Indiana GOP/Democratic Primaries

[Clippy]

So when does Kasich drop out ???

 

[Sycho Sid]

Most of those predictions are from early on and, based on polling and the betting odds, were reasonable.You've got a couple Sept. 2015 headlines there.

You actually remind me of democrats that were critical of Silver in 2012 (despite loving him in 2008). He was predicting, based on data, that things would be tighter.

Silver's good at what he does and has the track record to prove it.

And now the more vocal supporters of Sanders are saying he is unreliable.

He picked 2 states wrong.

 

[theBLADE1]

 

[Fawlty]

Most of those predictions are from early on and, based on polling and the betting odds, were reasonable.You've got a couple Sept. 2015 headlines there.

You actually remind me of democrats that were critical of Silver in 2012 (despite loving him in 2008). He was predicting, based on data, that things would be tighter.

Silver's good at what he does and has the track record to prove it.

Silver was screwing the pooch right up until the NE primaries, the criticism is deserved. He was offering bias as analysis. He has to take a knee on this one.

Post #4 in this thread I elaborate a little. http://forums.sherdog.com/threads/r...-to-secure-nomination.3223959/#post-115853133

 

[Jack V Savage]

Silver was screwing the pooch right up until the NE primaries, the criticism is deserved. He was offering bias as analysis. He has to take a knee on this one.

Post #4 in this thread I elaborate a little. http://forums.sherdog.com/threads/r...-to-secure-nomination.3223959/#post-115853133

It's not like personal bias. He was making projections based on evidence, but they didn't work well for this year's GOP primary. I think people would be right to be especially skeptical of projections that aren't based on polls, but it would be a mistake to write off Silver or his methods altogether.

 

[Greoric]

I'm neither here nor there with Trump, but this was fucking amazing. Good kill sir.

 

[Fawlty]

It's not like personal bias. He was making projections based on evidence, but they didn't work well for this year's GOP primary. I think people would be right to be especially skeptical of projections that aren't based on polls, but it would be a mistake to write off Silver or his methods altogether.

I agree it would be a mistake to write him off or his methods. But I have already and can give a number of instances where he did not base his predictions on good evidence or sound logic- such as handing Indiana to Cruz based on a couple instances of GOP collusion and a small Wisconsin upset. He failed to account for Indiana being, frankly, poorer and stupider than its neighbors. I believe he missed that due to his bias, as evidenced by his many anti-Trump statements and articles. Then there is the baseless calling of Montana and I believe South Dakota "Cruz Country" and the similar writing off of West Virginia. He performed poorly.

To his credit, he corrected himself significantly after the New York result (except for Indiana), and even with his bias he was closer to the mark than many of his peers.

 

[in the]

Most of those predictions are from early on and, based on polling and the betting odds, were reasonable.You've got a couple Sept. 2015 headlines there.

You actually remind me of democrats that were critical of Silver in 2012 (despite loving him in 2008). He was predicting, based on data, that things would be tighter.

Silver's good at what he does and has the track record to prove it.

Big surprise - that some headlines are from September. When the race was on. Time and time again this election, Silver had Trump losing - and it's not some recent phenomenon, either. Before the latest primary, he had Cruz and Hillary winning Indiana.

He might be pretty good at what he does, but he was a lot more credible before this election cycle and no amount of spin is going to change that. He's been sorely wrong on so many predictions he's made this primary season and instead of taking ownership of those mistakes, he externalized his own biases factoring into his data instead of looking at things in an objective manner. He is still worth listening to, but the impeccable luster has worn off.

 

[Fawlty]

Big surprise - that some headlines are from September. When the race was on. Time and time again this election, Silver had Trump losing - and it's not some recent phenomenon, either. Before the latest primary, he had Cruz and Hillary winning Indiana.

He might be pretty good at what he does, but he was a lot more credible before this election cycle and no amount of spin is going to change that. He's been sorely wrong on so many predictions he's made this primary season and instead of taking ownership of those mistakes, he externalized his own biases factoring into his data instead of looking at things in an objective manner. He is still worth listening to, but the impeccable luster has worn off.

I wouldn't take that harsh a view- was he really "sorely wrong on so many predictions" or did he just make Trump-related errors (which is kind of like making the same error in multiple ways)?

 

[7437]

So when does Kasich drop out ???

A few hours ago

http://www.cnn.com/2016/05/04/politics/john-kasich-drops-out/

 

[in the]

I wouldn't take that harsh a view- was he really "sorely wrong on so many predictions" or did he just make Trump-related errors (which is kind of like making the same error in multiple ways)?

You ended the quote too early, it answered your question. "He's been sorely wrong on so many predictions he's made this primary season..."
It wasn't just with Trump, either. He had Hillary winning Indiana. But most of his really bad predictions have come via Trump. He has let his bias get in the way of his stats when it comes to the GOP nominee. Not a good look. Like I said, some of the luster has worn off.

 

[Anung Un Rama]

I spit out my soda when Christie showed up on that 300 video. Unbelievably funny piece, an impressive.

 

[Jack V Savage]

I agree it would be a mistake to write him off or his methods. But I have already and can give a number of instances where he did not base his predictions on good evidence or sound logic- such as handing Indiana to Cruz based on a couple instances of GOP collusion and a small Wisconsin upset. He failed to account for Indiana being, frankly, poorer and stupider than its neighbors. I believe he missed that due to his bias, as evidenced by his many anti-Trump statements and articles. Then there is the baseless calling of Montana and I believe South Dakota "Cruz Country" and the similar writing off of West Virginia. He performed poorly.

Without getting under the hood, it's hard to say exactly, but I think the charge that he tweaked models specifically to show Trump doing worse is not credible. I know, for example, early on he noted (and he's not the only one) that very early polls tend to be far less predictive than the endorsement race, which caused him to underestimate Trump's chances. That's an evidence-based conclusion that happened to be wrong this time.

I think generally, we're just seeing that predictions are really hard, especially when it comes to primaries. Silver's not an oracle; just a guy working with the data (what gave him such a big edge when switched from baseball to politics is that unlike with baseball, there was almost no competition in terms of evidence-based analysis). Just as it's not that anyone who failed to foresee, say, Jose Bautista becoming a great hitter wasn't biased against him, Silver failing to predict Trump's success wasn't a result of bias.

 

[Clippy]

 

[Fawlty]

Without getting under the hood, it's hard to say exactly, but I think the charge that he tweaked models specifically to show Trump doing worse is not credible. I know, for example, early on he noted (and he's not the only one) that very early polls tend to be far less predictive than the endorsement race, which caused him to underestimate Trump's chances. That's an evidence-based conclusion that happened to be wrong this time.

I think generally, we're just seeing that predictions are really hard, especially when it comes to primaries. Silver's not an oracle; just a guy working with the data (what gave him such a big edge when switched from baseball to politics is that unlike with baseball, there was almost no competition in terms of evidence-based analysis). Just as it's not that anyone who failed to foresee, say, Jose Bautista becoming a great hitter wasn't biased against him, Silver failing to predict Trump's success wasn't a result of bias.

I'm questioning his inputs. Here's what he said on April 13th that really perked up my ears:

However, there isn’t a lot of recent polling in some states, so instead I’ll make inferences from Trump’s performance in nearby states, the state’s demographics, and so forth, tending to give a fair amount of deference to the panel’s original forecasts.

That sounds very reasonable in the case of Indiana until you consider that they are demographically Trump people, not exactly Midwest people. It is also surrounded by Illinois, Michigan, Ohio and Kentucky- where Trump beat Cruz each time (I don't think that's a major factor- Indiana should be looked at more in a vacuum- but I have to suspect his inputs were wrong here). What very little data was available also favored Trump. It's as if he hedged away Indiana completely, and without reason. It's a prediction he probably should have left blank.

Here is his argument in a nutshell:

"Last month, our panel gave Trump an average of 37 delegates in Indiana, which implied that he’s the favorite to win there. I don’t think I can agree with that after Wisconsin, however. The states are relatively similar demographically. In Indiana, as in Wisconsin, Trump doesn’t have much support from statewide elected officials. Moreover, the Midwest as a whole has been a middling region for Trump. Earlier in the calendar, he got away with some wins in the Midwest with a vote share in the mid-to-high 30s. Now that the field has winnowed and Republican voters have learned to vote tactically, he’ll often need 40 percent of the vote to win a state instead."

He's gifting republican voters a skill of tactical voting, which wasn't true. He's giving too much weight to Indiana being Midwest. He's not giving Trump enough credit for his success in neighboring states. Most importantly, he didn't defer to the panel. And he also should have factored the coming momentum from the northeast- everybody knew Trump would win every state that Tuesday after predictably crushing NY the week after Silver's April 13 predictions. I don't want to do too much criticizing after the fact, but I did point out that he was handing over Indiana for no good reason, and I did point out that the northeast had to change the race- a probability which was largely ignored by experts until April 19th. I have to guess why he messed up Indiana, and I guess that it's because he wasn't objective enough. I submit that any good "deterministic" prediction which had to rely heavily on subjectivity and guesswork must have favored Trump, or must have been biased.

 

[Jack V Savage]

That sounds very reasonable in the case of Indiana until you consider that they are demographically Trump people, not exactly Midwest people. It is also surrounded by Illinois, Michigan, Ohio and Kentucky- where Trump beat Cruz each time (I don't think that's a major factor- Indiana should be looked at more in a vacuum- but I have to suspect his inputs were wrong here). What very little data was available also favored Trump. It's as if he hedged away Indiana completely, and without reason. It's a prediction he probably should have left blank.

Here is his argument in a nutshell:

"Last month, our panel gave Trump an average of 37 delegates in Indiana, which implied that he’s the favorite to win there. I don’t think I can agree with that after Wisconsin, however. The states are relatively similar demographically. In Indiana, as in Wisconsin, Trump doesn’t have much support from statewide elected officials. Moreover, the Midwest as a whole has been a middling region for Trump. Earlier in the calendar, he got away with some wins in the Midwest with a vote share in the mid-to-high 30s. Now that the field has winnowed and Republican voters have learned to vote tactically, he’ll often need 40 percent of the vote to win a state instead."

He's gifting republican voters a skill of tactical voting, which wasn't true. He's giving too much weight to Indiana being Midwest. He's not giving Trump enough credit for his success in neighboring states. Most importantly, he didn't defer to the panel. And he also should have factored the coming momentum from the northeast- everybody knew Trump would win every state that Tuesday after predictably crushing NY the week after Silver's April 13 predictions. I don't want to do too much criticizing after the fact, but I did point out that he was handing over Indiana for no good reason, and I did point out that the northeast had to change the race- a probability which was largely ignored by experts until April 19th. I have to guess why he messed up Indiana, and I guess that it's because he wasn't objective enough. I submit that any good "deterministic" prediction which had to rely heavily on subjectivity and guesswork must have favored Trump, or must have been biased.

You're not necessarily proven wrong when you say that an event is unlikely and it happens anyway, and his argument seems fine to me. Your point at the end is fine (that is, you can certainly argue that a more qualitative analysis should have favored Trump more), but that just goes to the limitations of his methods. To take another baseball example, look at Hanley Ramirez. PECOTA wasn't super high on him because he didn't put up great numbers in the minors, but anyone who watched him could see that he was a monster talent. It doesn't mean that PECOTA is wrong or biased; it's just that it isn't able to incorporate all relevant information (and BTW, baseball scouts are *way* better at qualitative analysis than political pundits).

 

[UpaLoompa]

You're not necessarily proven wrong when you say that an event is unlikely and it happens anyway, and his argument seems fine to me. Your point at the end is fine (that is, you can certainly argue that a more qualitative analysis should have favored Trump more), but that just goes to the limitations of his methods.

Look, Silver was wrong on Trump for valid reasons. That invalidates his entire approach despite its demonstrated success.

What don't you get?

 

[UpaLoompa]

One thing I often like with Silver and his people is when they explain why their models were wrong in their predictions. That's often more informative to me.

"All models are wrong but some are useful"

 

[Fawlty]

Look, Silver was wrong on Trump for valid reasons. That invalidates his entire approach despite its demonstrated success.

What don't you get?

I wasn't saying that at all. I'm just saying I think he was biased against Trump and showed how, in detail, I came to that conclusion. That would be an exception to his usual level-headedness, and not discrediting or discounting him. Maybe there's a touch of idolatry happening here?

 

[UpaLoompa]

I wasn't saying that at all. I'm just saying I think he was biased against Trump and showed how, in detail, I came to that conclusion. That would be an exception to his usual level-headedness, and not discrediting or discounting him. Maybe there's a touch of idolatry happening here?

He's demonstrated to be very good at what he does. That's data driven rather than idolatry. Many of the examples of his bias (your example of indiana demonstrates him being wrong, not biased) date back to when the betting lines were saying the same thing he was.