11

Is there an intuitive reason why Hillary Clinton won the popular vote amongst people voting for President, yet the GOP won the popular vote amongst people voting for their Representative in the House? Is the answer 'lots of vote-splitting by people who expected Clinton to win and wanted a GOP House as a check on her power'?

President: 63m for GOP, 65.9m for Democrats (Wikipedia article on 2016 presidential election results)

House: 63.2m for GOP, 61.8m for Democrats (Wikipedia article on US House of Representatives elections, 2016)

  • But see also: 'Republicans captured the majority of the "popular vote" for the House on Election Day, collecting about 56.3 million votes while Democrats got about 53.2 million, according to USA TODAY calculations.' (Incorrect? USA Today article)
  • Wikipedia 2016 House Vote Summary, shows Republicans gaining 49.1% of House votes compared to Democrats at 48.0%

Note that comparing total votes cast, it seems an additional 4 million votes were cast for the presidential election compared to the House elections. This makes answers that just reference vote-splitting incomplete, to my eye at least...

EDIT: Since posting this question, I wanted to draw readers' attention to a pretty striking map that possibly explains part of the answer: Seats with only one major party candidate Source: Ballotpedia Hat tip: @lazarusL

It seems that there were fewer no-Democratic House districts in 2016 than no-GOP House districts. In these districts, Democratic voters would have voted for Hillary but would have been mechanically unable to vote for a Democratic House candidate.

Edit #2. Ballotpedia's margin of victory data for all 435 House races (see https://ballotpedia.org/United_States_House_of_Representatives_elections,_2016) lists 12 CRs with GOP unopposed and only 5 Congressional districts with Democrats unopposed. Let's say avg 250k per district, and perhaps 25% of them vote for the presidential candidate from the party that hasn't fielded a Congressional candidate. This leads to a rough estimate of half a million (net) votes to HRC that didn't go to a Dem Congressional candidate. So although the unopposed races story was really elegant, perhaps the bigger story is in fact vote splitting.

Edit #3: Ballotpedia list of unopposed districts doesn't count some districts where a candidate ran against another candidate from the same party, or from a third party. However, these omissions are if anything skewed towards Democrat House candidates running up the vote tally (see Brythan's answer), which makes it even less likely that the unopposed races story is a key component of the answer here.

  • That second edit might have been one too far, as the vote count difference could be for all sorts of different reason unrelated to popular vote difference. Feel free to roll that back. – Jontia Oct 29 '18 at 10:41
  • 1
    A strong answer to this question would include an examination of the broad scope of house races. How many of those races involved a third party that eroded voting results for the Democrat option? – Drunk Cynic Oct 29 '18 at 12:39
  • 4
    Odd choice for the colours on the "No candidate map". – Jontia Oct 29 '18 at 18:35
  • 1
    The thing that immediately rang my bell was that the GOP numbers are nearly the same while the Dems are different. An obvious, though probably wrong conclusion, is that 4M dems voted for presidential candidate but not the house candidate. Your map kind of points to that - if the blue areas minus the red areas equals 4M votes you might have answered it yourself. – CramerTV Oct 29 '18 at 18:42
  • 4
    Technically there is no "Presidential Popular Vote" either. But it does not stop people talking about it. – Jontia Oct 29 '18 at 22:54
10

The reason that there were fewer ballots cast in the house and the reason there were fewer votes for democratic candidates is that in many US house districts there is only one party on the ballot. Here's a map of the phenomenon. Democrats voted for the president in these districts, but had no House of Representatives candidate to vote for, so the total vote for Hillary Clinton was higher than the sum of the votes for Democratic House of Representative candidates.

By the ballotpedia map, there are 30 districts with only Democrat candidates (you have to zoom in) and 26 with only Republican candidates. This doesn't represent a huge disparity in number of unrepresented districts. However, a more detailed look might reveal that the concentration of Trump or Clinton voters in these districts skews things one way or the other.

So why weren't there just as many voters who didn't have congressional people of their party to vote for among Trump voters? The answer is that many conservatives and libertarians voted for third party candidates in the presidential election, but still voted for their local GOP house of representative candidate. Anecdotally, I know people who voted for, Gary Johnson, Evan McMullin, and even Hillary Clinton who voted for their local Republican house candidate. This "Never Trump" wing of the Republican party made the Republican votes in the house higher than that of the president, despite many heavily Democrat districts not having a Republican option.

Given the nature of secret ballots, this is a hard thing to prove, but there is some evidence. Trump received 46.1% of the popular vote in 2016 compared to Romney's 47.2% in 2012. However, in the house, Republicans got 49.1% of the popular vote in 2016 compared to 47.6% in 2012. With Romney the results were roughly the same, with Trump they widely diverge. This, combined with the fact that the raw number of GOP congressional voters was higher for Republicans and much lower for Democrats, supports the idea that third party voters supporting GOP congressmen caused the disparity in 2016.

  • 1
    Fantastic answer. ballotpedia.org/… – the_scheining Oct 29 '18 at 15:22
  • 1
    @elliotsvensson to my eye it's not the 3rd party candidates / NeverTrump dynamics so much as the asymmetric assortment of no-Dem vs. no-GOP districts (in 2016, there seem to have been a fair few more of the former). – the_scheining Oct 29 '18 at 15:44
  • 1
    @the_scheining, and whose fault is that? Furthermore, this is not a problem under the Electoral College. – elliot svensson Oct 29 '18 at 16:03
  • 1
    @the_scheining I think that's also very plausible. The map presentation does hide some of the "no GOP" districts in California and the Northeast, so it's not quite as lopsided as it look at first glance. (the opposite of the problem the GOP usually has with looking over-represented on maps). – lazarusL Oct 29 '18 at 16:28
  • 1
    @lazarusL Ballotpedia's Margin of victory data for all 435 House races (see ballotpedia.org/…) lists 12 CRs with GOP unopposed and only 5 CRs with Dems unopposed. Let's say avg 250k per district, and perhaps 25% of them vote for the Prez candidate from the party that hasn't fielded a Congressional candidate. This leads to a rough estimate of half a million (net) votes to HRC that didn't go to a Dem Congressional candidate. So although the unopposed CRs story was really elegant, perhaps the bigger story is in fact vote splitting. – the_scheining Oct 29 '18 at 20:45
7

Since both elections, both the presidential and the house, were elections on specific candidates, it does not surprise me at all. In both elections people vote for an individual candidate, in the case of the presidental election, it is one candidate nation-wide; in the case of the house elections, it is one candidate per district.

Since, you do not vote for a party in either election, the comparison is somehow moot, although interesting. In other political systems, the people vote for parties, and the members of parliament are selected based on the party's list of candidates.

A possible explanation could be, that the candidates of the Republicans for the House were on average more appealing to the electorate, than the Republican candidate for the presidency.

  • 1
    I agree with your first two paragraphs. The third: your hypothesis represents one possibility amongst many. – the_scheining Oct 29 '18 at 16:42
  • 2
    While the mechanism for voting is based on individual candidates, it is clear that many people do vote based on party affiliation. Here's some data on how voting practices changed when party affiliation was added to the Ballot in Virginia after 2001. eprints.lse.ac.uk/61957/1/… – Jontia Oct 29 '18 at 18:41
5

Also remember that Hillary Clinton did not win a majority of the popular vote, only the largest minority (what is called a plurality). A majority voted against her (and against every other candidate). So no vote splitting needed.

Looking at the numbers you cite, Republicans have roughly equal numbers of votes for president and the House. Democrats lost four million somewhere. This sounds more like Democrats weren't voting for the House but only president rather than vote splitting.

Your Ballotpedia list seems to be missing a number of no-Republican elections at first glance. E.g.

  • CA 12. Democrat and independent. No R.
  • CA 17. Two D. No R.
  • CA 29. Two D. No R.
  • CA 32. Two D. No R.
  • CA 34. Two D. No R.
  • CA 37. Two D. No R.
  • CA 40. D and independent. No R.
  • CA 44. Two D. No R.
  • CA 46. Two D. No R.

That's more just from California than the list for the entire country. If we add those nine to the five from the rest of the country, that's actually more than the twelve Republican-only districts. And remember that in seven of those districts, both candidates were nominally Democrats, so both candidates count towards the Democratic vote total. Also 8, 9, and 16 in New York had no Republican (in addition to 17, where the Democrat ran unopposed). So seventeen Democrat-only versus twelve Republican-only.

  • Thanks @brythan - I've added a link to your answer in the question itself. Note that in the first part of your answer, the logic is hard to follow; the fact that Hillary Clinton won a plurality rather than a majority would, if anything, point against her winning more votes than Dem Congressional candidates. – the_scheining Oct 30 '18 at 9:30
2

Remember that most polling had Clinton winning and a popular idea was to vote for Clinton for President, while electing a Republican Congress to keep her in check. In fact, Clinton carried 25 districts held by Republican House members. Indeed, some were expecting the Senate to go to Democrats (i.e. Sen. Pat Toomey was expected to lose)

  • 1
    How does this relate to the discrepancy between House of Representatives and Presidential popular votes? – lazarusL Oct 29 '18 at 15:10
  • 1
    @lazarusL Bah. I misread it. I've fixed it to answer – Machavity Oct 29 '18 at 15:33
  • Note that comparing total votes cast, it seems an additional 4 million votes were cast for the Presidential Election compared to the House Election. This makes answers that just reference vote-splitting incomplete, to my eye at least... – the_scheining Oct 30 '18 at 12:13
1

The simple answer is: a lot of voters split the ticket when voting.

In 2016, it is safe to say that both of the main presidential candidates had fairly obvious flaws, while the main independent turned out to be a bit of a clod (what is Aleppo), and missed a golden opportunity to gain credibility.

So the voters appear to have split the ticket. I don't really like this person, but I like the other one less, so I'll vote for my party's candidate, but vote for the other party for congress to keep them in check.

The outcome of the 2016 election indicates that a lot of Democrats didn't like Clinton or Trump, and chose Clinton but voted republican on the congressional race.

Keep in mind that Trump's victory was unforeseen. It is likely that most voters expected Clinton to win, but wanted a Republican congress to keep her from getting out of hand.

  • This answer seems plausible but lacks data to back up its claims. Also, the logic seems post-hoc in parts, i.e. the fact that HRC had "obvious flaws" would predict the opposite of her getting more total votes than Dem House candidates. – the_scheining Oct 30 '18 at 8:55
  • Also note that comparing total votes cast, it seems an additional 4 million votes were cast for the Presidential Election compared to the House Election. This makes answers that just reference vote-splitting incomplete, to my eye at least... – the_scheining Oct 30 '18 at 12:13
1

Split-ticket voting is driven by voter frustration, as is low voter turnout. Statistics suggest that such frustration is more prevalent in congressional races than in presidential races, for example voter turnout for the 2016 was roughly 59.3%, but the mid-terms of 2014 saw only 35.9% turnout. Both are quite low among developed countries and the 2014 mid-term turnout was the lowest in recent history.

A longer answer that focus on the why could be that increasingly complex voter behavior exposes flaws in simpler voter models. Political theories, such as the median voter theorem (which suggests candidates will move to the middle), are failing because attempt to model voter behavior as one-dimensional political spectrum or single-peaked distribution are substantially oversimplified. In other words, labels like left vs right or conservative vs liberal have only limited applicability. As a example, individual can be both fiscally conservative and socially liberal. An example of this in US history might be the Regan Democrats who were more moderate/libertarian than liberal.

So we've developed voters who do NOT vote a straight party ticket, perhaps even on principal; frustrated voters may not want to throw their support behind unified government. Studies have shown that those who vote a strait ticket are typically highly partisan (and thus predictable by simplified models), but that straight-party voting is on the decline. At the same time, it seems that many candidates have gotten substantially more partisan, leading to frustration.

In a two-party system (simplification to the major parties) is breaking down as voters get more information and less partisan while candidates head in an orthogonal direction. For the reasons above, the median voter may not have his interests well matched by either of just two candidates, and a realistic political map fills with gray, or at least lighter shades of the color representing political parties.

  • Note that comparing total votes cast, it seems an additional 4 million votes were cast for the Presidential Election compared to the House Election. This makes answers that just reference vote-splitting incomplete, to my eye at least... – the_scheining Nov 3 '18 at 9:02
  • I've endeavored to explain the why of split ticket voting, I'm not just referencing it. – Burt_Harris Nov 4 '18 at 0:32
  • The question itself postulates "So although the unopposed races story was really elegant, perhaps the bigger story is in fact vote splitting." I agree. Undervotes are another symptom of voter frustration, and the median voter theorem is failing. en.wikipedia.org/wiki/Median_voter_theorem – Burt_Harris Nov 4 '18 at 0:40
-6

Gerrymandering.

Both parties do it but over 1,000 state legislative seats and many governorships were lost from the Democrats for too many reasons to list here, under Obama's 8 years in office.

Lots of districts were redrawn in favor of the Republicans

  • 7
    The question is about the "popular vote" (i.e. total numbers of votes). Gerrymandering is about grouping votes to cancel out the popular vote. – Paul Johnson Oct 29 '18 at 8:43
  • 1
    @PaulJohnson It is conceivable gerrymandering disproportionately affects the turnout of particular demographic groups of voters, disfavouring the popular vote for the Democrats? However, if it does, the answer should point that out explicitly and with evidence. – gerrit Oct 29 '18 at 12:15

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.