U.S. Election - Because we know better than you do

[OnionAhaiReborn]

4 minutes ago, Fez said:

Five elections does not a valid sample make. Neither do the three I mentioned either of course.

We can't say for sure that February polling for the general election is "absolutely worthless" but we can certainly say that we have absolutely no idea what its value is.

Ok, then:

"In a comprehensive analysis of elections between 1952 and 2008, Robert Erikson and Christopher Wleizen found that matchup polls as early as April have generally produced results close to the outcome in November. "

That's a large sample. If April polls have value it's reasonable to assume that polls a month earlier have some value too. In fact, it would be pretty crazy to imagine there was some threshold in March that turned polls from useless to predictive.

And, again, it's a long article with many different arguments (it directly addresses the case of Dukakis, for instance). You should take the time to read it.

[Fez]

5 minutes ago, OnionAhaiReborn said:

Ok, then:

"In a comprehensive analysis of elections between 1952 and 2008, Robert Erikson and Christopher Wleizen found that matchup polls as early as April have generally produced results close to the outcome in November. "

That's a large sample. If April polls have value it's reasonable to assume that polls a month earlier have some value too. In fact, it would be pretty crazy to imagine there was some threshold in March that turned polls from useless to predictive.

And, again, it's a long article with many different arguments (it directly addresses the case of Dukakis, for instance). You should take the time to read it.

But April is very different from February. By April nominations are usually locked up whereas in February things are still contested and supporters of candidates are sometimes pretty salty about the idea of supporting one of the other ones.

Also, even that's only 13 elections. You generally want a sample of at least 60 to start having any statistically validity, and even than its going to be pretty low.

ETA: maths

[A Horse Named Stranger]

Well, but this election cycle seems to be a bit off. At least compared to the previous US election I more or less followed. The splintering on the right has reached a new low (or high), and with a self-proclaimed socialist on the left with a fair chance of winning the Democratic ticket (and early signs of splintering on the left). So I am really not sure how that circus compares to any previous election. Are you really willing to bet, that the old common wisdom is still applicable here?

[DanteGabriel]

2 minutes ago, TerraPrime said:

In 2004, when Kerry was running against the incumbent W. Bush, the bishop (or archbishop?) of his region made a point saying that politicians who support the right to choose on abortion are not true Catholics, and that the priests who lead the congregation to which such politicians belong should contemplate whether they ought to be denied communion.

Rightfully, imo, that caused a minor uproar amongst Kerry supporters because it was seen as using religion to pressure politics. 

The Pope declaring that building a wall to stop immigration as anathema to being a Christian differs only slightly from that example, and that difference is insufficient to warrant immunity from similar criticism imo.

And, to be fair, of course the Pope is entitled to make statements about what is and isn't "true" Christian, since last I checked, the RCC still declares that they are the one true Church for Christ and that all salvations must come through them. I don't expect the RCC to abandon that point of view any time in the next 2 millennia. I should have been more clear to say that the Pope is free to remind people that he presumes to know what is a true Christian and what isn't, just like any number of other spiritual leaders, from Pat Robertson to the local street preacher, do routinely. And just like in those cases, the rest of us are free to disregard these proclamation and make up our own minds on what defines a Christian. While the RCC does have a lot more followers than many other types of Christian faith, from the perspective of a non-believer, what the Pope says on the issue matters no more, and no less, than any other religious figures' similar statements.

There were a handful of American bishops who said stupid things about John Kerry, and Pope Francis demoted the most outspoken of these.

I'm comfortable cheering Pope Francis's weighing in on the border wall and his throwing shade on trickle-down economics, even while I thought the bishops critical of Kerry were asshats. Maybe I'm just such an irredeemable team partisan that I take a small amount of satisfaction in seeing an influential religious leader advocating for positions my party holds, after decades of watching evangelical leaders nakedly shilling for the GOP. I can live with that inconsistency.

[Kalbear]

Ok, then:

"In a comprehensive analysis of elections between 1952 and 2008, Robert Erikson and Christopher Wleizen found that matchup polls as early as April have generally produced results close to the outcome in November. "

That's a large sample. If April polls have value it's reasonable to assume that polls a month earlier have some value too. In fact, it would be pretty crazy to imagine there was some threshold in March that turned polls from useless to predictive.

And, again, it's a long article with many different arguments (it directly addresses the case of Dukakis, for instance). You should take the time to read it.

The problem there is that this election really isn't like any of the others; it's been a long time since both parties did not have front runners that were well established.

Also, in 2000 the polls showed gore winning. The polls in 2004 showed Kerry winning. Both by small margins. The reason Obama won in 2008 and especially 2012 has little to do with national polls and a lot more to do with demographics at state levels than they did national averages.

Like I said, the polls in Feb coinciding with the winners by 2% points speaks more to dumb luck, and to the fact that the race is close, period.

[OnionAhaiReborn]

Just now, Fez said:

But April is very different from February. By April nominations are usually locked up whereas in February things are still contested and supporters of candidates are sometimes pretty salty about the idea of supporting one of the other ones.

Also, even that's only 9 elections. You generally want a sample of at least 60 to start having any statistically validity, and even than its going to be pretty low.

Essentially you are now calling into question the validity of any work done by political scientists on the question of polling and US Presidential elections (like the work cited in the Jacobin article), on the basis that they do not ever have enough evidence. I'm inclined to trust that they know a little something about the work they're doing, YMMV.

3 minutes ago, Kalbear said:

The problem there is that this election really isn't like any of the others; it's been a long time since both parties did not have front runners that were well established.

Also, in 2000 the polls showed gore winning. The polls in 2004 showed Kerry winning. Both by small margins. The reason Obama won in 2008 and especially 2012 has little to do with national polls and a lot more to do with demographics at state levels than they did national averages.

Like I said, the polls in Feb coinciding with the winners by 2% points speaks more to dumb luck, and to the fact that the race is close, period.

You say it's dumb luck. The evidence cited above suggests some real relationship. As I said above, you can accept that these people have some idea what they're doing or be skeptical off the bat. YMMV.

Also, once more and to everyone- there are several arguments in that article and it's worth reading.

[Hereward]

3 hours ago, MerenthaClone said:

Yeah, okay guy.  I thought that's why we paid for having 10 supercarriers to the rest of the world's grand total of 2 (the UK's, neither of which are seaworthy at the moment).  We also have two more under construction.  So, uh, who is doing the invading? 

I can dream.

[Kalbear]

I'm talking about the February ones, not the April ones. I agree that their conclusion is probably accurate - that partisanship is going to play a more clear role in results. The problem is what you're doing, which is applying the results of multiple undecided primaries to the 5 prior elections, all of which were significantly different.

Another way to say it is this: if their conclusion about partisanship is right then the current polls are not useful. If they are wrong then the current polls might or might not be useful.

[Fez]

22 minutes ago, OnionAhaiReborn said:

Essentially you are now calling into question the validity of any work done by political scientists on the question of polling and US Presidential elections (like the work cited in the Jacobin article), on the basis that they do not ever have enough evidence. I'm inclined to trust that they know a little something about the work they're doing, YMMV.

Actually I call into the question the validity of any work done by political scientists. Period. Even other social scientists laugh at how little rigor political science has, and that's saying something considering how much disdain hard scientists have for social scientists.

[OnionAhaiReborn]

2 minutes ago, Kalbear said:

I'm talking about the February ones, not the April ones. I agree that their conclusion is probably accurate - that partisanship is going to play a more clear role in results. The problem is what you're doing, which is applying the results of multiple undecided primaries to the 5 prior elections, all of which were significantly different.

Another way to say it is this: if their conclusion about partisanship is right then the current polls are not useful. If they are wrong then the current polls might or might not be useful.

That is not what I'm doing- that is the Matt Karp's relaying of the evidence on February polling since 1996 from the work of political scientists Robert S. Erikson and Christopher Wlezien. He could be relaying it inaccurately. They could be wrong in some crucial way. Accept or be skeptical. Agree to disagree. YMMV.

2 minutes ago, Fez said:

Actually I call into the question the validity of any work done by political scientists. Period. Even other social scientists laugh at how little rigor political science has, and that's saying something considering how much disdain hard scientists have for social scientists.

Good to know! I completely disagree with you.

[James Arryn]

4 minutes ago, Fez said:

Actually I call into the question the validity of any work done by political scientists. Period. Even other social scientists laugh at how little rigor political science has, and that's saying something considering how much disdain hard scientists have for social scientists.

This statement could be made about any social science. If that's your rubric, and you're throwing out all social sciences, I'd call that ill-advised. I mean I get the argument, but it's pretty narrow-minded.

[alguien]

1 hour ago, OnionAhaiReborn said:

"As partisan polarization has increased over the last three decades, there’s some evidence that early polling has become more predictive than ever.

Not according to Nate Silver or Jill Lepore, anyway. 

[lokisnow]

Fez and Kal, you both would love this website unskewed polls it is explicitly for people cognitively cousins of yourselves: it's a place for people to gather when they don't like the data a poll gathers and want to wishfully invent data they would prefer instead, exactly like you both are doing.

Everyone there is very fond of phrases like "that can't possibly be right" and "throw out the baby with the bath water"