Tag Archives: 2020 Generic Ballot

THE 2018 MIDTERMS CONTINUE: NORTH CAROLINA’S 9TH

The 2018 midterms will finally end next month. North Carolina’s Ninth Congressional District will hold a special election on September 10 for a seat that should have been decided in November 2018. Those election results were thrown out by the state’s election board due to allegations of election fraud against McCrae Dowless, a member of Republican candidate Mark Harris’s campaign staff. He was charged with improperly collecting absentee ballots, forging signatures on absentee ballots and changing or filling in votes.

The District is traditionally Republican — No Democrat has won there since 1962 and Trump carried it by 12%. It runs from southeast Charlotte eastward through more rural and majority black counties. Union County, which includes the exurbs of Charlotte, was Harris’s stronghold. He won by 20% (over 17,000 votes) in Union County which allowed him to lose both Mecklenburg County (home to Charlotte) as well as the more rural and less populated Districts to the east and still win the election.

In the (invalidated) 2018 election, Republican Mark Harris beat his Democratic opponent, Dan McCready, by just 905 votes.  Such a close election was bound to set off a competitive rematch. And so it has: Cook Political Report and Inside Elections have the race rated a “Toss Up” and Sabato’s Crystal Ball has it as “Leans Republican”.

Dan McCready, a marine veteran and renewable energy businessman, is again the Democratic nominee. He ran unopposed in the primary and is running as a moderate — disavowing some of the party’s more progressive figures and vowing not to support Nancy Pelosi for speaker in the 2018 race. On the Republican side is Dan Bishop, a state senator most famous for sponsoring the controversial “bathroom bill” to require transgender people to use public bathrooms that align with the sex indicated on their birth certificates.  Also on the ballot: Libertarian candidate Jeff Scott and Green Party candidate Allen Smith.

Polling has been scarce and hasn’t told us much about who is leading. The only external poll in the race found Bishop with a 46% to 42% lead over McCready. This poll, though, was taken back in May and found 10% of voters were undecided and had a 5.2% margin of error. The only other poll is from McCready’s camp and had the two candidates tied at 46% with 8% of voters and a 4.6% margin of error. These two polls don’t give us much indication of a favorite. They both show a tight race and have margins of error and enough undecided voters that the race could go either way.

The current polling doesn’t tell us much about who is winning. So, what other indicators do we have?

Good Signs for Democrat Dan McCready:

  • McCready had no primary opponent. Primaries can cause intraparty schisms and drain candidates’ bank accounts, hurting them in the general election. McCready was able to glide through the primary season while his opponent, Dan Bishop, was battling it out with fellow Republicans.
  • McCready has been campaigning for nearly 28 consecutive months. He announced his 2018 campaign in May of 2017. His opponent, on the other hand, has been running for 5. McCready has had ample time to get his name recognition up, meet voters and spread his message.
  • McCready has dominated Bishop in fundraising. As of July 30, McCready had outraised his opponent $3.4 million to $1.2 million. McCready also had $1.8 million left in the bank compared to Bishop’s $340 thousand. As I wrote in a previous post, candidate fundraising in the general election usually doesn’t actually change the race that much. But it will allow McCready to stay on the air through election day, which could tip the scales in a very tight race. More importantly, though, it is an indicator of grassroots support and heightened enthusiasm.
  • The district has an urban(ish) core and lots of black voters. The district did vote for Trump by 12 points in 2018, but is the archetypal district that is swinging towards Democrats. It comprises a good portion of southeast Charlotte and its suburbs, meaning there are likely plenty of suburban voters — the kind we saw swing towards Democrats in 2018, handing them the House. And while the district stretches eastward into rural counties, these counties have high proportions of black voters, a core Democratic constituency.
  • Lastly (and in my opinion an underrated factor) is that Democrats did not commit election fraud. The Republican in 2018 received national bad press for tainting the election process, something Americans on both sides of the aisle view as sacrosanct. It’s not hard to imagine this recent Republican betrayal driving Democrats to turn out and Republicans to stay home.

Good Signs for Republican Dan Bishop:

  • This is historically a Republican leaning District. Both Trump and Romney carried it by 12 points. No Democrat has won the seat since 1962. A 12 point swing in three years is a huge jump and a lot of things would have to go wrong for Bishop and right for McCready for it to happen.
  • 2018 was a banner year for Democrats. They won the House popular vote by 8.6% and netted 41 House seats. But that may have passed. And if the national mood of the country is not as friendly for Democrats, it will be tough for them to win in a district that Trump carried by 12%. According to FiveThirtyEight’s 2020 poll aggregator, Democrats are currently leading in the generic ballot by about 6%. Special elections, though, are hard to predict and don’t always align with the national environment. But the possibility of a less Democratic national mood is a good sign for Bishop.
  • Outside spending is heavily weighted in Bishop’s favor. While McCready has dominated in direct campaign contributions, Bishop has the weight of the NRCC and Congressional Leadership Fund behind him. The two groups have reserved around $3.8 million in ad spots through election day. The DCCC pledged to spend $2 million on the race. The $1.8 million gap effectively neutralizes McCready’s advantage in campaign fundraising and may even be more effective because outside groups are often more amenable to drafting negative ads.
  • This is an off-year election and off year elections have lower turnout. Generally, low turnout elections benefit Republicans because their base of older white voters are more likely to turn out than younger minority voters who lead Democratic.
  • Dan Bishop avoided a runoff in the Republican primary, allowing the general election to happen in September rather than in November. Charlotte’s municipal elections take place in November and will likely turn out Democratic voters in the city. Bishop will benefit from the lack of overlap between the municipal and congressional elections.

The winner of this seat is not going to affect the current House power balance. Democrats hold 235 seats and Republicans effectively hold 1991There is a vacancy in NC-03 due to the death of Representative Walter Jones, but this is a heavily Republican district and expected to elect a Republican in the September special election. After this election the House balance will be either 236-199 or 235-200 — an insignificant difference. More importantly, the winner will have the incumbency advantage in 2020 and a good chance of holding onto the seat. Depending on how close the House race is next year, one extra seat could be meaningful.

And perhaps even more importantly, the results will tell us a few things about the current national environment. If McCready wins the election — or even if he loses but the race is much tighter than Trump’s 12-point margin — Democrats can breathe a bit easier. This would mean that the blue wave of 2018 has not completely receded and may stay through 2020. On the other hand, if Republicans again carry the seat by close to 12 points, they may have a better shot at winning back the House than the generic ballot indicates.

Lastly, the election is a testing ground for the parties’ 2020 messaging. Dan Bishop has been attacking Democrats as socialists, focusing his jabs against progressives like Ilhan Omar, Alexandria Ocasio Cortez and Bernie Sanders. McCready has run as a moderate — distancing himself from those same progressive figures and party leaders like Nancy Pelosi. His campaign slogan is “country over party”.

We are one month from election day and the race is up in the air. The polls show a tight race and no other indicators show a clear frontrunner. So, without any clear indicators, I’m moving from clear, hard data to opaque, soft guesswork. The previous Republican’s dabbling with election fraud caused this special election and left the district without congressional representation for nine months. This betrayal of America’s most fundamental institution — free and fair elections — leaves voters with a bad taste.

The election is still up in the air, but election fraud is, understandably, not popular. This could cause Republicans to stay home and drive Democrats to the ballot box. Election fraud is not a good look for Republicans and could give McCready the win in this traditionally Republican district. 

THE GENERIC BALLOT: SEAT ALLOCATION VS THE POPULAR VOTE (PART 2)

The 2020 elections are still 16 months away and yet pollsters are out in force, giving us just enough information to break out our crystal balls and make wildly irresponsible predictions. This is the last post in a four-part series looking at the Generic Ballot and its utility as an election predictor. You can read the first post in the series here

The Republican Party’s structural electoral advantage in federal elections is well documented in liberal circles. Republicans won two of the last five presidential elections while losing the popular vote. The Senate — where states get equal representation — gives rural, Republican-leaning states undue voting power relative to their populations. Republicans have so heavily gerrymandered the House that Democrats are consistently underrepresented relative to their portion of the popular vote.

Our focus here at ESY is the House — questions about the presidency and Senate will have to wait. We’ll concentrate on the last claim: Do Democrats face a structural disadvantage in House elections?  To start, it’s important to understand two measures of bias in the House. We’ll then look at how these biases have come into play historically and, finally, how that relates to the generic ballot.

The Median House District
One way to measure bias in the House of Representatives is to compare the median House district’s electoral margin to the national popular vote. If you were to line up all the House districts from most Democratic to most Republican, the median House district would be the one directly in the middle. Number 218 out of 435. The district that would tip control of the House from one party to the other. The distance between this district’s margin and the national popular vote is one way to measure the House’s bias.

In 2018, Democrats led the House Popular Vote by 8.6%. In the median House district, California’s 10th, Democrat David Harder won by 4.5%.  If the whole nation voted 4.5% more Republican, Democrats would still have carried the popular vote by 4.1%, but lost CA-10 and, with it, control of the House. In this way, Republicans had a 4.1% structural advantage in the House in 2018.

David Wasserman of the Cook Political Report uses this measure to argue that Democrats have faced a disadvantage in the House since the 1960s.

The Seat Bonus
The other way to measure bias in the House bias is to measure each party’s “seat bonus”. A party’s seat bonus is the gap between their share of seats in the House and their share of the popular vote.

In a perfectly proportional system, a party that wins 53% of the vote would receive exactly 53% of the seats. The House, though, is decided by 435 individual elections rather than proportional allocation, making a perfectly proportional distribution of seats unlikely. In 2018, Democrats won 53% of the popular vote and won 235 of the House’s 435 seats, which translates to 54%. The Democratic seat bonus, therefore, is 1%.

This is perhaps a more intuitive way to measure the House’s bias. It also paints a more complex picture of the parties’ structural advantages and disadvantages in this legislative body. Unlike the “Median House District” measure, the “Seat Bonus”, over the years, has swung back and forth between the parties.

Tracking the “Seat Bonus”
The following graph shows Democratic and Republican overrepresentation in the House, measured by their proportion of seats minus their percentage in the national House popular vote, from 19722The 1972 redistricting was the first to take place under the Supreme Court’s “one person one vote” ruling which mandated districts of roughly equal proportion, making that election a natural starting point. to 2018. It’s a graph of each party’s “seat bonus”.3All data in the graph is rounded to the nearest 1%


Data: Brookings Vital Statistics on Congress

The graph shows three eras in the House. From 1972-1992 Democrats consistently benefited from a seat bonus while Republicans faced a heavy seat penalty. From 1994 – 2008, things were a bit more muddled, with Republicans and Democrats both usually benefitting from a seat bonus, but Republicans typically getting more of a bump. Lastly, from 2012-2016 Republicans got a strong seat bonus while Democrats faced a penalty. The most recent election 2018, seems to break this last era of Republican overrepresentation, but we will have to wait to see where the trendline goes in 2020 and after.   

These three eras line up almost perfectly with which party had control of the House. Democrats had majorities from 1972-1992, from 2006-2008, and again in 2018 while Republicans did from 1994-2004 and from 2010-2016. The party that won a majority in the House almost always received a bigger seat bonus. Out of 24 elections, only 1994 breaks the trend. And over these 24 elections, the party that won the House majority received an average 6% seat bonus. This bonus, however, has decreased from an average of 8% between 1972-1982 to 4% from 2008-2018.

The reason for the seat bonus deserves its own blog post and will be the focus next week. Some preliminary explanations:

1) Swingy seats are (proportionally) overrepresented. In 2018, there were 91 seats that fell in the range of +10 Republican to +10 Democratic. Since seats can range from +100R to +100D — a total range of 200 points — the 20-point range of +10R to +10D represents 10% of the total possible range. One might expect 10% of House seats to fall within this range. Instead, 91 seats (21% of all House districts) fell in this range. Because these close seats are overrepresented, a party will be disproportionately rewarded for marginal gains around 50% of the two-party vote.

2) Seats have different levels of elasticity — meaning they respond differently to changes in the national environment. According to FiveThirtyEight, for every change in 1% of the national popular vote, Michigan’s 5th District — the most elastic district in the nation — will move 1.24%. On the other side is Pennsylvania’s 3rd, which will only change .72%. Perhaps the swingy seats are also the most elastic, meaning that a party will be disproportionately rewarded for marginal gains around 50% of the two-party vote.

Again, these are just ideas. Next week we’ll dig into the data.

Below is chart similar to the one above, but this one uses the two-party popular vote, which excludes any votes for third parties.


Data: Brookings Vital Statistics on Congress

It looks almost identical to the first graph, with one big exception. In this graph, both party’s trendlines are consistently a few percentages lower.  When third party votes are included in the popular vote total, both Democrats and Republicans over-perform in House to a greater extent than when these votes are excluded. The reason for this is clear: third party votes rarely translate to real representation in the House. Third party votes increase the raw number of votes in the popular vote, meaning that Democrats and Republicans will receive a lower percentage of the total vote percentage. But because third party candidates almost never get elected, Democrats and Republicans do not see their overall representation in the House go down.

Tracking the Median House District
The Median House District measurement has shown a consistent bias in favor of Republicans for decades. Take a look at this graph charting the median seat in the House and Senate relative to the presidential popular vote. Every presidential year from 1968-2016 (excluding 1980), Republicans have had a two to six-point advantage. And that advantage has risen since the turn of the century, reaching about 5.5% in 2016.

This discrepancy comes from how Republican and Democrats are distributed among districts. Democrats are heavily concentrated in cities and urban areas; Republicans tend to be more spread out among rural, exurban and suburban districts. There are more districts with very high portions of Democratic voters than there are districts with very high portions of Republican voters. While party density makes it very easy for the individual seats, it also leads to a lot of wasted votes. Read my last post for more detail, but due to natural sorting and gerrymandering, Democrats waste more votes than Republicans. These wasted votes largely explain why the median House district is more Republican than the national popular vote. In 2018, Democrats won 180 seats by a 15% margin or greater while that number was just 126 for Republicans.

But…the Generic Ballot?
The Generic Ballot estimates the House popular vote which is used to project seat allocation between the parties. To understand the Generic Ballot’s utility at projecting seat allocation, it’s necessary to understand how well it predicts the House popular vote (which we covered in the first two pieces in this series) and how well that popular vote translates to seat allocation (which we covered in the last post and here).

According to FiveThirtyEight’s poll aggregator, Democrats lead the 2020 two-party Generic Ballot 53.7% to 46.3%, a 7.4% margin (as of July 13). This number is bound to change over the next year and a half. As 2020 approaches it will likely mirror the presidential election polls and in the meantime will track President Trump’s approval rating.

A 7.4% lead, though, would probably push Democrat’s past the Republican “Median House District” bias, giving them control of the House. Along with control of the House, Democrats would probably benefit from the “Seat Bonus” bias, giving them a greater than 53.7% (234 seats) share of House seats. Democrats didn’t get that “Seat Bonus” in 2018, winning about 8.6% in both the Popular Vote and seat allocation. However, as we saw from Republicans in 1994 and Democrats in 2006, it is common for a party to not receive a big bonus the first year that they re-take the House but then to see the bonus factor in the next election.

As we learned in the first post of this series, the Generic Ballot is pretty good measure of the national popular vote. It’s a good indicator of the national mood and predicts well how many House seats each party will win. We will end with two caveats to this. One: The Generic Ballot is much less reliable this far out from election day. Two: The House of Representatives is not decided by the national popular vote. It’s decided by 435 individual single-member districts with local factors and unique candidates. In this sense, there’s nothing at all general about House elections.

THE GENERIC BALLOT: SEAT ALLOCATION VS THE POPULAR VOTE (PART 1)

The 2020 elections are still 16 months away and yet pollsters are out in force, giving us just enough information to break out our crystal balls and make wildly irresponsible predictions. This is the third post in a four-part series looking at the Generic Ballot and its utility as an election predictor. You can read the first post in the series here

The Generic Ballot is a poll question that aims to measure the national popular vote for the House of Representatives. Respondents are asked to choose between a nameless Republican and Democrat for Congress. Gallup asks it this way: “If the elections for Congress were being held today, which party’s candidate would you vote for in your congressional district — the Democratic Party’s candidate or the Republican Party’s candidate?”

Each party’s number of seats in the House of Representatives, however, is not determined by the national popular vote. A party that gets 45% of the popular vote will not always (or usually) receive 45% of seats in the House. Instead, we hold 435 individual elections, district by district. There are, therefore, two degrees of separation between the Generic Ballot and House seat allocation. The Generic Ballot predicts the national House popular vote which then can be used to estimate each party’s seat allocation.

How U.S. House Elections Work
The current conception of U.S. House elections — 435 individual districts with one representative each — is not mandated by the Constitution. Regarding seat apportionment among the states, the Constitution stipulates that states have House representation proportional to their population and that each state has at least one representative. As for elections, the Constitution says that only these representatives should be “chosen every second Year by the People of the several States.”

Instead, the Uniform Congressional District Act, a federal law passed in 1967, mandates the use of single member districts in all states with more than one representative except Hawaii and New Mexico.4These two states were given exceptions, allowing them to continue electing representatives at-large. Both states have since done away with this practice. Without this statute, a state could theoretically establish multi-winner elections, where all of a state’s voters choose from the same slate of candidates and the candidates with the most votes fill the number of open seats in order of votes received. If, for example, there were eight candidates running for three open statewide seats, the top three vote getters in the statewide election would fill those three seats. As of now, though, this is against federal law.

Maine did run ranked choice elections in 2018 for House and Senate, rather than traditional plurality elections. But even in this system, the state is split up into proportional districts according to population and each district gets one representative. With this caveat, the this single-member district, first past the post system of choosing representative dominates U.S. House elections. So that’s what we will focus on here.

Why Party Seat Allocation Differs From The Popular Vote
Single member districts mean that some votes will not be represented. The votes of Republicans in a heavily Democratic districts, Democrats in heavily Republican districts and third-party voters are essentially useless. And in close elections — say a candidate wins 51% to 49% — nearly half of the electorate’s votes go without representation. Unsurprisingly, on a national scale, this means party representation does not match the national popular vote.

A key concept is the ‘wasted vote’. There are two kinds of wasted votes. Type One is a vote that does not go the winner. So, in a 60-40 election, the 40% of votes that did not go to the winning candidates are Type One wasted votes. A Type Two wasted vote is any vote for the winning candidate over the threshold to win the election. It is essentially an additional vote that the candidate would have won without. In an election with a 50% winning threshold, any vote over that 50% mark is a Type Two wasted vote. One might think that, on a national level, the number of Republican and Democratic votes would balance out. There are two big reasons that this does not happen.

1) Natural Sorting
Democratic and Republican voters are not spread evenly across the country or within states. Democrats are concentrated in urban areas while Republicans are spread out over larger, more rural areas. When congressional districts are drawn, Democrats are often naturally placed into districts that are overwhelmingly Democratic due to their heavy concentration in cities and urban areas. This creates a lot of Type Two wasted votes. If an urban district is 85% Democratic, that 35% over the 50% mark are unnecessary for Democratic representation and wasted votes. On the other hand, Republicans tend to live in suburban, exurban and rural areas. They are more evenly dispersed around the country, making it rarer for a district to be overwhelmingly Republican. This distribution is more efficient for Republican voters, because if they live in more 60-40 districts than 85-15 districts, they have cast many fewer wasted votes.

2) Gerrymandering
Gerrymandering is the drawing of congressional districts to favor one party over another. Because state legislatures are responsible for drawing congressional districts, they often try to maximize their party’s federal representation.

The two methods used in gerrymandering are ‘packing’ and ‘cracking’. Both involve maximizing the opposite party’s number of wasted votes. Packing is drawing a small number of congressional districts that heavily overrepresent the opposition party’s voters, creating a lot of the Type Two wasted vote. Cracking is the opposite of packing — diffusing the opposition’s voters into districts so as to create Type I wasted votes.

Most gerrymanders are a combination of Packing and Cracking. Imagine a state that has 100 voters: 50 Republicans and 50 Democrats and five congressional seats of 20 people each. A Democratic legislature could draw the districts to look like this:

1) 20 Republicans
2) 7 Republicans + 12 Democrats
3) 7 Republicans + 12 Democrats
4) 8 Republicans + 13 Democrats
5) 8 Republicans + 13 Democrats

The Republican Packing in District 1 along with the Republican Cracking in Districts 2 – 5 allowed Democrats to win four out of five seats in a state that his half Republican.

Natural Sorting and Gerrymandering are the two biggest reasons that the national popular vote and seat allocation among the parties do not match. There are, of course, other reasons: third party votes, unequal district sizes, differing voter turnout in districts, etc. But these two are the biggest structural and geographic factors of our system that ‘distort’ House representation away from the popular vote. Now that we know what causes the gap between popular vote and House representation, we can look at how significant this gap has been in the past, how much it matters today, and which party tends to benefit from it. Next week: SEAT ALLOCATION VS THE POPULAR VOTE (PART 2)

THE GENERIC BALLOT: 18 MONTHS OUT

The 2020 elections are still 18 months away and yet pollsters are out in force, giving us just enough information to break out our crystal balls and make wildly irresponsible predictions. This is the second post in a four-part series looking at the Generic Ballot and its utility as an election predictor. You can read the first post in the series here


Last post I argued that the Generic Ballot — the polling question that asks respondents if they plan to vote for a Republican or Democrat for Congress — is fairly predictive of election results when the poll is taken right before or on election day. But what about a year and a half out, as we are now from the 2020 election?

Early Generic Ballot polling is tough to track down, but Real Clear Politics has aggregated such polls since the 2004 Bush v. Kerry presidential race. The sample here is very small, only four presidential election cycles, so adjust skepticism accordingly.

In the chart below, I have averaged the results of the first five Generic Ballot polls taken in an election cycle, starting 19 months out from the election5Notice that not every cycle has polling data going back 19 months before the election. 2004 in particular, lacks very early polling data..


Data: RealClearPolitics

In 2008, 2012, and 2016, the early polling was surprisingly accurate, missing real election results by a net average of only 1.7%. The 2004 election polling, though, breaks this trend. The earliest polling projected a Democratic win of 10%, but Republicans won the House Popular Vote by 2.6%, leading to a 12.6% polling miss.

Lacking more historical data, it’s hard to determine if the error in 2004 is an outlier or not. If it is, and early Generic Ballot polling is generally within 1 to 3% of real election results, we could say that the numbers we see now are predictive of what’s to come in 2020. And while control of the House could hinge on this 1-3% polling error, the Generic Ballot polls would still be able to show the world of possible outcomes and which is most likely.

If 2004 is not an outlier, though, and Generic Ballot polling really will miss the mark by double digits about a quarter (or more) of the time, the Generic Ballot this far out cannot show us the world of possible outcomes, or the likelihood of these outcomes. Predicting that the House Popular Vote could be anywhere between a net +6 for Democrats and a net +6 for Republicans does not narrow election outcomes in any helpful way.  Because we can’t determine how likely a polling error like 2004 is the Generic Ballot this far out it should not be treated as predictive.

With that caveat, here are all the Generic Ballot polls taken within the past two months from pollsters receiving at least a B- pollster rating by FiveThirtyEight. Note that Morning Consult and Politico have conducted all but one of these polls, which raises the probability that there is a consistent statistical bias in their results, meaning that the data should be treated with even more skepticism.


Data: FiveThirtyEight

Clearly, Democrats currently have a consistent advantage in the Generic Ballot, averaging out to 7%. And while, again, this is not predictive 18 months out from election day, it does expose three important things about the current political environment.

1) The wave that swept Democrat’s into power in the House in 2018 has not dissipated. In that election, Democrats carried the House Popular Vote by 8.6%. So, Republicans may have closed this gap by a point or two, but the pro-Democratic sentiment largely remains.

2) Most voters’ Generic Ballot responses are determined by their feelings about Donald Trump. His net approval rating during this two-month period has ranged from about -9% to -13%. Not exactly the same as Generic Ballot polling, but close enough to give the impression that Trump’s approval is a big factor in down ballot decision making (at least in polls).

3) The small gap between Trump approval and the Generic Ballot average is important, though. While most voters who approve of Trump will vote a Republican Representative, and voters who disapprove of Trump will vote Democratic Representative, there is a small number of voters who distinguish between the top of the ticket and down-ballot races, And while this population is small and shrinking, it still exists. When control of the House can hinge on just one seat, any swing can be determinative.

The big question is if Trump will be able to turn his poor approval rating around. He has been stuck around -9 to -13 for most of the last year. If he can pull it closer to even, Republicans have a good chance to win the Generic Ballot and to take the House. If he continues to slum it down near negative double digits, 2020 will almost certainly see continued Democratic control of the House, likely control of the Presidency and possible control of the Senate. That’s the order Democrats are likely to hold or take power: House -> Presidency -> Senate. It’s hard to imagine a world where Democrats win the Presidency and lose the House or where they win the Senate and lose the Presidency and/or House.

Control of the House, though, is not determine by the Generic Ballot or the House Popular Vote. It is decided district by district. Whichever party wins in a majority (or plurality if there are 3rd party winners) of the 435 House seats will hold the House. In the next post we’ll look at how closely the House Popular Vote correlates to seat allocation between the parties and what this means for the 2020 election. The big question we’ll aim to answer: how much do Democrats or Republicans need to win the House Popular Vote to take control of the House?

THE GENERIC BALLOT: PSEPHOLOGY’S CRYSTAL BALL

The 2020 elections are still 18 months away and yet pollsters are out in force, giving us just enough information to break out our crystal balls and make wildly irresponsible predictions. This is the first post in a four-part series looking at the generic ballot and its utility as an election predictor


Democratic presidential primary polls have been dominating election headlines, but some congressional polls have been been released too,  albeit to much less fanfare. We should probably ignore these polls — it’s too early for them to be predictive. Prognosticating off these preliminary numbers is rash, reckless and generally unwise. And so that’s exactly what we will do here in a three-piece series. This post will focus on how predictive the “generic ballot” is on, or close to, election day. The next post will look at how valuable these numbers are now, given that the election is still a year and a half away. The last will look at the relationship between popular vote and seat apportionment.

In “generic ballot” polls, respondents are asked to choose between a nameless Republican and Democrat for Congress. Gallup asks it this way: “If the elections for Congress were being held today, which party’s candidate would you vote for in your congressional district — the Democratic Party’s candidate or the Republican Party’s candidate?” It measures national support for the two parties without some of the baggage carried by their polarizing national figures (think Donald Trump, Bernie Sanders, Mitch McDonnell, Nancy Pelosi, Alexandria Ocasio-Cortez, etc.). It can also bring forward nuanced voter trends. For example, some voters might support a Democrat for President but Republican for House because they want the grinding impotency of split government. Another might vote the same way but because they support Trump but want continued House oversight. Another might choose to reward the current power dynamic because the economy is strong.

Final generic ballot polls — the ones taken within a few days of the actual election — are a better indicator of election outcomes in midterms years than in presidential years. Since 1948, the final generic ballot has missed the real midterm vote by an average of only 2%. The fallibility of the generic ballot in presidential years, though, is clear in the chart below.6Galup data from 1988 was not available.7Data is based of ‘likely voters’ for 1976 and 1996-2006 and ‘registered voters’ otherwise8The two-party generic ballot was used when available (2004-2016)9Data from Real Clear Politics was rounded to nearest integer to make it consistent with Gallup’s data


Data: Gallup101968-2000, Real Clear Politics112004-2016

Since 1968, the generic ballot has missed the real House popular vote by an average of 4% and until 2008, it consistently overestimated Democratic support.  Both of these problems have been ameliorated in recent years, resulting in a more accurate and balanced  generic ballot since the late 1990s. And even though the generic ballot is less accurate in presidential years, it is still correlated with election results.12Data is based on the chart above. However, data from Real Clear Politics (2004-2016) is rounded to two decimal places rather than the nearest integer as in the chart

A party that performs better on the generic ballot will generally earn more votes in the election. The R2 in the bottom left corner indicates how much of the variance in the dependent variable (Democratic Margin in the Two-Party Vote) can be explained by variance in the independent variable (Democratic Margin in Generic Ballot). In English: R2 shows how well Variable 1 can explain or predict Variable 2. Here, R2 equals almost exactly 0.5, meaning that about half of the variance in the popular vote can be explained by the generic ballot.

Looking at the equation above the R2 value, the trend line has a slope of 0.54. This means that in general a 1% increase in a party’s margin in the generic ballot translates to about a 0.54 % bump in the House popular vote. There is still plenty of variation, though (see how the data points are not clustered too closely along the trend line), so this is not at all a perfect measure for any individual election. Even with this variation, and the knowledge that a lead in the generic ballot usually overestimates electoral success, a party is better off leading the generic ballot than trailing. And their better off leading with a larger margin than a smaller one.

Two caveats to this data. First, the makeup of the House of Representatives is not determined by the popular vote. Seat apportionment, gerrymandering, demographic & partisan sorting and other structural imbalances mean that a party’s share of the national House vote can differ substantially from the share of seats they actually win. Second, this data is from polls within a few days of the elections. We’re still a year and a half away from November 2020. The next two posts will focus on understanding these qualifications to the generic ballot.