Understanding those Senate numbers

As you’ve probably noticed in past years, American elections take a while to finalise their results. A number of seats are still doubtful from Tuesday’s midterms; the Democrats look to have won a majority of about 25 in the House of Representatives, 230-205 (a gain of 35 seats), but that could still vary by a couple either way.

The Senate is more definite. Three seats are officially still doubtful: Arizona, Florida and the Mississippi by-election. But the first two look unlikely to shift from Democrat and Republican respectively, and while Mississippi is going to a runoff on 27 November, it’s as certain as anything can be that Republican incumbent Cindy Hyde-Smith will win it.

Recall that the Republicans started with a 51-49 Senate majority (counting the two independents in the Democrat column), so the Democrats needed to win two seats to take control.

They did indeed win two seats, Nevada and (apparently) Arizona. The problem is that they also lost four of their own seats: Indiana, Missouri, North Dakota and (apparently) Florida. So in net terms they’ve gone backwards by two seats, producing a 53-47 Republican majority.

In previewing the election, I tried to explain why the House and Senate were so different. Now we can see just how that worked.

But we need to focus on the right thing. There’s a popular meme going around showing the Democrats leading in the Senate popular vote by 55.4% to 43.0%, yet still losing seats, and inferring that something is badly wrong with Senate representation. There is something wrong, but that doesn’t show it.

That’s because it’s comparing a stock with a flow, as accountants would say. Knowing how many votes you won doesn’t tell you anything about how many seats you might expect to gain or lose, until you compare it with the votes you won last time.

In fact, the Democrats look to have won more of the popular vote than they did in 2016. But that’s not very informative either, because since only a third of the Senate is elected each time, a different set of states were voting then.

You can try to compare apples with apples, by just looking at the 16 states that voted both times (leaving out California, for reasons that will become clear later). That shows, by my reckoning, a median swing to the Democrats of around four per cent. So why didn’t they pick up seats?

But hold on! If you just compare those states with 2016, they did pick up seats. Four states (Arizona, Ohio, Pennsylvania and Wisconsin) elected Republican senators in 2016 and Democrats this week. But that’s not reflected in the tally of gains and losses, because they’re different seats: those 2016 senators won’t be up for re-election until 2022, and what happens to them will get counted then.

So the real comparison is with 2012, when this year’s senators were elected. And the problem is that 2012 was a very good Democrat year, so they were always likely to lose by comparison: they had many more seats to defend than their opponents, and they were doing it in a midterm election, when voters tend to be richer, older and whiter.

Moreover (although this might take another post to analyse properly), the particular states on the map this year seem to be disproportionately the sort of places where the Republican Party has been gaining ground, relatively speaking, in the Trump era: Indiana, Michigan and Missouri, for example, but not Colorado, North Carolina and Oregon.

The popular vote in 2012 was Democrat 53.4% to Republican 41.8% (official figures here), or 56.1% to 43.9% two-party. So if the figures currently being quoted are about right (they can only be approximate, since counting is still going), it looks as if there’s been very little swing at all.

But that’s misleading too. In the biggest state, California, the Democrats won in 2012 by a bit more than three million votes. But this year – because California runs a “jungle primary”, or first round election – there was no Republican on the ballot paper. Two Democrats ran on Tuesday, and collected something like seven million votes between them, improving their nationwide advantage over the Republicans (in a wholly artificial way) by about four million.

When the national Senate vote tallies are finally calculated, I expect that if you leave out California, they’ll show a two-party swing to the Republicans (as compared to 2012) of around three per cent, which would easily account for their net gain of two seats.

One more question. The Democrats have to contend with the problems of huge differences between large and small states, and a smaller and less representative turnout in midterms. But haven’t those things always been problems? Why do Democrats seem to be having a particular problem with the Senate now?

At least part of the answer is the growth of partisanship and straight-ticket voting. Once upon a time, popular Senators could entrench themselves even in states that were otherwise hostile to their parties. A few still do, but it’s become much more difficult. The four Democrats who lost were all in states that had voted for Donald Trump in 2016.

That means that the full effects of the small-state bias in the Senate are now being felt, and that’s bad news for the Democrats. But at least the map will be more favorable for them in 2020 and 2022.

8 thoughts on “Understanding those Senate numbers

  1. At least part of the answer is the growth of partisanship and straight-ticket voting.

    I guess, but not in Texas where the Republican governor got re-elected in a landslide but Ted Cruz lost 13 points compared to last time (but still scraped back by 3 points; note: despite being backed/bear-hugged by Trump!). And in fact there was a blue-wave across many of the other elections in Texas but just not enough to quite overcome the red threshold. I haven’t seen an analysis but I suspect those new millennial voters still didn’t turn out enough for Beto O’Rourke; I mean with 1.4m new voters …. Still, he has announced he isn’t going anywhere so that should be good for 2020. There is a lot of denial going on in GOP camp but the state is purple and ineluctably turning blue.

    The other big factor, that I guess you covered, is that many of the required shifts to the Dems occurred but not always or often enough in the states needing it. PBS-Newshour last night (ie. Wednesday US time) did a good analysis. For example that thing with college-educated white suburban women who either voted Trump in 2016 or stayed away because of Hillary-hate, did vote more and they did vote a lot more for Dems, however not in some red states. (I’m not sure why they picked Indiana for the example, since I don’t think it was expected to turn blue–it voted back Senator Pence, but his brother!–but this showed a dramatic still-Trump amongst these voters. In fact maybe there was an overall Pence factor?).

    Still, I persist with my earlier complaints. The Dem party has to get its shit together, namely with a clearer leader much earlier than just the presidential season. That is why Beto O’Rourke’s near-miss is such a disappointment as he was seen as a bit of a messiah for them. Nevertheless, with less gerrymandering and no blockage of ex-felons, and a natural higher turnout of new voters, the next round in 2020 should be good for Dems. But one has come to almost expect them (a bit like UK Labour or Australian Labor) to manage to shoot themselves in the feet again.


    1. Thanks Michael. The straight-ticket voting doesn’t apply as much to state races; we’re seeing Governor typically differ more from Senator/House/President than they do from one another. But even then not always – the alignment between Governor & Senator in Florida is quite striking.

      And agree completely about the need for the Dems to get their act together, and their apparently unlimited capacity to foul things up. I think an early retirement date for Nancy Pelosi would be a good start.


      1. I was wrong about the millennials. They did go and vote for O’Rourke, and that also helped several other Dems defeat incumbent GOP Reps. In fact with all the voter segments that turned out for him, it is surprising and disappointing he didn’t defeat Cruz.
        Anyway here is what the Guardian wrote:

        O’Rourke attracted the votes of 39% of Texan white women – compared with the 29% who backed Hillary Clinton in 2016.

        In 2016, Clinton attracted the votes of 55% of the 18-29 age range in Texas, to Trump’s 36%. This week, O’Rourke won a stunning 71%, to Cruz’s 29%.
        Not only did he win over young people in far greater proportions, he also crucially managed to unlock a door that has been frustratingly closed to progressive causes in vast swaths of America for years. He persuaded young voters who usually overwhelmingly opt to stay at home in midterm elections to get off their couches, get over to the polling stations, and vote.
        Figures for overall Texas turnout have yet to be completed, but early voting data is again stunning. The number of 18 to 29-year-olds casting an early ballot this year was five times greater than in the 2014 midterms.

        In addition to being young, he is also a native Spanish speaker. Makes one kind of wonder just why the Dems don’t search for and select such candidates. Not rocket science. Half their problem, with money (old & corporate) and connections etc being more important than genuine electoral appeal. Like that young New Yorker who has just become the youngest ever elected Rep; of course that race is not competitive for GOP so the real race was the primary where she defeated some old fart who had been in power forever and expected to remain forever, and this shocked the establishment Dem party. This election showed what the Dems need to win but I’m still not sure they really have got the message, or the will where it counts. The young(er) will just have to grab it from the old-guard and that is a pretty big ask.

        The depressing thing is that he spent $70m which apparently is more than spent on a single senate or House race, ever. I suppose it’s a slight compensation that it was all raised from small donations, not a single dollar from corporates. Overall these elections spent $5bn, though someone pointed out that in the same week Americans spent $9bn on Halloween stuff.


  2. The comparison with 2012 falls down somewhat because the Democrats failed to win the House that year. So whichever way you look at it, they’ve gone forward in the House and backwards in the Senate. That’s unusual.

    I think your point about partisanship really nails it. Also too, the Senate hinges on just a few competitive races, So the small quirks in a particular Senate race make a big difference. For instance, two of the defeated Democratic senators, Claire McCaskill and Joe Donnelly, benefitted from facing awful Republican opponents in 2012. No such luck this time.


    1. Thanks David. Yes, because the number of competitive Senate races is so small, a lot of it comes down to luck. It looks as if Nelson is going to lose Florida, for example, by just a handful of votes. And yes, McCaskill really should never have won in the first place, it’s just the Republicans shot themselves in the foot so badly in 2012.

      True, the Dems still didn’t win back the House in 2012, but I think the big thing there was the “sophomore surge”: you had this big cohort of Republican members who came in in 2010 and now were incumbents for the first time, and the advantage of incumbency is huge. But I’d want to look properly at the figures for that to see if there was something else going on.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.