Protected: Two ways to estimate primary outcomes without polls (transcript)
This content is password protected. To view it please enter your password below: Password:...
The Takeaway – today!
Today on The Takeaway, John Hockenberry and I discuss whether a realignment is brewing (spoiler: probably not). On WNYC-AM at 9:00am Eastern, WNYC-FM at 3pm. Broadcast nationwide, and here is the recorded podcast.
Topics:
Will the appearance on The Takeaway be available as a podcast that can be downloaded later?
I was able to find it on my podcast client. Also, there was a bonus Don Delillo episode. Thanks!
great interview. too bad he kept calling you sam wong.
Actually, I believe Wang does rhyme with “Song” and not with “Sang.”
Correct.
I agree. there will be a few odd switches, but nothing permanent. I just wish people would stop talking about the end of the Republican party.
If the Ds get the Congress back, they will just lose it again in 2018.
And the Rs will just go back to business as usual. and in 2020 they will be saying “Donald who? I don’t know no Donald.”
What’s worse is that if/when Hillary wins this year, it will just set the R up for 2020 and another census year and more gerrymandering.
Just my 6 cents. 2 for each topic.
Actually, unless the GOP really is cracking up, they are set to win in 2020 either way. If they win this year, then 2020 would be a re-election year. Typically, a party going for its second term in power wins big (exceptions in the last century are 1980 and 2012, but even in 2012 Obama won). If the Democrats win, they would be trying for a fourth consecutive term, which has only been accomplished once in American history (1944); third terms are relatively common.
The only way for Democrats to avoid Republicans tightening their grip on state legislatures, and therefore the House, in 2020 is to get behind a non-partisan re-districting movement now.
@alurin
That’s not right. The Republicans won 4 in a row with McKinley-Roosevelt-Taft and then 6 with Lincoln-Grant-Hayes-Garfield-Arthur (although with Lincolns death Democrat took over, but even if you excluded that they’d still have 4 in a row). Going way back the Democratic-Republicans won 7 in a row with Jefferson-Madison-Monroe-Adams.
In short, this is the 58th election, and in those 58 elections one party has won 4+ elections in a row 4 times (not counting duplicates). Of those 58, 22 (38%) have been part of one of these streaks. I would say this is not an unusual enough situation to predict a Republican victory in 2020.
Alurin, about arguing from precedent, xkcd has you covered.
https://xkcd.com/1122/
The human need to see patterns where they may not exist, whether it be lines connecting the stars in the sky or number of consecutive elections won by a party, is very strong and needs to be continually fought against.
Unless you can posit some credible mechanism as to why the number of consecutive presidential wins by a party is naturally limited (voters develop antibodies?) I’m afraid arguing from statistics is not going to work on its own.
I think there’s, at least, a plausible just-so story. The world is rarely as nice as people want it to be. When a party is out of power for a long time, they don’t actually have to govern, and gradually take the role of “generic alternative”, while the party in power gradually gets everything bad that happens attached to it.
Of course, there are mechanisms pushing in the opposite direction: incumbency is an advantage for many reasons, and if we’re talking about holding the Presidency for a fourth election (rather than a third, as now), the incumbent party’s candidate would most likely be an incumbent.
And when generalizing about presidential elections, small sample sizes are always a danger–the party system can completely change before you have a reasonable statistical sample.
Hi Matt, good points. I’m sure you know the counterexamples better than I do (FDR in the US, Labour in UK 1997-2010, LDP in Japan). My only point is that if there are so many exceptions to a pattern, maybe there is no pattern.
It might be appropriate to see how often a party is elected once, twice in a row, three times, and so on – and then to do a statistical test for randomness. However, Amit is probably right – there’s unlikely to be a pattern. This sounds one of those political superstitions.
Fundamentals-wise, it stands to reason that 2020 could be a good year for the Republicans to defeat a President Clinton. We may already be seeing the seeds of the next recession in the international news, and while I doubt it’s going to go acute before November, if it really pops sometime after she takes office, that’s always a hard thing for a President and their party to shake.
But it all requires Republican voters to somehow nominate a candidate who could appeal to a plurality of the electorate, and that’s getting harder and harder for them to do.
There is little indication that the streak of Democratic wins are anything other than the result of shifting demographics and the Republican’s inability to appeal to non-white voters. I don’t see why that would change unless the Republicans change, and that would carry risks with portions of their white base.
Points taken.
I have no idea what would constitute data for this hypothesis, but @Matt McIrvin’s just-so story is what I have always assumed. You have competition between the advantages of incumbency and the fact that people get tired of the same faces in power. And of course after 8 years you can’t blame the previous administration any more.
As for the stats… Omitting Washington and Obama, the mean run length is 2.4 terms (median 2 terms, unsurprisingly). The number of runs is 22, somewhat less than the 26-27 that would be expected by chance, but that’s probably not significant.
Still, the distribution of runs is pretty skewed, with only 7 runs of length > 2, and only 4 runs > 3 terms.
Of the 22 times a party has claimed the White House from another party, 14 times (63.6%) they have won a second term. Those second terms have led to a third term 7 times (50%). Four of those seven third terms have led to a fourth term (57.1%), and three of those fourth terms have led to fifth terms (75%). Then there are two sixth terms (67.7%) and one 7th term (50%). So there seems to be a bias to keep the same guys in power which declines over time, so that getting a third term is like flipping a coin. But once you have a third term, your odds get better again. Of course, that’s based on just four long runs… This also assumes that presidential politics is stationary, but the really long runs occurred in the 19th century, including the 7-term run of the Democratic-Republicans and the 6-term run of Lincoln’s Republicans (counting 1864 as a Republican victory). Runs have been getting shorter over time. Again, the Ns are small.
So, my original claim was half wrong: IF Clinton wins this year, the odds are slightly in her favor to capture a 4th term for the Democrats. However, should the Republicans win, they would be strongly favored to win again in 2020.
Interesting–to me, that analysis seems entirely consistent with nothing more than a weak advantage in favor of incumbents, with small-number randomness providing the other variations.
Hockenberry seemed to associate your election expertise to your job as a neuroscientist rather than someone versed in statistics, so much so that he wanted you to use your superpowers to predict the election. I’m glad you called that “science fiction”.
Just to put this “realignment” idea to bed (in my head anyway). In mechanics any many-body motion can be broken up into center of mass (CM) motion, and motion of bodies relative to the CM.
I am picturing realignment as relative motion, such as the South moving to Republicans. Blowout election such as Reagan, W. Clinton involved the entire electorate (motion of the CM) so are not considered “realignments”.
That is a good analogy!
Even if they are not considered realignments, blowout elections probably have residual effects. For example, Bill Clinton’s election seems to have switched 10 states from R to D, and these have not changed back to R since.
I think Sam’s analysis would treat that as just the effect of the gradual realignment that had been happening since 1976, finally manifesting in election results after it was masked by the nationwide Reagan/Bush victories.
@Matt McIrvin
Agree. But in the fantasy scenario that Trump turns 10 states from Democrat to Republican, and they stay that way for 5 presidential elections, we would be calling it a realignment, even if it was an incremental push over 50% for all those states.
@anonymous: Well, the pundits would. But a quantitative analysis might disagree.
I enjoyed listening to you Sam and have enjoyed your site for the last 5 years! Thanks you!
Congratulations for the common cause award!
Eitan Hersh at 538 reaches for an excuse to say that we don’t know anything!!!