- 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: Nate silver: the slowdown in CV19 cases in NOLA is unlike anything we have seen elsewhere
Posted on 4/13/20 at 5:22 pm to moneyg
Posted on 4/13/20 at 5:22 pm to moneyg
quote:
It is you who has a fundamental misunderstanding of probabilities.
quote:Polls are known to be "off". On average by a margin of about 2%, if I'm not mistaken. This is well-known and accounted for.
If you don't think the polling was off in the 3 main states we've been discussing, I don't know what to tell you.
quote:Of course they would have been different. But on what grounds would you justify randomly moving those 3 states to 50/50?
If you moved those states to 50/50 contests, Silver's overall prediction would have been much, much different.
quote:You seem to think that randomly changing data points based on feelings makes a probability more "right".
You think naming something a probability means the probability value it is scientifically valid. It's not.
1
Posted on 4/13/20 at 5:25 pm to Antonio Moss
quote:
You can reasonably infer that it would higher. Plus, at the time you had the ongoing saga in Italy to refer to.
Is that the standard? Higher than any flu season? You think that justifies Trillions in economic damage?
If you don't mind me asking, what line of work are you in?
quote:
Less important
It's critical to making a sound decision. With Trillions at stake, it's absolutely critical to understand what that number is. I realize it isn't an exact number. But, we need to know.
For example, if we ended up cutting our overall deaths from 200K to 50K, this was a collosal mistake.
quote:
Do you honestly believe that the U.S. was going to be the only country in the world to no implement any level of social restrictions?
Any level? We're discussing whether or not shutting down the economy was justified. Don't move the goalposts.
quote:
Depends on who "We" is.
The United States.
quote:
There is no federal mandate; only recommendations.
Irrelevant. The national guidelines are pushing the state level shutdowns. And, the shutdowns are in line with what the Federal government is recommending. Moreover, the Federal government hasn't modified or clarified its guidance to suggest that states were going overboard...even as damage is done to our economy.
quote:
State governors have taken the lead on this - as they should - and there have been different levels of restrictions. I believe some states have gone overboard.
Isolating hot zones and focusing on metro areas going forward should be the path along with rapid testing and tracking.
That said, I will give governors a little slack because we haven't had the ability to rapidly test until this week and the misinformation from the WHO really damaged our preparation timetable.
This is reasonable.
But, even if the missteps (presumably) were understandable, they are still missteps. And, we've already taken a huge hit economically due to the guidance provided at the national level. That guidance was based on something. The sooner we figure out that what we based it on was off the mark, the faster we can adjust course.
Every day that goes by increases the damage.
If we've overreacted, then the guidance was either based on a bad model, or a model with very bad data. The only thing that would be worse is if it comes out that they didn't even consider what the guidance was based on. Instead, they just offered guidance because distancing was better than not distancing.
Posted on 4/13/20 at 5:31 pm to Korkstand
quote:
Because that's exactly what you're doing.
You must be unable to read.
quote:
Poll data is known to have flaws. Do you think the 538 model does not account for this?
What are you saying? That the 538 model accounts for non-representative sampling?
quote:
Again, just because an underdog wins doesn't mean the odds were underestimated.
It sure as hell doesn't mean the odd were accurate, correct?
So, the question is whether or not you think, at the time of the election, Michigan, Pennsylvania, and Wisconsin were closer to 50/50 contests or 80/20 contests in favor of Hillary? What is your opinion?
Posted on 4/13/20 at 5:34 pm to moneyg
quote:Yeah, and it happens in every single poll. Again, this is known and accounted for as best as the data allows.
anyone who understands sampling and the requirements to come up with a valid probability off of said sampling knows this is true. If the sampling wasn't representative of the actual population, you get skewed results. That's exactly what happened.
quote:Well, what are you suggesting 538 should have done? Just drastically alter poll results because "reasons"? I mean, you keep saying shite was wrong and flawed, but you haven't even attempted to explain what could have been done better. Or why basically no one else in the world, including Trump's own campaign, wasn't giving him better odds.
This is the guy who just said I was misrepresenting is comments.
quote:I couldn't help but notice that you didn't answer the question.
This is a PERFECT example of you limited logic skills.
If that's ALL we had to go on (national polling), and that's the only data that was available, you'd get odds that weren't close to even odds in favor of the candidate winning the national polling. That doesn't make the model valid because...
If the actual data showed that the difference in votes was based in two states, and ALL OTHER STATES were polling the other way, the model you reference would be very wrong. And, a more accurate model (based on electoral college) would put high odds in the opposite direction.
This should be obvious.
Instead, you tried to call out my "limited" logic skills by going on to lay out an even more ridiculous example.
I know how polling works. I know how the EC works. I know how polling and margins of error are applied to project state by state outcomes, and how that relates to EC votes.
But, more importantly, I also understand probability in general.
If we for some random reason had moved the 3 states in question to 50/50, can you calculate the odds of Trump winning all 3 again?
Posted on 4/13/20 at 5:36 pm to moneyg
quote:
Is that the standard? Higher than any flu season? You think that justifies Trillions in economic damage?
You just jumped to several assumptions. I never opined on the economic ramifications of the shutdown. In fact, I don't think I have ever posted an opinion on that.
The conversation in this thread was about the validity of the models and the false narratives surrounding them.
quote:
If you don't mind me asking, what line of work are you in?
I'm not answering that.
quote:
It's critical to making a sound decision. With Trillions at stake, it's absolutely critical to understand what that number is. I realize it isn't an exact number. But, we need to know.
You don't think the WH and governors had numbers in from of them?
quote:
For example, if we ended up cutting our overall deaths from 200K to 50K, this was a collosal mistake.
That is your opinion.
quote:
Any level? We're discussing whether or not shutting down the economy was justified. Don't move the goalposts.
Any level. I'm not moving goal posts; we were not discussing the validity of the economic shutdown. We were discussing the valifty of the models.
My point was the IMHE model was, and is, of more weight when compared to the IC for several reasons but the biggest being that we were never going to operate as normal. There was going to be some level of restriction and thus the IC is of less value.
quote:
The United States.
Well that is a bad standard because there is no federal mandate. Restrictions vary state to state.
quote:
Irrelevant. The national guidelines are pushing the state level shutdowns. And, the shutdowns are in line with what the Federal government is recommending. Moreover, the Federal government hasn't modified or clarified its guidance to suggest that states were going overboard...even as damage is done to our economy.
There are different standards for different states. I don't know how else to say this.
quote:
But, even if the missteps (presumably) were understandable, they are still missteps. And, we've already taken a huge hit economically due to the guidance provided at the national level. That guidance was based on something. The sooner we figure out that what we based it on was off the mark, the faster we can adjust course.
Every day that goes by increases the damage.
If we've overreacted, then the guidance was either based on a bad model, or a model with very bad data. The only thing that would be worse is if it comes out that they didn't even consider what the guidance was based on. Instead, they just offered guidance because distancing was better than not distancing.
I guess we will see
Posted on 4/13/20 at 5:39 pm to Korkstand
quote:
Polls are known to be "off". On average by a margin of about 2%, if I'm not mistaken
So, you claim I don't understand statistics. Do you not understand how margin of error is calculated? It doesn't seem like you do.
And, to this point, with representative sampling, wouldn't you expect to see polling sway back and forth in each direction due to the inherent margin of error? Is that what happened?
quote:
Of course they would have been different. But on what grounds would you justify randomly moving those 3 states to 50/50?
This is irrelevant. Let's pretend that the bias that existed was totally impossible to predict. Let's pretend it was based on a group of voters who would never allow themselves to be counted in any poll. Let's remove all fault from Silver.
The data IS STILL FLAWED. And, Trump's chances of willing are still going to be understated. But, it doesn't make him right.
I can tell you my opinion on what I think the reasons are that the polls were wrong. I can tell my opinion on why people like yourself didn't see it coming. None of that really matters though. What is evident is that Silver didn't get it right. He got it wrong. He underestimated Trump's chances. If he his conclusion was Hillary 55/45 and that the 3 states in question were tossups, his model IMO would have been VERY accurate (even though his ultimate prediction ended up being wrong).
Posted on 4/13/20 at 5:45 pm to moneyg
quote:That's exactly what I'm saying. Do you not realize they do this?
What are you saying? That the 538 model accounts for non-representative sampling?
quote:It sure as hell doesn't mean they were inaccurate, or "wrong".
It sure as hell doesn't mean the odd were accurate, correct?
quote:Why does it matter what I think, especially after the fact when we know what happened? Hindsight is 20/20, and it obviously adds bias to one's opinion. This is extremely evident in yours.
So, the question is whether or not you think, at the time of the election, Michigan, Pennsylvania, and Wisconsin were closer to 50/50 contests or 80/20 contests in favor of Hillary? What is your opinion?
Posted on 4/13/20 at 5:45 pm to Korkstand
quote:
Well, what are you suggesting 538 should have done? Just drastically alter poll results because "reasons"?
This is why you have this crazy illogical stance.
I'm not necessarily suggesting 538 should have done anything. The only thing I'm suggesting is that his model, because of the data within it, wasn't right. It was signficantly off the mark.
Your illogical self thinks that because there was nothing 538 could have done differently, that it must have been accurate.
quote:
Instead, you tried to call out my "limited" logic skills by going on to lay out an even more ridiculous example.
I know how polling works. I know how the EC works. I know how polling and margins of error are applied to project state by state outcomes, and how that relates to EC votes.
But, more importantly, I also understand probability in general.
I'm not convinced of that at all. Honestly, your other post indicated that you don't understand how margins of error are calculated, or what the requirements were to get to a valid probability.
quote:
If we for some random reason had moved the 3 states in question to 50/50, can you calculate the odds of Trump winning all 3 again?
All 3 would 12.5%. Now, what's your point?
Posted on 4/13/20 at 5:59 pm to Antonio Moss
quote:
Any level. I'm not moving goal posts; we were not discussing the validity of the economic shutdown. We were discussing the valifty of the models.
It's implied. The social separation that we chose had huge economic ramifications. I don't care what your opinion is. I may disagree. But, your opinion should be based on some understanding of the cost and benefit of each possible path we could have undertaken.
You've stated a few times, that we didn't need to know the cost of not implementing drastic social distancing policies to decide that implementing them was the right thing to do.
quote:
The conversation in this thread was about the validity of the models and the false narratives surrounding them.
Correct. How can we gauge the validity of the models if your position is that we might not have even considered a model that predicted deaths if we didn't do anything? To the degree that there are false narratives, we can't evaluate that without understanding the models and values that were used to make the decision.
quote:
I'm not answering that.
You don't have to be specific. Can you share the industry?
quote:
You don't think the WH and governors had numbers in from of them?
#1, when I suggested that it was obvious that the decision makers had used some model to estimate the number of deaths with no action, you questioned that.
more importantly
#2, I'm absolutely NOT convinced that I agree with the decision we made by simply knowing that they had numbers in front of them. I'd like to know what the numbers were. Then I can opine on whether that was a good decision. Then, I'd like to compare those numbers to what we've seen since those predictions. Then I can opine on whether our leaders were misled. And, if I come to the conclusion that they were misled, I'll probably opinine on whether or not that was intentional.
quote:
For example, if we ended up cutting our overall deaths from 200K to 50K, this was a collosal mistake.
That is your opinion.
Yes, that is my opinion. Do you disagree?
quote:
Any level. I'm not moving goal posts; we were not discussing the validity of the economic shutdown. We were discussing the valifty of the models.
My point was the IMHE model was, and is, of more weight when compared to the IC for several reasons but the biggest being that we were never going to operate as normal. There was going to be some level of restriction and thus the IC is of less value.
The validity of the models is important because it justifies our action. We aren't discussing the validity of the models in a vacuum. Clearly, there is a reason the validity is important. This line of argument from you really makes me want to know what your background is. I hope you reconsider and let us know.
quote:
Well that is a bad standard because there is no federal mandate. Restrictions vary state to state.
Distinction without a difference. I love Trump. But, what has happened has happened with his blessing. He may have been misled. But, the national guidance supports the state shutdowns and there has been no pushback.
quote:
There are different standards for different states. I don't know how else to say this.
I'm not sure what point you are making.
Posted on 4/13/20 at 6:15 pm to moneyg
quote:I do, but maybe I wasn't clear in what I was talking about. Margin of error is a statistical calculation, which expresses the confidence that the result will be within the given range of the poll. Margin of error is calculated before a result. The error I was referring to was the error in historical national polls vs. election results, calculated after the fact. This type of data is considered when calculating the margin of error in polls, among numerous other factors.
Do you not understand how margin of error is calculated? It doesn't seem like you do.
quote:That is what we might expect with representative sampling, yes. And no, that did not happen. What happened in the case of 538's data, they considered 80 polls for Wisconsin, for example. Of those 80, Trump was favored in... 3. After 538 adjusted for various biases, they had Trump favored in 9 of them. Actually, just from glancing at the data, they adjusted in favor of Trump in the vast majority of the polls. They clearly were aware of the biases and adjusted the data based on historical data.
And, to this point, with representative sampling, wouldn't you expect to see polling sway back and forth in each direction due to the inherent margin of error? Is that what happened?
quote:
This is irrelevant.
So, you definitely think they should have adjusted the data in order to be more "right", but the grounds for making such adjustments is irrelevant?
quote:Ok.
Let's pretend that the bias that existed was totally impossible to predict. Let's pretend it was based on a group of voters who would never allow themselves to be counted in any poll. Let's remove all fault from Silver.
quote:But does it make him "wrong"?
The data IS STILL FLAWED. And, Trump's chances of willing are still going to be understated. But, it doesn't make him right.
quote:Now that is irrelevant.
If he his conclusion was Hillary 55/45 and that the 3 states in question were tossups, his model IMO would have been VERY accurate
Posted on 4/13/20 at 6:21 pm to Korkstand
quote:
That's exactly what I'm saying. Do you not realize they do this?
You are getting caught up in weeds.
I think you are discussing adjustments to random sampling to get a population that is closer to a representative sample. That's what most polls do.
538 is melding these polls together. To the degree that they are adjusting the raw data to fit a "representative" population, the results are only as accurate as how close they hit the mark. I don't know the specifics of how they are doing that. I'd call this their special sauce.
In the end, they were very far off in the 3 states we've been discussing.
quote:
It sure as hell doesn't mean they were inaccurate, or "wrong".
Agree. Underdogs do win. And, sometimes, the oddsmakers just get it wrong. And, when 3 midwestern states all err in one direction, and polling didn't identify that trend at any point in the election cycle, it's the oddsmakers who were wrong.
quote:
Why does it matter what I think, especially after the fact when we know what happened? Hindsight is 20/20, and it obviously adds bias to one's opinion. This is extremely evident in yours.
Because reality tells the real story. As an analog, Deontay Wilder was a favorite to beat Tyson Fury in their rematch. In hindsight, that was inaccurate. The oddsmakers and betting public had reason to think it was 50/50. Wilder won the first fight. He had one punch knockout power that showed in the first fight.
But, in hindsight, we know differently. Nobody on earth now thinks if they had that fight 10 times, it would be split 5-5.
This post was edited on 4/13/20 at 6:28 pm
Posted on 4/13/20 at 6:25 pm to Korkstand
quote:
That is what we might expect with representative sampling, yes. And no, that did not happen. What happened in the case of 538's data, they considered 80 polls for Wisconsin, for example. Of those 80, Trump was favored in... 3. After 538 adjusted for various biases, they had Trump favored in 9 of them. Actually, just from glancing at the data, they adjusted in favor of Trump in the vast majority of the polls. They clearly were aware of the biases and adjusted the data based on historical data.
They didn't adjust enough. They underestimated the bias...and because of that they were wrong.
quote:
So, you definitely think they should have adjusted the data in order to be more "right", but the grounds for making such adjustments is irrelevant?
No, I'm saying that basing your model off of inaccurate data, even if it is the only data available, doesn't mean your model is predictive.
quote:
But does it make him "wrong"?
It makes his proability value off the mark. Some will call that wrong. It is significantly off the mark that "wrong" isn't a crazy term.
quote:
Now that is irrelevant.
It's a hypothetical. So, I agree it's irrelevant. But, that hypothetical was very close to reality as we know with real data and hindsight.
Posted on 4/13/20 at 6:26 pm to moneyg
wtf is even this thread right now
Posted on 4/13/20 at 6:37 pm to moneyg
quote:
This is why you have this crazy illogical stance.
quote:And you base this opinion on absolutely nothing other than your feelings.
I'm not necessarily suggesting 538 should have done anything. The only thing I'm suggesting is that his model, because of the data within it, wasn't right. It was signficantly off the mark.
quote:While I would argue that it was as accurate as a prediction could be, I am really arguing against your repeated claims that the projection was "wrong". That is just an absurd statement.
Your illogical self thinks that because there was nothing 538 could have done differently, that it must have been accurate.
quote:Very good. And the point is that, even with "good" data, Trump winning all 3 of those states would have been a relative longshot. Just because something is improbable doesn't mean the probability was "wrong" when it happens. And 538 gave Trump TWICE as high a probability to win the entire election, and you just think that was "wrong" for some unknown reason.
All 3 would 12.5%. Now, what's your point?
Posted on 4/13/20 at 7:06 pm to moneyg
quote:I'm caught up in something, that's for sure.
You are getting caught up in weeds.
quote:A lot of the specifics in their methods are explained on their site. It's broken down state by state, poll by poll, and includes a lot of their adjustments such as a given poll's historical accuracy, how much they have adjusted it, how heavily they weight it, etc. Perhaps you should look into all that before proclaiming that someone was "wrong"?
I think you are discussing adjustments to random sampling to get a population that is closer to a representative sample. That's what most polls do.
538 is melding these polls together. To the degree that they are adjusting the raw data to fit a "representative" population, the results are only as accurate as how close they hit the mark. I don't know the specifics of how they are doing that. I'd call this their special sauce.
quote:Wisconsin
In the end, they were very far off in the 3 states we've been discussing.
538 projection: Hillary by 5.3
Actual: Trump by 0.7
Error: 6.0 to Hillary
Michigan
538 projection: Hillary by 4.2
Actual: Trump by 0.3
Error: 4.5 to Hillary
Pennsylvania
538 projection: Hillary by 3.7
Actual: Trump by 0.7
Error: 4.4 to Hillary
These are not unexpected errors! This is not "very far off". They happen all the time in every election. If 538 had gotten these results much closer, given the available data, Silver would be praised as a fricking prophet. Instead, you're here just calling him "wrong".
quote:Except 538 very prominently identified the trend. It was actually a relatively big deal in the media that 538 was giving Trump much higher odds than everyone else.
when 3 midwestern states all err in one direction, and polling didn't identify that trend at any point in the election cycle, it's the oddsmakers who were wrong.
Posted on 4/13/20 at 7:27 pm to moneyg
quote:It's pretty clear at this point that you think the probability value should have been closer to the voting percentages. Yet another example that shows that you do not understand probability as well as you think you do. Because, take Pennsylvania for example, where 538 predicted Trump would get 45.2% of the vote. He ended up getting 48.2%. Do you not realize how insanely accurate that is? You are seriously just going to call that flat out "wrong"?
It makes his proability value off the mark. Some will call that wrong. It is significantly off the mark that "wrong" isn't a crazy term.
Posted on 4/13/20 at 7:29 pm to HailHailtoMichigan!
Nola’s spread of covid was artificial
We don’t live extremely dense like other hotspots. Mardi Gras is a spot in time issue, not an ongoing one
We 100% had this mapped out in our model at Ochsner. If anyone is truly surprised with this I have no faith they aren’t a retard
We don’t live extremely dense like other hotspots. Mardi Gras is a spot in time issue, not an ongoing one
We 100% had this mapped out in our model at Ochsner. If anyone is truly surprised with this I have no faith they aren’t a retard
Posted on 4/13/20 at 7:33 pm to VolsOut4Harambe
Despite the amount of upvotes this will still be unappreciated years from now.
Well done.
Well done.
Posted on 4/13/20 at 7:36 pm to notiger1997
quote:I guess so, NOLA is full of liberal sheep pussies. I’m proud of our D rating
but we got an A in the states score card for doing the right thing
Posted on 4/13/20 at 7:38 pm to Antonio Moss
quote:
Birx - "The worst case would be to do nothing which we believe could be 1.6-2.2 million deaths in a worst-case scenario. We have never tried to do social restrictions in this country but we believe that with these restrictions we could bring that number down to 100K-200K . . ."
It sounds to me that she’s saying that the worst case scenario is 1.6 to 2.2 million dead if we do nothing, but if we put the social restrictions in place we could achieve as few as 100,000 deaths. In other words, a best case scenario.
It could be read the way you are suggesting, but that is an unlikely reading. Plus it does not comport with what they were consistently saying.
Page 14 of 15
Popular
Back to top
Follow TigerDroppings for LSU Football News