That guy, geeze. I like to think I’m a statistician and I … I just wish I had that kind of unearned confidence once and a while.
3 Likes
@beschizza it’s worth noting that Clinton’s lead in June 2016 was almost exactly the same size as Biden’s is now.
1 Like
I’m all for data and facts and such, but I find poll-driven election coverage exhausting and pointless. Once the two candidates are selected, there’s literally no sense in which polling data makes you better informed.
The only reason polls get covered at all is that they’re such a reliable source of vacuous word count, and an evergreen pretext for opinion-as-news pieces. Where the opinion is nearly always “this election will be too close for you to relax and stop consuming news for a single moment”.
To my mind, the fact that every election is always close to 50/50 is highly suspicious. How come a presidential race never breaks down as like 70/30? If you really think about it – and the possibility that it’s because the media forces races to be close – the implications are profound and not in a good way.
1 Like
Larger!
https://twitter.com/Beschizza/status/1269741072030347264
Sadly, general elections are not held in June.
4 Likes
I guess the argument is, as more and more information becomes available, campaigns make fewer mistakes, and every chess game becomes nearly a draw
For one candidate to get 70%, the other side would have to screw up very badly and get 30%
A more cynical way to look at it is, the politicians really work for big campaign donors, and they don’t want to give voters anything they don’t have to, to get that 51% and win
There’s no reason for them to run up the score and enact policies people actually like
2 Likes
That’s kind of tautological since it’s a zero-sum game, but I understand the idea - this is a kind of hungry hungry hippos model, where candidates run around gobbling up undecided voters, and there’s no reason why one competent candidate should find it harder than another. That makes sense if you imagine a Norman Rockwell picture where two blandly polished candidates with roughly identical politics go across the country shaking one hand at a time.
But even then, it seems numerically unstable; the less different the candidates are, the smaller the differences people will decide on, and the more one candidate is perceived as having a lead, the more that in itself will distinguish them. This would arguably drive the result away from being close.
And of course, the candidates might be very far from identical (e.g. Clinton and Turmp); in that case, a close race requires that two different offerings just happen to have the same number of supporters, which you would not in general expect to be true.
Suppose there was a competition where two different teams had to pick a $5 Secret Santa gift, and the whole population picks which they’d rather receive. If it’s possible to find a single perfect choice, then two competent teams will offer the exact same thing, making it a simple coin toss, and you’d expect consistent 50/50 outcomes. However, what if it’s not that easy, because it’s a multi-variable optimisation with imperfect data? Maybe one team offers a $5 hat, and the other a $5 mangosteen. Then you would expect the outcome to be normally distributed: 50% is the mean split, but you’d mostly expect anything in the 50-60% range, and sometimes 70%, and occasionally 90%.
I ramble, but it seems like almost however you model this, the results we see can only be explained by some mechanism that actively prevents any candidate gaining a clear lead. Which amounts to saying there is a (strong) systemic bias against any candidate who is clearly the better choice.
Fun trivia tidbit: since 1992, the Republican presidential candidate received more national votes than the Democrat candidate only once.
2 Likes
He isn’t even a statistician.
I’d trust him to do statistics like I’d trust a statistician to do neurosurgery.
1 Like
Indeed, published CNN polls indicate the foregone conclusion that Biden will prevail, ultimately, just like his opponent Clinton was to win the 2016 campaign, based on published polling analysis, and by CNN, naturally, which was later explained away as fake news, a term, un-ironically and mockingly coined by the victor, current President Trump, who happens to be an asshole. “Fake News”, the headline plastered across the webpage banners of CNN on the day after? Just get the vote out, people.
1 Like
I would rather say don’t let hope lead to complacency. Rather, let it fuel you, give you the motivation to beat expectations.
3 Likes
The nearly 50/50 split is what you would expect with a first past the post system with two strong parties and repeated occurrences. The idea even has a name, the Hoteling-Downs model. If you imagine a linear spectrum of voters from the most radical leftist to the most extreme right wingnut, each candidate selects a position to try to maximize votes. If one candidate moves away from the center mass of the voters, they will lose votes on average, because the other candidate can move right next to them and pick up the abandoned votes. That loss tends to push the party back to the center. Even with that we have the Reagan/Mondale race which was nearly 60/40. We also had sizeable gaps in the 1992 and 1996 elections, but not as extreme as that.
2 Likes
1992 was 43% for Clinton vs 37% for Bush and 19% for Perot. 1996 was 49% for Clinton vs 40% for Dole and 9% for Perot.
I think Dole was just an empty suit candidate. I know who he is, obviously, but I don’t know why anyone would vote for him.
1 Like
Sure. I’m not saying that there is only active voter suppression just on election day, but that the election day results show the success of orchestrated, long term voter suppression tactics, including questionably targeted voter roll purges, gerrymandering, lack of early voting, lack of mail in ballot access, lack of planning for sufficient polling sites, lack of placement for polling locations near people w/o transportation, etc.
Does that model factor in voter suppression & disenfranchisement? Because that’s how the GOP has hacked the system. They don’t have to convince the majority of voters to vote for them; they just have to convince the majority of voters not to vote (or prevent them from voting, selectively).
It would be the middle of whatever electorate is left. No matter what other forces are applied to the electorate you should expect the parties to fall near the middle of those who vote. That won’t represent those who might be eligible in an ideal system, but will still give you numbers in the 50% ballpark.
1 Like
This topic was automatically closed after 5 days. New replies are no longer allowed.
