Alright, I'm gonna push back a bit here. Obviously the monikers we attach to political figures are tongue-in-cheek (Queen, Daddy, etc), but the Trash Nate and GodNate stuff I find weird and a little "console war-esque".
It is without a doubt that Nate screwed up with Trump by not deferring to the data. It is without a doubt that the model he's created for 2016 is more reactive than others, even by his own admission. He's explained himself a few times that he chose a very reactive model due to what he believes is a more volatile election (perhaps overcompensating for the Trump effect he missed previously) and due to the high number of undecideds. We can definitely debate the inclusion of certain polls and their various weights assigned, but Nate has been very open about his firm belief in including just about every poll (probably due to his strong 3rd party-type sympathies).
I'm by no means a statistical expert. Personally, I just received my Masters in Epidemiology and Biostatistics, and I'm very familiar with STATA, the software Nate does most of his work on. Now, retrospective infection surveillance is very different from prospective prediction modeling, so I'm typically really hesitant to critique a field I'm unfamiliar with. I love that PoliGAF really digs into the crosstabs and such. I encourage more people to learn more about statistics, without a doubt, but I also ask that we have a little bit of patience. Science is often about scrutiny, not dismissal, until results are seen. 538 will not decide the election, so it's all just methodology debates.
The fights between Silver, Cohn, Wang, etc have been fascinating this cycle. We are currently watching really smart people debate over a field that is still essentially brand new about how to approach these types of models. Silver has been fairly open about what he's doing and we'll see on Nov 8th how close he really is. Rewarding and giving voice to those other individuals has been really cool, this cycle. But picking teams just feels silly to me (outside of tongue-in-cheek stuff).
Again, I'm not saying we can't critique him. His choices this cycle are definitely suspect, and Nate has a big, big smug contrarian streak in him. I, personally, just refuse to dismiss him as "Trash Nate" like he's Sony or Nintendo or something just because he's trying something a little more sketchy than his peers or because his numbers are less reassuring. At least until we see more direct evidence that he's incorrect.
It is without a doubt that Nate screwed up with Trump by not deferring to the data. It is without a doubt that the model he's created for 2016 is more reactive than others, even by his own admission. He's explained himself a few times that he chose a very reactive model due to what he believes is a more volatile election (perhaps overcompensating for the Trump effect he missed previously) and due to the high number of undecideds. We can definitely debate the inclusion of certain polls and their various weights assigned, but Nate has been very open about his firm belief in including just about every poll (probably due to his strong 3rd party-type sympathies).
I'm by no means a statistical expert. Personally, I just received my Masters in Epidemiology and Biostatistics, and I'm very familiar with STATA, the software Nate does most of his work on. Now, retrospective infection surveillance is very different from prospective prediction modeling, so I'm typically really hesitant to critique a field I'm unfamiliar with. I love that PoliGAF really digs into the crosstabs and such. I encourage more people to learn more about statistics, without a doubt, but I also ask that we have a little bit of patience. Science is often about scrutiny, not dismissal, until results are seen. 538 will not decide the election, so it's all just methodology debates.
The fights between Silver, Cohn, Wang, etc have been fascinating this cycle. We are currently watching really smart people debate over a field that is still essentially brand new about how to approach these types of models. Silver has been fairly open about what he's doing and we'll see on Nov 8th how close he really is. Rewarding and giving voice to those other individuals has been really cool, this cycle. But picking teams just feels silly to me (outside of tongue-in-cheek stuff).
Again, I'm not saying we can't critique him. His choices this cycle are definitely suspect, and Nate has a big, big smug contrarian streak in him. I, personally, just refuse to dismiss him as "Trash Nate" like he's Sony or Nintendo or something just because he's trying something a little more sketchy than his peers or because his numbers are less reassuring. At least until we see more direct evidence that he's incorrect.
