In an off-year long on election commentary but short on actual elections, the two main events on a Spartan political calendar are now upon us: New Jersey and Virginia will elect new governors next week, and the stakes are high, particularly for Democrats.
As we have argued before, the only way the Democrats can be judged to have a good night is if they sweep both governorships. Republicans already hold a massive 34-15 advantage in state governorships (there is one independent, Gov. Bill Walker of Alaska). The two races decided next Tuesday are currently split — Democrats hold Virginia, Republicans hold New Jersey. A flip-flop, with Democrats winning the Garden State but Republicans capturing the Old Dominion, would represent no net gain for Democrats, and a maintenance of the net gubernatorial status quo would represent a win for Republicans. A double win by Republicans, which is difficult to fathom, would be a triumph, particularly given the Republicans’ difficulties right now (more on that in a second).
A Democratic sweep, meanwhile, would represent progress for the minority party, although Republicans could say that Democrats had an advantageous environment in both states. They would be correct, but the challenge for Republicans is that such an advantageous environment could very well be present again in many places this time next year, where far more is at stake and Republicans have far more to lose, and Democrats far more to gain.
The president often hinders his party in an off-year election, particularly if that president is unpopular, and President Donald Trump looks about as weak now as he has at any point in his still-young term. His approval ratings have dipped below 40% in all three major polling averages (FiveThirtyEight, HuffPost Pollster, and RealClearPolitics), and these same three averages show a Democratic edge in the House generic ballot of between six to 10 points, around the kind of numbers that might suggest a Democratic wave next year and a flip in party control of the House. We’ve got a long way to go until the midterm, but 2017’s elections are in a few days, and the environment for Republicans is poor.
Monday’s bombshell indictment of Trump’s former campaign chairman, Paul Manafort, also seems poorly timed for Republicans heading into these elections, although we’re not sure how much it affects these races. Similarly, it’s also difficult to say if Tuesday’s lone-wolf terrorist attack in New York City changes anything.
An upset in New Jersey by Lt. Gov. Kim Guadagno (R) over former Ambassador to Germany Phil Murphy (D) is hard for anyone to imagine.
However, an upset in Virginia by former Republican National Committee Chairman Ed Gillespie (R) over Lt. Gov. Ralph Northam (D) is not difficult at all to fathom. Gillespie has run a nasty, tough campaign against Northam, blanketing the airwaves with ads arguing that Northam and the current Virginia Democratic administration restored voting rights to a sex offender and that Northam backs sanctuary cities (something Virginia doesn’t have), thus unleashing scary gangs in the suburbs. From the left, Latino Victory Fund (an outside group) produced an ad portraying a Gillespie backer chasing nonwhite children in a truck (LVF quickly took the ad down and the money behind it was a relative pittance, but it has generated a lot of headlines and Republicans hope to use it as a rallying cry in the final days).
Democrats argue that Gillespie’s ads have ended up hurting the Republican’s favorability, but while Gillespie’s tougher campaign is hardly a tribute to high-minded civics, that does not necessarily mean it will not be effective, and politics is rightly or wrongly a game measured through wins and losses. Carter Eskew, an Al Gore campaign veteran writing in the Washington Post, echoed a lot of the chatter we’ve heard and read about the potential effectiveness of Gillespie’s messaging versus Northam’s.
The Northam campaign has not hit Gillespie as hard, and if the Democrat loses, there will be a lot of painful “what ifs” for Northam and his team to consider in the coming weeks, months, and years. Indeed, in the event of a loss, the comparisons to Hillary Clinton’s 2016 national disaster will run rampant in Democratic circles.
Polls in the commonwealth have been all over the place. While most show a Northam lead, the spread is huge, ranging from a Hampton University poll showing Gillespie up eight points to a Quinnipiac University poll showing Northam up 17. If anything, one cannot accuse the pollsters of “herding” together at the end: There are going to be at least some pollsters who finish far from the eventual margin. Steve Shepard of Politico has noticed that many pollsters surveying Virginia have used polls based on calling people on voter lists, as opposed to random digit dialing calling a larger universe of people. The voter list polls, which mimic the techniques used by campaigns, find a narrower range of horse race predictions, from Gillespie by one to Northam by seven. That’s roughly the range of the internal campaign polls we’ve heard about throughout the race.
The RealClearPolitics average puts Northam’s lead at about 3.5 points — not big enough to consider him more than a modest favorite, and only then because the small polling lead may be reinforced by the generic advantages any Democrat might have in this race (meanwhile, if Clinton was president, Gillespie might be the one favored right now, for reasons we get into in more detail below). Like in his narrow loss to Sen. Mark Warner (D-VA) three years ago, Gillespie may close hard at the end, although our sources on both sides do not indicate that there has been much movement in this race either way for weeks, and the bulk of the evidence continues to suggest a small Northam lead. But a Gillespie win would not be as surprising as Trump’s own national victory was a year ago.
To get ready for next week, we’ve identified five numbers, or sets of numbers, that we’ll be watching next week in the gubernatorial contests and in some other races down the ballot. After the results come in, we’ll report back to you with what these numbers tell us.
1. Will Gillespie buck the typical anti-White House party pattern in Virginia?
Rightly or wrongly, Virginia’s gubernatorial election is now viewed as one of the first indicators of how voters are responding to a president. This is obviously the product of being one of the two regularly-scheduled statewide elections in the year following a presidential election, meaning it takes place roughly 10 months after Inauguration Day. While the commonwealth’s gubernatorial election has been hit-or-miss in forecasting the vote in the succeeding federal midterm, it has usually gone for the party outside of the White House. From 1977 to 2009 — nine elections — the out-of-power party won the governorship in Virginia. Though that streak came to an end in 2013 when now-Gov. Terry McAuliffe (D) defeated then-state Attorney General Ken Cuccinelli (R), the result didn’t upset a long streak of the president’s party performing worse in a gubernatorial contest than in the presidential election the year before — at least based on the two-party vote. The reality is, for the past 40-plus years Virginia voters have usually shifted away from the president’s party in the gubernatorial race, even when the president’s party won the governorship.
The 1969 gubernatorial election is often viewed as the start of the modern period of competitive two-party politics in Virginia. Although Republicans had captured Virginia’s electoral votes in four of the previous five presidential elections, with 1964 as the lone exception, no Republican had been elected to statewide office. But in 1969, Linwood Holton (R) became the first out-and-out Republican to ever be elected governor of Virginia. In that cycle, Holton out-performed Richard Nixon’s (R) 1968 percentage in Virginia by about nine points (52.5% versus 43.4%). However, the 1968 election featured George Wallace’s third-party bid for president, which complicates a one-to-one analysis. One way to overcome this is to just look at the two-party vote to look solely at Democratic and Republican votes. Moreover, we know from polls at the time that Wallace voters largely preferred Nixon in a two-party race against Hubert Humphrey (D), so Nixon’s result likely would have been more favorable without Wallace in the race (Wallace won 24% in Virginia in 1968). By the two-party vote, Holton won 3.5 fewer points than Nixon did in 1968. From there, as Table 1 indicates, the same pattern in the two-party vote has continued through the 2013 election: the nominee of the presidential party always performs worse in the succeeding gubernatorial election.
Table 1: Comparative performance by presidential party in Virginia gubernatorial elections
Notes: *Indicates that a major third-party candidate won at least 10% of the vote in the presidential election. ^Indicates that while was technically no Democratic nominee for governor in 1973, then-Lt. Gov. Henry Howell (I) had previously been a Democratic state senator and the Democratic State Central Committee “commended” Howell to the voters in his election against former Gov. Mills Godwin (R), a former Democrat.
Source: Virginia Department of Elections
Over the 12 pairs of presidential and gubernatorial elections in this period, the presidential party’s nominee has, on average, performed about nine points worse in the two-party vote in the gubernatorial contests, though the decrease has ranged from 18.5-points worse for Mills Godwin (R) in 1973 to just 0.6-points worse for McAuliffe in 2013. It should be noted that Godwin, who had previously been elected governor as a Democrat in 1965, faced off against Lt. Gov. Henry Howell (I), who had previously been a Democratic state senator before winning a 1971 special election for the lieutenant governorship as an independent. It was a transitional period for party affiliation in Virginia, but Howell was “commended” by the Democratic State Central Committee, essentially telling Democratic voters to back him. Besides the 1968-69 and 1972-73 elections, the only other set of elections that isn’t cut-and-dry is the 1992-1993 pairing. In 1992, independent Ross Perot won 12% of the vote in Virginia. Partly because of Perot’s success, state Attorney General Mary Sue Terry narrowly out-performed Bill Clinton in raw vote percentage (by 0.3 points), but she did six points worse in the two-party vote while losing badly to then-former Rep. George Allen (R).
Gillespie will be seeking to end this streak of worse performances by the president’s party. President Trump won 47.2% of the two-party vote in Virginia, so that will be the number Gillespie must best to end it. For what it’s worth, an average of the last five polls puts Northam ahead 49%-44%, which would work out to about to about 53%-47% in the two-party vote. Given the closeness of the race, Gillespie may be able to out-perform Trump, particularly if he can get close to Trump’s margins in rural Virginia while improving on Trump’s percentages in the state’s three major metropolitan areas. Democrats also argue that the smaller off-year Virginia electorate is less friendly to them than a presidential-year one is, which gives Gillespie an opportunity even as the overall environment is unfavorable to him right now.
2. Will Phil Murphy outrun Hillary Clinton?
Unlike Virginia, which has seemed close and competitive for the entire election even as Northam has led the lion’s share of polls, New Jersey has seemed like an easy Democratic takeover the whole cycle. Phil Murphy, a wealthy Democrat who served as ambassador to Germany in Barack Obama’s first term, has held a towering lead in polls over Lt. Gov. Kim Guadagno (R), and national Republicans have not seemed all that interested in making a run at defending this open seat. A big part of Guadagno’s trouble is unpopular term-limited Gov. Chris Christie (R), who despite winning a smashing 22-point victory in 2013 now has approval ratings in the teens.
New Jersey’s Democratic lean is also a challenge for Guadagno, although the state actually hasn’t been that Democratic for very long. Prior to 1996, the state regularly voted a little bit more Republican than the nation in presidential races, although typically only by a few points. More recently, it has become reliably Democratic, most recently giving Hillary Clinton a 14-point margin of victory, a margin about a dozen points higher than her national performance.
Compared to Virginia, New Jersey’s gubernatorial results don’t seem quite as historically tied to the White House party. Still, over the past four presidencies before Trump (George H.W. Bush, Bill Clinton, George W. Bush, and Barack Obama), the non-presidential party has won each of the seven gubernatorial elections. The Democrat, Murphy, is in an excellent position to extend that streak to a fifth president, Trump.
It may be an unfairly high bar for Murphy, but given all of his advantages in the race, we’ll be judging his performance on whether he can match or exceed Clinton’s 14-point margin. His lead in polls is a little bit above that right now (about 16 points, according to RealClearPolitics).
3. How will state legislative performance compare to last year’s presidential results?
Part of the reason we’re using the 2016 presidential race as a measuring stick in both New Jersey and Virginia is because both gubernatorial races are open-seat contests, so there’s no strength of incumbency that can make it harder for the challenging party. That makes these races somewhat similar to a larger group of elections we and others have been tracking so far this cycle: special legislative elections across the country. These include the handful of special congressional races, but the list is mostly composed of special state legislative races.
There have been 42 such races so far, and the Democrats have outperformed Hillary Clinton’s 2016 margin in 31 of them, often by a lot: their average improvement is about 11 points, and their median improvement is close to 12 points. This is a reversal from four years ago, when Democrats routinely ran behind Barack Obama’s 2012 showing by similar amounts, as RealClearPolitics’ Sean Trende found in an overview of special elections and the New Jersey and Virginia state legislative slate in December 2013. This did end up being a bad sign for lousy Democratic turnout in the 2014 midterm. That doesn’t mean Democratic overperformance is predictive this time, but it is worth monitoring.
The state legislative elections in New Jersey and Virginia, with a couple of major caveats, will provide us with many more data points. All 100 Virginia House of Delegates seats are up (all 40 state Senate seats are elected two years from now), and all 80 New Jersey General Assembly and 40 state Senate seats are up. So we can compare the results with the 2016 presidential results and see if the pro-Democratic trend continues.
Here are the caveats, though. For one thing, not every race features two-party competition. In Virginia, just 60 of the 100 races feature candidates from both parties, although that represents a high water mark for two-party House of Delegates competition over the last two decades. New Jersey is different in that nearly every district features two-party competition. In the Senate, Democrats have a candidate in all 40 races and Republicans do in 37. In the General Assembly, Democrats have a full slate of 80 candidates in the 40 districts (each New Jersey district elects two General Assembly members and one senator), while Republicans are fielding a full slate in all but two districts, where they are fielding just one candidate apiece. So we’ll throw out the races that don’t feature full two-party competition next week when making comparisons to 2016.
Additionally, and perhaps more importantly, the specials held so far and the regular state legislative elections in New Jersey and Virginia differ in another important way: While specials, as a rule, do not feature elected incumbents, almost all of the New Jersey/Virginia races with two-party competition do feature incumbents next week. Given that incumbents usually perform at least a little bit better than one might otherwise expect — U.S. House incumbents won roughly three points more of the vote (or about six points in margin) than they otherwise might have been expected to because of an incumbency advantage in 2016, according to FairVote — comparing open-seat specials to incumbent-held regular elections may be a little misleading.
The special elections we’ve seen so far don’t have elected incumbents running in them, whereas the lion’s share of the New Jersey and Virginia races do: Of the 60 Virginia races with two-party competition, 54 feature incumbents. In New Jersey, 34 of the 37 two-party contested Senate races have incumbents, and in the General Assembly, the 38 districts featuring a full slate of candidates feature 67 incumbents (New Jersey data from Ballotpedia).
Party control in either of the New Jersey chambers or the Virginia House of Delegates is not at stake, although Democrats are hoping to make gains in both places (they already control the New Jersey legislature comfortably but are in a big deficit in Virginia).
Also next Tuesday, there will be 18 more special state legislative elections outside of New Jersey and Virginia (plus one congressional special in Utah) featuring one Democrat and one Republican. These are not all technically open seats — four races in Washington state feature appointed incumbents — but none feature elected incumbents. We’ll be looking for trends in these races, too.
The single most important state legislative election in the whole nation is taking place in Washington state. If Democrats win the state Senate race in SD-45 — a true open seat that the appointed incumbent, former Senate and gubernatorial nominee Dino Rossi, is leaving to run for an open U.S. House seat — they will take control of the Washington state Senate. The district voted for Clinton by a punishing 65%-28% score, although Republicans do better down the ballot.
If Murphy wins in New Jersey and Democrats win that Washington Senate special, they will take unified control of the governments of both states. That would give them eight so-called “trifectas,” where they control the governorship and both houses of the state legislature, according to Ballotpedia. Republicans, who dominate so much of state government across the country, already hold a whopping 26 trifectas and could add a 27th if Gillespie wins in Virginia. The remaining states have some form of divided government.
4. How much variation will there be in the Virginia’s three statewide races?
With high levels of political polarization making straight-ticket voting very common, a straight-ticket result where the same party sweeps the governorship, the lieutenant governorship, and the attorney generalship seems fairly likely. Still, if the Gillespie-Northam race is quite close, it’s possible that Virginia could have a split-ticket outcome, with both parties winning at least one of the commonwealth’s three statewide offices. If neither side has a major coattails advantage, it could happen. Table 2 lays out the two-party election results for governor, lieutenant governor, and attorney general from 1969 to 2013. It also shows the standard deviation in the election results: A lower standard deviation indicates that the results are close to the party’s average result while a higher standard deviation indicates that the results are spread out over a wider range.
Table 2: Two-party election results for Virginia’s three statewide offices and variation in performance
Notes: ^Indicates that while was technically no Democratic nominee for governor in 1973, then-Lt. Gov. Henry Howell (I) had previously been a Democratic state senator and the Democratic State Central Committee “commended” Howell to the voters in his election against former Gov. Mills Godwin (R), a former Democrat.
Source: Virginia Department of Elections
Most recent elections have had only small variations in performance for the three candidates from each party. The 2013 cycle saw a slightly higher standard deviation than 2005 or 2009, in part because Northam led the Democratic ticket in vote share while defeating a weak GOP opponent for lieutenant governor. Still, he only won 55.3%-44.7% in the two-party vote, a far cry from a more-traditional landslide win like the 20+ point margins put up by Mary Sue Terry (D) in the 1989 attorney general’s election and Jerry Kilgore (R) in the 2001 attorney general’s race. Northam’s 2013 party mates, McAuliffe and state Attorney General Mark Herring, both won by narrower margins, with Herring winning the closest statewide election in Virginia’s history (by just 0.01 points). But the deviation from the Democratic mean of 52% or so was still fairly small compared to many previous gubernatorial cycles. This is mainly because six of the 10 elections between 1969 and 2005 saw each party win at least one office, a product of ticket-splitting by voters. But odds are that the standard deviation in the 2017 results will be fairly small. Most polls that have asked about the lieutenant governor and attorney general races have found results that are in line with the gubernatorial topline. Nonetheless, if things are quite close, we could still see a mixed outcome.
Should there be a split-ticket result, one thing to watch for is whether or not Herring’s incumbency will matter. Because governors cannot run for immediate reelection in Virginia, incumbency in state-level offices is only an occasional feature of elections. Many times, the lieutenant governor and/or attorney general can opt to run for the top office instead of seeking reelection. For what it’s worth, no incumbent has ever lost a general election in Virginia. In the modern period of Virginia elections (1969 on), incumbents are four for four — Attorney General Andy Miller (D) in 1973, Terry in 1989, Lt. Gov. Don Beyer (D) in 1993, and Lt. Gov. Bill Bolling (R) in 2009. So Herring’s opponent, John Adams (R), will be fighting history on Nov. 7.
5. What are some Virginia trends to watch out for on Election Night?
Polls close at 7 p.m. EST and some localities in the state should start reporting around 8 p.m. or so. Typically, the earliest areas to report are mostly rural, Republican-leaning parts of the state, so expect Gillespie to hold an early edge. However, one early-reporting county to particularly watch is the fifth-largest locality in the state by population: Chesterfield County, a suburban county south of Richmond. It’s traditionally a GOP-leaning area, but in recent times it has become more competitive. In 2013, Cuccinelli won Chesterfield by about nine points in the two-party vote (Libertarian Robert Sarvis won 9.5% there, though). In the 2014 U.S. Senate election, Gillespie also won Chesterfield by nine points while narrowly losing statewide to Sen. Mark Warner (D). But in 2016, Trump only carried the county by two points as Chesterfield, like many other suburban areas of the state, moved away from Republicans. Watch to see whether Gillespie can get the GOP margin in Chesterfield closer to the high single-digit edge of 2013 and 2014 — if he does, Gillespie will probably be in a position to win or at least come very close.
In recent elections, Fairfax County — at about 14% of Virginia’s population, it’s by far the state’s largest locality — has typically been one of the later-reporting parts of the state. This is particularly true of its large absentee voting precincts. On Election Night in 2016, Trump still led in Virginia late into the evening, only to suddenly fall behind when Fairfax’s remaining precincts reported. A little more than one-fifth of the county’s votes were absentee or provisional votes, and together those 121,368 votes were 3% of the statewide vote (outside of Fairfax County itself, only five other localities in the state had more votes cast than Fairfax’s combined absentee and provisional ballots). By margin, Fairfax County’s absentee and provisional votes were 12 points more Democratic than the Election Day votes there. So if the gubernatorial contest is quite close with much of the commonwealth’s vote reporting, watch out for the late Democratic hammer in Fairfax to put Northam over the top.
More broadly, an interesting question is just how much Virginia’s 2016 shifts are replicated in 2017. That is, will the map look more like the 2013 gubernatorial and 2014 Senate races, or more like the 2016 presidential election? Maps 1 and 2 show the change in the two-party vote margin for Virginia localities from 2013 to 2016 and from 2014 to 2016, respectively. The maps are cartograms, with each county and city in the state weighted by its average share of the statewide vote for the 2013-2016 period. They show the striking change in margin in different parts of the state, particularly Southwest Virginia (trending sharply Republican) and Northern Virginia (trending Democratic).
Maps 1 and 2: Change in two-party vote margin by Virginia locality, 2016 presidential election compared to the 2013 gubernatorial election and the 2014 U.S. Senate election
Source: Virginia Department of Elections
However, there are a couple of notable differences between the two maps. Perhaps most obvious, Fairfax County (the largest blue blob in Northern Virginia) is a darker shade of blue in the second map. This reflects the fact that Gillespie did better in Fairfax and other parts of Northern Virginia, so Clinton’s advantage in that region was larger compared to Warner’s in the 2014 Senate race than it was compared to McAuliffe’s. At the same time, much of Southwest Virginia is a darker shade of red in the second map because Warner did better in that part of the state and other areas outside of the “Urban Crescent” — Northern Virginia, Greater Richmond, and Hampton Roads — than Clinton (or McAuliffe in 2013). That means that Trump’s margins were much larger in some rural areas of the state than Gillespie’s in 2014. Together, these two differences are important to 2017: Will Gillespie be able to hold down Democratic margins in Northern Virginia? Will Northam not get absolutely crushed in much of rural Virginia? To some extent, the 2017 result may see a slight retreat from 2016-level margins in places like Northern Virginia and Southwest Virginia, but the extent to which Democratic gains in urban and suburban areas and Republican gains in rural areas are preserved will help determine the outcome of the Gillespie-Northam contest. Some polling suggests that Gillespie may not win by as crushing a margin in parts of the state that shifted very hard toward Trump in 2016, such as Southwest Virginia, but also show Northam possibly falling short of Clinton’s levels in areas of the state where she surged, most notably Northern Virginia. The specifics will help determine the winner.
To conclude, we present four tables of data about where the Virginia vote came from by region in the 2013, 2014, and 2016 election cycles, and where the parties were both strongest and weakest.
Table 3: Two-party vote margins in the Urban Crescent and the rest of Virginia
Table 4: Share of the two-party statewide vote by region
Table 5: Share of the statewide Democratic vote by region
Table 6: Share of the statewide Republican vote by region
Note: Calculations based on Virginia Department of Elections data. Metropolitan areas based on the Office of Management and Budget’s 2015 definitions for the Richmond, VA MSA, the Virginia Beach-Norfolk-Newport News, VA-NC MSA, and the Washington-Arlington-Alexandria, DC-VA-MD-WV MSA.
Table 3 lays out the regional two-party vote margins for the 2013 gubernatorial election, the 2014 U.S. Senate election, and the 2016 presidential election. The last two columns compare the two earlier elections with the presidential cycle, providing regional data that helps explain the changes shown in Maps 1 and 2 discussed above. As Table 3 shows, Gillespie only lost Northern Virginia by 11.1 points in 2014, better than Cuccinelli’s 16.5-point loss in the 2013 two-party vote and far better than Trump’s 27.6-point loss last November. Taking advantage of his appeal in Northern Virginia to limit Democratic margins will once again be vital to Gillespie’s chances. Meanwhile, Northam will want to replicate Clinton’s margins in the Urban Crescent as best he can while reducing her margins outside of the three major metropolitan areas. Whereas McAuliffe lost in the rest of Virginia by 20.2 points and Warner lost by just 16.9 points, Clinton lost by 24.8 points.
In Table 4, the shares of the vote from each region and the rest of the commonwealth are presented for the 2008 to 2016 period. Northern Virginia’s vote share has grown over time, but it has tended to decrease slightly in gubernatorial elections compared to the presidential contest the year before. Gillespie will be hoping for something in line with 2008-2009 drop off while Northam will prefer 2012-13. Also note the higher percentages for Hampton Roads in the 2008 and 2012 cycles compared to the other years. Very likely, those high marks can be attributed to higher African-American turnout due to Barack Obama’s candidacy as a plurality of the commonwealth’s black voters live in that region. Minority turnout usually falls to some degree in off-year Virginia elections, making it a challenge for Northam.
Tables 5 and 6 show where the parties tend to garner most of their votes (or fewer of them). Unsurprisingly, Democrats are most reliant on Northern Virginia. But note the large increase in 2016 compared to 2013 and 2014 in that region. The DC suburbs and exurbs made up 42% of Clinton’s vote in 2016, making her even more reliant on the region than McAuliffe or Warner. On the other side, Trump drew a larger share of his votes from the areas outside of the Urban Crescent than Cuccinelli or Gillespie did. The extent to which the 2016 shifts are maintained in 2017 will be something to keep an eye on.
If the 2017 election looks more like 2013 or 2014, that would suggest that the state’s polarization has taken at least a small step back from the heights (or lows) of 2016. But if 2017 looks more like 2016, that may suggest that the urban-rural divide in the Old Dominion will continue to be quite pronounced going forward.