Should I show any special respect or deference to American University Professor Allan Lichtman because he has correctly predicted the winner of every presidential election since 1984? Think about it. Who did you think was going to win the elections of 1984, 1988, 1992, 1996, 2000, 2004, 2008 or 2012? Did you get any of those wrong?
I mean, there was a point in 1988 after the Democratic convention when Michael Dukakis had a solid polling lead, but by late September it was obvious he was going to lose. Going into election night in 1992 and 1996, I had no doubt that Clinton would win. I never had any doubts that Obama would win either. The only time I was wrong was in 2004, and I think the entire Republican establishment thought Bush had lost after the preliminary exit polls came out. And, in any case, I didn’t know who would win or have a strong opinion about who would win until I got swept up with enthusiasm on Election Day itself. As for 2000, I don’t even know what it means to have predicted the winner. Did Lichtman predict that Bush would lose the popular vote and get saved by a ballot design in Palm Beach County? It seems to me that if you predicted Bush would win, you were more wrong than right.
I really only see 2004 as an election where predicting Bush would win should earn you any analytical credit. Before that (and excluding 2000 for the aforementioned reasons), I think you have to go back to 1980 to find an election where it was difficult to foresee the winner more than a month out from the election.
So, we have a professor who claims to be eight for eight in predicting the outcome of our presidential elections and the average American probably has a record of seven for eight. It’s a small sample to begin with, and the deviation from the norm is doubly insignificant.
So, Lichtman says Trump will win.
I don’t think I owe his opinion any more deference than the man on the street.
If he had predicted in mid-September the winner of the last eight World Series, then I’d be impressed.
Should we start a prediction thread early?
Clinton by 5.4 points. (Followed by 5.4 months of analysis proving she didn’t win sufficiently convincingly, and has no mandate.)
Trump by 1
Clinton +3 (95 percent CI of +1 to +5).
In other words, between 2008 and 2012.
Clinton wins the Obama 2012 states plus North Carolina minus Iowa.
Clinton by 11.7.
I wanted to say +10 but lost my nerve. If you’re within two points, I’ll donate $20 to your favorite charity!
Do you feel this is what a generic Democrat should be winning by over someone like Trump? I just don’t see this as a realistic margin given the polarization and strong third party showing. Let’s see if HRC manages to shore up her soft support but you seem to be setting expectations quite high.
2012 replay; close enough to keep Republicans looking for every last white vote for a while.
No.
Even 1980 was only hard because Reagan was such an outside the mainstream candidate at that point. On fundamentals any warm Republican would have won.
Carter was dead meat and everyone knew it. That’s why Kennedy challenged him. The misery index inflation plus unemployment was around 20 all during 1980. And the Iran hostage crisis was debilitating. Then we boycotted the Moscow Olympics because the Russians invaded Afghanistan. The stench of failure was everywhere.
That’s why the Democrats tried a rule change to release all the delegates from their primary/caucus commitments (same as Republicans tried this year). But with no viable compromise candidate available (Kennedy was unacceptable), it failed.
The historical judgment on Carter (as president) was made in late 1979 and really has never changed.
No disrespect, but you need to try harder:
Internal surveys must have been much worse. The Democratic senators were distancing themselves from Carter and the Senate flipped.
And the problem with the public polling accuracy is self-evident since they missed it by roughly a magnitude of 10.
As I said, Reagan’s image as being outside the mainstream was the smokescreen that kept the establishment/pundits/pollsters in denial about Carter’s demise.
The convention summary:
https://www.campaignsandelections.com/campaign-insider/remembering-1980-kennedy-s-brutal-political-f
istfight
And the misery index:
http://inflationdata.com/articles/wp-content/uploads/2016/08/US_Misery_Index_September_2016.jpg
That’s practically Herbert Hoover territory from an on-the-ground voter perspective.
This is a link to a database with Presidential Election Polling since 1968.
https://docs.google.com/spreadsheets/d/1TA4NktYbU_qtvRKnIaacJOYDIoidLZTabKW5wobUTT0/edit#gid=0
There are three things to note in 1980:
1. One can argue the polling wasn’t wrong, it simply was reflecting a number of events that happened late in the cycle.
And this part frustrates me. In the last 10 days there was a debate (which most think Reagan won) and a weekend development regarding the hostages in Iran.
Events. They change opinion. And yet we are conditions to think that the polls were wrong, and not that people actually changed their minds.
This is a link to the Sunday New York Times before the election. The electoral college map in the front page looks bleak for Carter, and the headline doesn’t match the analysis.
But it also shows the extent to which Iran was dominating the political discussion.
http://timesmachine.nytimes.com/timesmachine/1980/11/02/issue.html
I never saw anyway Carter could be re-elected.
and
frankly, Carter is a pretty unlikable person at the retail level and comes across as petty in a lot of ways at the higher level and did during the campaign.
One fact that confirmed this for me decades later – Carter is about the only person on the planet that Bill Clinton can’t stand.
Which raises Carter a notch or two in my opinion.
And then there was the infamous October surprise. Whether or not true, still contested.
Did you/do you know Carter well enough to make the judgement that he was terrible at the retail level? I met him when he was campaigning for Governor and observed him campaigning for President. I disagree that he was terrible and that he was unlikeable. He was a different sort of politician for his era. He wasn’t an elite, if you will. Certainly not an elite within the Democratic party which is perhaps the basis for saying he had alienated his own party?
I know that we in his home state were proud and thrilled when he won the nomination. I’m still proud of the man that he is and the President that he was. Rosalyn’s not too shoddy either.
Half-way agree. The Beltway elites are snobs. They like wealth and style. The right schools, right family, and right friends. They’ll even accept a decent facsimile like the Reagans. Technically, the Bushes had all the right stuff, but somehow with all their advantages, they didn’t get the high-class fairy dust. So, the Carters didn’t measure up in any way for them.
However, the larger problem for Carter was that he was barely a Democrat. Deregulation got rolling under his administration. Then there was Billy, Bert Lance, and Hamilton Jordan. He also went with Z-Big over Cyrus Vance which led to the Soviet invasion of Afghanistan and that in turn led to the boycott of the Olympics and grain embargo. How many athletes missed their one chance for an Olympic medal? How many farmers were left with grain they couldn’t sell? How many grain handlers and loading dock workers were without work? For what? To train a bunch of jihadis that twenty years later would strike the US? Would the Tehran embassy have been take over if Carter had ignored the Shah’s request to enter the US for medical treatment? Oh, and the ridiculous hostage rescue mission.
Carter is personally a decent man, but … A shame the Democratic Party didn’t have a strong candidate to primary him in 1980 and didn’t have the sense to recognize that Kennedy wasn’t that person.
There is no way to be sure that choosing Vance over Brzezinski would have prevented a Soviet invasion of Afghanistan. Claiming that choosing Z-Big over Vance led to the invasion removes the agency of the Afghan resistance movement against the takeover by the Communists.
8 elections. If they were “fair” coin flips, 4 out of 1000 flips would have correctly predicted all 8 outcomes, on average. Of course, WaPo wouldn’t be writing about the other 996 coins.
Isn’t this article basically another version of the octopus that predicted World Cup match winners?
Sorry, “4 out of 1000 flips…” should read “4 out of 1000 sets of 8 flips…”
Also…it’s 8 goddamn data points. 8!!
Dunno, from the actual article he is actually saying a generic republican would win, which seems likely. As for Trump himself his ‘method’ suggests Dems are almost finished but not quite which accords with the existing polling of HRC a touch ahead.
Not that I’d give him particular deference his ideas just look like a version of the ‘fundamentals.’
Every election, one of these guys gets trotted out to make their (contradictory) prediction, using some model that was exquisitely fitted to previous data.
There was a similar guy for 2012 and yet another one for the 2008 election. They were wrong, which is why I and everyone else has completely forgotten about them and their stupid models.
Yeah. Look, I don’t want to blow off negative predictions out of hand, but when I look at his “16 Keys to the Election”, I don’t know where those keys came from. They certainly wouldn’t be the same “Keys” that would explain the 1988 election. It seems more like Zeitgest reading than a prediction model. How does he operationalize “charisma” in cases like 1988 and 2000 and 2004 where both candidates lacked that quality? Or does he, say, foist a universe on us where Bush I was more charismatic than Dukkakis. I really don’t see a world where one could refine a measure of charisma where it would be clear that one was better than the other in that regard. So we add in “War Hero”. Yeah, Bush I was actually a combat hero. Kerry wasn’t charismatic but was a war hero as well. Bush II was neither charismatic nor a War Hero. At some point it would appear that Charisma and War Heroness doesn’t really apply to elections where it might have mattered.
Not charisma per se, but comparative. I find it easier to measure by first watching the debate with the sound off.
It is a requirement for discussion of Bush the Younger that we accept without dispute that he was a terrible candidate.
This is certainly not true by any objective standard. He beat an incumbent governor, won re-election, and won the nomination as easily as anyone really ever has on their side.
His favorable numbers in ’00 were very good, and his debate performance in the first debate was good enough to change the race.
I’m not saying that he was a terrible candidate. I am looking specifically at these “16 Keys”, scientifically derived that determine electoral outcomes. “Charisma/War Hero” is the combination that the author touts as explaining electoral results. So in 2000, you have “War Hero/Dull” vs “Not War Hero/Dull”. What difference does that “Key” make at all? And no, I don’t think Bush met any definition of Charisma. Yes, he won elections. But you can’t say “Well he must have had charisma because he won elections and therefore charisma explains who wins elections.”
. . . fitted to previous data.”
Wow, is that ever an important, and very underappreciated point!
I well recall the point drilled into me in grad school stats courses: it’s not hard to fit a model to some particular dataset (i.e., the “training” dataset from which the model is derived) that produces a high R-squared and low p-value (i.e., for those lacking statistics background: a model that seems to “explain” the observed data very well).
Problem being that when you then apply the same model to a completely different, but analogous dataset, the “fit” may degenerate into insignificance.
One recommendation to avoid this is to hold back half your dataset, fit the model to the other half, then test the resulting model on the withheld half of the data. If the prediction success on the withheld data is at least significant (it’s not likely to be as good a fit as for the original “training” dataset), then the model probably has some significant explanatory/predictive value. (If not, it likely reflects “overfitting” of independent/predictor variables to the dependent/response variable of interest.) This can be a hard thing for a researcher to discipline her/himself to do, though. Refraining from using all the data you expended considerable time, effort, money, and other resources to obtain — which would usually produce an even better fit for the model being derived from it — requires some self-convincing (at least in my experience).
It’s a good way to avoid professional embarrassment when somebody else tries, but fails, to replicate your result, though!
Sure would help if he disclosed when in the election cycle he made his predictions. Polling data in the week before election day (excluding Gallup) seem to do well n picking the winner — tend to flub a bit on the margins.
Notice he didn’t include 1980.
The only one that I flubbed was 2000. Honestly didn’t occur to me that team Bush could make a 2-5% FL win for Gore disappear.
I have not been wrong since 1980 on a Presidential election. I take some pride in 2012.
I wrote this the day before:
http://www.dailykos.com/story/2012/11/06/1156224/-Obama-gains-ground-in-8-of-10-battlegrounds-clear-
evidence-of-Obama-Surge
And I said in April that Obama was in a stronger position than people thought:
http://www.dailykos.com/story/2012/04/27/1086946/-State-Polling-Summary-Obama-has-a-95-chance-of-bei
ng-re-elected-BUT
But so fucking what. The leader in EVERY election since 1952 who led 10 days after the last convention has has won with the exception of 2000.
No one is better than the polls. There are some things to know:
1. An incumbent President under 45 loses. See 1980. If the right track wrong direction number is bad AND the consumer sentiment number is bad you can predict with good certainty the race will break against the incumbent party.
This is what is a little confusing in this cycle. The right track wrong direction number is awful,GDP is not very good, BUT consumer sentiment is high. So the “fundmentals” are a bit conflicted.
There is a parallel to 1980 in the sense that Trump is an idiot and Reagan was too. People had significant doubts about him. Things were more fucked up in 1980, and Reagan makes Trump look like a Washington, but there is a parallel there.
2. Polling misses of more than 3 15 days out have happened in a MAJORITY of elections. In most cases they haven’t mattered because the races aren’t close.
So basically we are overdue for a surprise. They happen in other countries with regularity, but for a variety of reasons we haven’t had one that mattered.
Think of it this way: there is a 33% chance of a 4 point shift in the last 10 days. Empirically this is true.
So if you haven’t been wrong it is only a matter of time until you will be.
I sure hope the “overdue for a surprise” is postponed for at least another 4 years.The GOP needs to go down, but the rest of us don’t need to go down with them.
I take it as a good sign that my brother, ever the believer in his side winning, sent me a link to this article and started crowing a bit. He truly thought Romney was going to win in 2012 and I didn’t and won’t push back. That’s sort of my personal good luck talisman. Save all that for the day after. Fingers crossed.
Yea, I got the WaPo article in my mailbox and concluded it wasn’t worth a mention as a comment. Prof. Lichtman, or how can I get the media focus on me.
I guess Ted Cruz expects him to win, which probably reflects internal Republican polling. That’s how I read the endorsement. It’s too late for him to gain from it now, and it throws away the “I told you so” button he could wear later. It only makes sense if he expects this guy to become President and ruthlessly crush him for embarrassing him at the convention.
God help us.
Teddy might be looking at the TX power bases. The Bushes loathe him. Gov Abbot has been on-board with Trump for months and he seems to carry more political weight in the state than Perry ever did. So, TPTB might be looking to primary Ted.