the two-party presidential vote in 1980, so Stump fit right in. The current district has nearly 80,000 more registered Republicans than Democrats, a far cry from the roughly 24,000 voter registration edge for Democrats in PA-18.

In other words, while PA-18 and AZ-8 voted very similarly for president (the former by 20 points, the latter by 21), their partisan makeups and histories are significantly different.

The April 24 special general election will feature former state Sen. Debbie Lesko (R) and physician Hiral Tipirneni (D). Republicans avoided a headache when Lesko won the special primary amidst a crowded field that included former state Sen. Steve Montenegro (R), whose campaign was rocked by a story that he had been exchanging inappropriate text messages with a staffer. Lesko doesn’t have Montenegro’s baggage although she has had to deal with some campaign finance questions. Because of Republicans’ persistent underperformance in special elections across the country this year, we’re moving AZ-8 from Safe Republican to Likely Republican. We don’t expect we’ll have to go further but we’re curious to see if Republicans have to sound the alarm over this district too, particularly because Republicans have avoided close calls in only one of six previous Republican-held special House elections this cycle so far (Republicans’ only double-digit victory was in UT-3, which now-Rep. John Curtis won easily in November).

OH-12, which has its special primary concurrent with Ohio’s regular primary election on May 8 followed by an Aug. 7 special general election, is potentially more interesting, and large primary fields on both sides will battle for their respective party nominations. Trump did significantly worse in OH-12 than in either AZ-8 or PA-18, but the district is more similar to AZ-8 in terms of its reliable Republican history.

Ohio has a unique party registration method in that the state doesn’t really have formal party registration at all, but rather the state determines whether someone is affiliated with a party based on which primary ballot a person takes in the most recent primary election in which they participated. Based on this, Republicans have a huge advantage statewide based on far greater turnout in their 2012 and 2016 presidential primaries (in 2012, the Democrats of course didn’t have a contested presidential primary, and in 2016 there was much more interest in the Republican presidential primary than the Democratic one — the Ohio Secretary of State’s office produced some helpful information relating to statewide party affiliation following the 2016 primary). So for what it’s worth, Republicans have a more than two-to-one affiliation advantage in OH-12, although more than half of its registered voters are unaffiliated.

A better gauge than party affiliation of the district’s inherent Republicanism involves a closer look at the counties that comprise it. About half of the district’s population lives in three counties located northeast of Franklin County (Columbus): Delaware, Licking, and Morrow. None of them have voted Democratic for president since at least 1964, and Delaware hasn’t voted Democratic since 1916, the longest Republican-voting streak of any county in Ohio. That said, Delaware is also the most affluent and educated county in Ohio, and it’s also the fastest-growing. It’s also one of only four (of 88 total) counties in Ohio where Donald Trump actually performed worse than Mitt Romney (to put that in context, Trump improved 11 points on Romney’s statewide margin in Ohio, turning a three-point loss into an eight-point win in a traditional bellwether state). Overall, OH-12 voted for Trump by about 11 points and Romney by 10, so it hardly budged at all from 2012 to 2016. That’s probably because the district is more highly educated than the national average (40% of residents over 25 have a four-year college degree, compared to about 30% nationally), and white college-educated voters are significantly more skeptical of Trump than white non-college voters. If there’s a parallel to PA-18, it would be if the areas of the district closest to Columbus swing heavily to the Democratic nominee as a protest against Trump, much like areas of PA-18 did on Tuesday. It’s this possibility that prompts us to move the rating from Likely Republican to Leans Republican. To be clear, Lamb ran better than Clinton did in PA-18 throughout the district, but he only won the district’s Allegheny County portion, where a strong showing and high turnout allowed him to overcome a deficit in the rest of the district.

Tiberi, the now-former congressman, won a version of this district in 2000 after then-Rep. John Kasich (R) retired (Kasich is now finishing up his second term as Ohio governor). Kasich first won the seat in 1982 with an assist from redistricting, and he was one of the few GOP success stories in an otherwise bad year for Republicans nationally and in Ohio — he beat then-Rep. Bob Shamansky (D), who himself was something of an anomaly as a first-time Democratic House winner amidst the 1980 Reagan Revolution election. At the time, Shamansky became the first Democrat to represent Columbus in Congress in four decades, although there have been others since then: former Rep. Mary Jo Kilroy (D) won OH-15 for a single term in 2008 but was defeated by now-National Republican Congressional Committee Chairman Steve Stivers (R) in 2010, and Columbus effectively has its own representative now in a safe Democratic seat, Rep. Joyce Beatty (D, OH-3), who was elected in 2012. She represents about 60% of Franklin County, with the rest divided between the vacant OH-12 and Stivers’ OH-15. If there is to be an upset in OH-12 — or OH-15 in the fall (more on that district below) — it likely will be led by the pieces of those districts in Franklin County, which is fast becoming a Democratic stronghold (Hillary Clinton won it by 26 points, up from Barack Obama’s 23-point win in 2012), and it soon will surpass Cuyahoga County (Cleveland) as the largest single source of votes in the state. That said, Stivers and Tiberi both won their pieces of Franklin County comfortably, although neither faced much real Democratic opposition as Clinton was simultaneously carrying their portions of Franklin.

We’re also making three other changes in Ohio, and one of them involves the aforementioned Stivers. We’re moving the open OH-16 from Safe Republican to Likely Republican, and we’re making the same changes to the districts held by Reps. David Joyce (R, OH-14) and Stivers (R, OH-15). These districts cover significant portions of suburban Columbus (OH-15) and Cleveland (OH-14 and OH-16), and the areas of these districts closest to their respective big cities have the potential to mimic the suburban portions of Pittsburgh that powered Lamb to victory on Tuesday. All three districts are also similar to PA-18 in that they have at least average four-year college attainment and above-average median income, and while the Democratic challengers in those seats are unproven, Republicans may eventually have trouble in one or more of the seats, each of which Trump won by double digits (11 points in OH-14, 16 in OH-15, and 17 in OH-16). Of the three, OH-16 is probably the most vulnerable if only because it’s an open seat (Republican Rep. Jim Renacci is running for U.S. Senate). Stivers and Joyce are quality incumbents although they could be pushed if things go further south for Republicans this fall.

In any event, the list of potentially competitive Republican seats is growing, and we suspect it will continue to grow. Hypothetically, the GOP lineage of AZ-8 and OH-12 should make them easier to hold than PA-18, and the Democratic nominees in those districts may not catch fire the way Lamb did. But Republicans are in a position where they can take little for granted.