- My Forums
- Tiger Rant
- LSU Recruiting
- SEC Rant
- Saints Talk
- Pelicans Talk
- More Sports Board
- Fantasy Sports
- Golf Board
- Soccer Board
- O-T Lounge
- Tech Board
- Home/Garden Board
- Outdoor Board
- Health/Fitness Board
- Movie/TV Board
- Book Board
- Music Board
- Political Talk
- Money Talk
- Fark Board
- Gaming Board
- Travel Board
- Food/Drink Board
- Ticket Exchange
- TD Help Board
Customize My Forums- View All Forums
- Show Left Links
- Topic Sort Options
- Trending Topics
- Recent Topics
- Active Topics
Started By
Message
re: Sensing a little bit of unease from Nate Silver today
Posted on 10/27/20 at 11:46 am to nc_tiger
Posted on 10/27/20 at 11:46 am to nc_tiger
quote:And the end game showed that he was wrong.
and at another point before Election Day he had the odds near 50/50
Probability is irrelevant if you cannot repeat the event and test the hypothesis.
There is no “less wrong” when it comes to this.
If Nate Silver states that Trump has a 1% chance this election and Trump wins, he is equally as wrong as he was in 2016.
His probability narrative is simply a way to give himself an “out” if his predictions are wrong.
Nate giving anything above 0% odds means that he will always be right, no matter the outcome.
Hence why probability is nothing but a propaganda tool, and it is one of my biggest issues with Nate Silver.
This post was edited on 10/27/20 at 11:48 am
2
Posted on 10/27/20 at 11:50 am to Flats
quote:
How do you "ace" an election if all you give are percentages?
easy: the states with higher percent chance of going Obama... actually went Obama. the model aced the winner-by-state predictions, that doesn't mean it was quantitatively perfect in the odds it gave each race.
quote:
Since you claim to know how statistics work, you know that 2 iterations of something this complex tell us nothing.
538 has done a lot more than the 2012 and 2016 presidential elections... c'mon you know I'm just pointing those out as examples worth highlighting in the current context.
Posted on 10/27/20 at 11:52 am to Scruffy
quote:Jesus
Hence why probability is nothing but a propaganda tool
Posted on 10/27/20 at 11:56 am to nc_tiger
quote:
to build a statistical model that's based on a set of priors and available polls to predict the probability of election outcomes, not the outcome it self.
The issue being those "priors and available polls" are flawed to begin with.
Garbage in, garbage out.
Posted on 10/27/20 at 11:58 am to Powerman
quote:I have already stated that I think Biden wins, but these probability factors mean very little.
Jesus
This isn’t like batting average or poker hands.
There is no way to test the validity of the results or the accuracy of the data.
And like I said, Nate can give any number and still claim he was correct.
Hell, he was wrong in 2016, yet still is lauded as the cream of the polling crop because people believe that he was “less wrong” than everyone else.
He would be equally as “correct” in his mind if he gives Trump a 10% chance, and Trump ends up winning.
The very concept itself is flawed.
This post was edited on 10/27/20 at 11:59 am
Posted on 10/27/20 at 11:59 am to nc_tiger
quote:
easy: the states with higher percent chance of going Obama... actually went Obama.
Then by that metric he blew 2016. You can't say when his percentages were right they count and when they were wrong they were still above zero so it's not that bad. He either nailed it or he didn't, so in presidential elections he's at 50%.
Anybody can claim their model predicts this presidential race accurately and assign a non-zero percentage to each candidate. Silver's model may be the greatest thing since sliced bread. My model might too. We won't know until we've got a lot more than 2 iterations to judge by.
Posted on 10/27/20 at 11:59 am to Centinel
quote:
The issue being those "priors and available polls" are flawed to begin with.
of course. the pollsters (and downstream models like 538) claim to have made adjustments after learning things post-2016 election. we'll see on nov 3
Posted on 10/27/20 at 12:00 pm to nc_tiger
quote:
538 has done a lot more than the 2012 and 2016 presidential elections... c'mon you know I'm just pointing those out as examples worth highlighting in the current context.
538 was working with the NYT from 2010 to 2013 when it was bought by ESPN in 2013 and placed under the ABC News umbrella in 2018. Which means Nate’s boss is George Stephanopolous
Why would we think anything about this dude’s product is genuine?
Posted on 10/27/20 at 12:01 pm to Scruffy
quote:
This isn’t like batting average
It's exactly like batting average.
If you look at a new MLB player and compute his batting average after his first 3 at bats.
Posted on 10/27/20 at 12:02 pm to Flats
quote:Exactly.
You can't say when his percentages were right they count and when they were wrong they were still above zero so it's not that bad. He either nailed it or he didn't, so in presidential elections he's at 50%.
That is what annoys me the most about those who defend his models.
There is no “less right” in a one off event.
You are either right or you aren’t.
Stating that someone had a very low chance of winning and having that person win, doesn’t mean anything when you cannot reproduce the scenario.
The answer is either a “yes” or a “no”.
Posted on 10/27/20 at 12:02 pm to Scruffy
The problem is you are looking at it as an all or nothing approach. If someone says that x has a 66% probability of happening and it doesn't happen you think they're "wrong" which says everything about your own stupidity. Low probability events happen all the time.
Posted on 10/27/20 at 12:04 pm to Scruffy
quote:
Probability is irrelevant if you cannot repeat the event and test the hypothesis.
Emotional right wingers apparently can't think clearly.
Look, if I have one dice roll and I need a 6, I have only a 16.7 chance of rolling a 6.
Does that mean I will never roll a 6? NO.
Does that mean if I roll a 6 the statistician who pinned my odds at 16.7 percent was wrong? NO.
Sometimes when the random stuff happens (like actually rolling the dice or millions of people voting), you roll the 6. Usually you don't. But that doesn't mean it's impossible and it doesn't invalidate the statistical odds if the less likely thing occurs.
In the last election Trump was given right around a 30% chance of winning. That's almost as much as saying he would win if he rolled 5 or 6. That's pretty good odds.
This time he's got to roll a 6. That's still not impossible.
Posted on 10/27/20 at 12:04 pm to Powerman
quote:
The problem is you are looking at it as an all or nothing approach. If someone says that x has a 66% probability of happening and it doesn't happen you think they're "wrong" which says everything about your own stupidity.
It says something about your own stupidity that you think we know that the 66% probability in your hypothetical is accurate.
Posted on 10/27/20 at 12:05 pm to TBoy
quote:
Emotional right wingers apparently can't think clearly.
Look, if I have one dice roll and I need a 6, I have only a 16.7 chance of rolling a 6.
Does that mean I will never roll a 6? NO.
Does that mean if I roll a 6 the statistician who pinned my odds at 16.7 percent was wrong?
Think really hard and you'll see the problem with your analogy.
Posted on 10/27/20 at 12:06 pm to Flats
in 2016 538 had 4 or 5 states at > 50% hillary that she lost. I think it was 4 or 5 I can't remember exactly and I'm too lazy to look. Therefore I'd say the model performed poorly in comparison to how it performed in 2012. Not a 50/50 metric. I'm basing my opinion of 538, which I would level at "reputable", on more than just presidential elections, but everyone is free to make their own judgements.
Posted on 10/27/20 at 12:12 pm to TBoy
quote::facepalm:
Look, if I have one dice roll and I need a 6, I have only a 16.7 chance of rolling a 6.
Does that mean I will never roll a 6? NO.
Does that mean if I roll a 6 the statistician who pinned my odds at 16.7 percent was wrong? NO.
Sometimes when the random stuff happens (like actually rolling the dice or millions of people voting), you roll the 6. Usually you don't. But that doesn't mean it's impossible and it doesn't invalidate the statistical odds if the less likely thing occurs.
In the last election Trump was given right around a 30% chance of winning. That's almost as much as saying he would win if he rolled 5 or 6. That's pretty good odds.
This time he's got to roll a 6. That's still not impossible.
You just proved my general statement.
Using dice, you have 6 possible outcomes, with each afforded the exact same percentage.
You can repeat the rolling of dice and if you do it for eternity, the numbers will end up with essentially equal 16.7% for all faces.
With the election, you have two possible outcomes.
I cannot repeat this event.
A probability prediction cannot be validated.
People can make whatever probability predictions that they want, but it comes down to a “1” or “0”, win or lose.
In a singular event, there is no “less right”.
It is right or wrong. Probability is irrelevant and means nothing in this scenario.
I mean, how is stating that Trump had a 1% chance in the previous election any different than stating that he has a 30% chance?
Both predictions are correct.
This post was edited on 10/27/20 at 12:16 pm
Posted on 10/27/20 at 12:12 pm to the808bass
All those numbers suggest precision, but he might as well be a an Augur reading entrails. There’s so much human input into his system.
Your gut is probably a more reliable predictor.
Your gut is probably a more reliable predictor.
Posted on 10/27/20 at 12:17 pm to nc_tiger
quote:
which I would level at "reputable", on more than just presidential elections,
Based on what?
The fact he rates "YouGov" as a reliable poll should tell you everything you need to know about Nate Silver.
Posted on 10/27/20 at 12:19 pm to tigerskin
Louisiana folks know about the term crawfishing I am sure
probably came up with it
probably came up with it
Posted on 10/27/20 at 12:19 pm to Powerman
quote:
Low probability events happen all the time.
When you have controlled parameters and a set game with known variables.
Too many unknowns in polling nowadays.
This post was edited on 10/27/20 at 12:22 pm
Page 4 of 6
Popular
Back to top
Follow TigerDroppings for LSU Football News