The most prestigious law school admissions discussion board in the world.
|
 |
 
|
Nate Silver's thoughtful day-after musings on why the models were so wrong again
Poast new message in this thread
Date: November 6th, 2024 12:42 PM Author: An 8-INT Orc AA'ing their way into mage school I'll link it here if he winds up publishing something (http://www.autoadmit.com/thread.php?thread_id=5628539&forum_id=2\u0026mark_id=5310880",#48300075) |
Date: November 6th, 2024 12:45 PM Author: 38 BMI and Age Cat Mom Sobbing on TikTok (gunneratttt) this outcome is within the first std. deviation of probable outcomes.
not even nate anymore (worse) but 538 has has 297 evs for trump as the third most likely outcome. with 312 being the most likely.
https://projects.fivethirtyeight.com/2024-election-forecast/
how do you think probability works? |
 |
Date: November 6th, 2024 12:47 PM Author: ;..........,,,...,,.;.,,...,,,;.;. I can't find that anywhere (http://www.autoadmit.com/thread.php?thread_id=5628539&forum_id=2\u0026mark_id=5310880",#48300098) |
|
Date: November 6th, 2024 2:16 PM Author: An 8-INT Orc AA'ing their way into mage school "We had Trump winning somewhere between 219 and 312 votes 68% of the time!"
Lmao at this cope. I could have done that too.
|
|
Date: November 6th, 2024 2:25 PM Author: 38 BMI and Age Cat Mom Sobbing on TikTok (gunneratttt) then do it. statisticians at this level make bank. ezpz. you'll be the next lichtman! (http://www.autoadmit.com/thread.php?thread_id=5628539&forum_id=2\u0026mark_id=5310880",#48300980) |
|
Date: November 6th, 2024 2:36 PM Author: LathamTouchedMe What's their value when they can't beat the RCP "no toss ups" map, which got 49/50 states? (http://www.autoadmit.com/thread.php?thread_id=5628539&forum_id=2\u0026mark_id=5310880",#48301069) |
|
Date: November 6th, 2024 2:42 PM Author: 38 BMI and Age Cat Mom Sobbing on TikTok (gunneratttt) do you think that rcp doesn't use probability too? all the "no toss up" map does is average out the polls and assign a win based on polling average. it's a model just like any other, except that it just weighs every poll equally and spits out only the most likely result. other models are more advanced because they weigh polls differently and give you the probability of each result.
hard to tell if you're being obtuse or really just a dunning kruger midwit. but if you don't find value in statistical models no one is forcing you to pay them any mind. |
|
Date: November 6th, 2024 2:53 PM Author: LathamTouchedMe You're the one being obtuse. You don't see the difference between a simple poll aggregator and Silver's or 538's crazy complicated black-box model that includes shit like Fed Reserve economic data and "convention poll bounce smoother"? If they're going to make something more complicated, they're going to have to show why it's worth the trouble.
|
|
Date: November 6th, 2024 2:54 PM Author: DrakeMallardxo (🦆) Silver has jumped the shark into laughably false precision that creates a model with no explanatory power (http://www.autoadmit.com/thread.php?thread_id=5628539&forum_id=2\u0026mark_id=5310880",#48301209) |
|
Date: November 6th, 2024 3:06 PM Author: 38 BMI and Age Cat Mom Sobbing on TikTok (gunneratttt) i never see anyone criticizing this shit putting their money where their mouth is and making their own predictions. it's just "god damn it, most probable outcome didn't happen, fucking thing is useless!"
it's as if i said "if you flip a coin twice there's a 75% chance it won't land on heads twice", and then when it does, you rage about how stupid my prediction was. |
|
Date: November 6th, 2024 3:04 PM Author: 38 BMI and Age Cat Mom Sobbing on TikTok (gunneratttt) "all the "no toss up" map does is average out the polls and assign a win based on polling average. it's a model just like any other, except that it just weighs every poll equally and spits out only the most likely result. other models are more advanced because they weigh polls differently and give you the probability of each result."
you're retarded. that's exactly what i said. what do you think is the difference between a "aggregator" and a "model" aside from the more advanced analysis of the data? yes, it does other things beyond weighing polls, but you didn't seem to even understand that much.
state your most advanced statistics course before i engage you further. |
|
Date: November 6th, 2024 3:17 PM Author: LathamTouchedMe You're deep diving into some semantics about model or aggregator, which I was never arguing about. I'm simply saying, Silver and his colleagues have not presented a good case justifying their far more complicated models. They're adding shit on like a Defense Contractor and it doesn't appear to produce better results. I think the aggregators proved their worth over the past 20 years. It's better than relying on a single poll, like pundits did in the olden days. (http://www.autoadmit.com/thread.php?thread_id=5628539&forum_id=2\u0026mark_id=5310880",#48301407) |
|
Date: November 6th, 2024 3:34 PM Author: 38 BMI and Age Cat Mom Sobbing on TikTok (gunneratttt) you're the one that brought up aggregator versus model.
it does produce results when the models are used with things that can be repeated. the issue is that an election cannot be repeated and tested against the model. you have no idea whether if you ran the election 100 times how the distribution would work out. but there will always be outliers.
in events that can be repeated these models prove how predicative they are. as well as when you consider individual events over a long period of time.
people that just say "wow, an outlier happened, therefore this is crap" don't understand what a stupid statement that is. outliers are *guaranteed* to happen. and in this case, the results were well within silver's most probable results. if you think you can do better, do it, because then you'd be making millions instead of passionately arguing about shit you don't even know the fundamentals of.
hth. |
|
Date: November 6th, 2024 12:46 PM Author: ..................,.....,.,,.,..... Nate: “by the way, I REALLY need bus fare” (http://www.autoadmit.com/thread.php?thread_id=5628539&forum_id=2\u0026mark_id=5310880",#48300092) |
|
Date: November 6th, 2024 3:35 PM Author: infraphysics is hetero (http://www.autoadmit.com/thread.php?thread_id=5628539&forum_id=2\u0026mark_id=5310880",#48301545) |