Open Side Menu Go to the Top
Register
Rachel Maddow Shows Stupidity Rachel Maddow Shows Stupidity

07-30-2015 , 11:00 AM
Yeah, I mean, if we're playing heads up education for rollz, I'll still take Maddow (Stanford BA, Rhodes Scholar at Oxford) over Trump (started at Fordham, finished his Bachelors at Wharton).
07-30-2015 , 11:48 AM
Quote:
Originally Posted by pvn
I wouldn't call that wishcasting as much as the "well Mr. X went to the same school I went to so obv he's probably pretty smart QED" part.
Giving DS benefit of the doubt, I think his argument was that Trump had to take a course in it (which DS knows by the fact that Trump went to Wharton), whereas Maddow did not.

Still, I have a hard time imagining that Trump understands this, or much of anything beyond the level of a 12-year-old. Being required to take a course in something 40 years ago != understanding what was taught in that course.
07-30-2015 , 11:59 AM
Quote:
Originally Posted by David Sklansky
By the way, I think her error may actually be even greater than stated. I think she may have averaged five polls and used the MOE for one. If so that would be one more mark against her intuition.
dude..

Quote:
Originally Posted by goofball
Watching the segment now, the other thing it looks like she's missing, is that averaging polls impacts the standard error. Basically, averaging polls increases your N to be the total N of all the polls. So saying "all these polls have a standard error of 3" doesn't mean "when I average these polls the average will have a standard error of 3." It means "When i average these polls they have a standard error much less than 3."

edit: I think the pollster doesn't really correct her because her point is still well taken: using polling to decide places 8/9/10/11/12 is going to be largely arbitrary, and will likely have a huge impact on the fortune of the candidates in spots 11 and 12. And yes, she should have done her due diligence on standard error, but shows are produced daily and move fast etc. The idea that this should be taken as any evidence of her being stupid is laughable.
07-30-2015 , 12:23 PM
Clearly, the guy who hates Mexicans is better at math.
07-30-2015 , 02:49 PM
Quote:
Originally Posted by pvn
I wouldn't call that wishcasting as much as the "well Mr. X went to the same school I went to so obv he's probably pretty smart QED" part.
Can I get a probability that two Wharton grads have a dildo in the glove box?
07-30-2015 , 04:32 PM
Quote:
Originally Posted by Trolly McTrollson
Clearly, the guy who hates Mexicans is better at math.
Way better. Oh wait you mean Trump.
07-30-2015 , 05:56 PM
Quote:
Originally Posted by David Sklansky
It would be close to impossible that #s 14, 15, or 16, were actually in the top ten and she was implying otherwise. And made a big deal of it without checking.
There's a zillion polls out there right now. Don't some of them have #10 at 2.2% or whatever and #'s 11, 12, 13 and 14 at 1.6% etc? You're talking about extremely small differences.
07-30-2015 , 05:59 PM
Quote:
Originally Posted by vhawk01
How big do you think they should/need to be?
Considerably bigger than 800-1000. This is infinitesimal and unreliable. I mean, the Presidential election will have somewhere between 135m and 145m vote in it.
07-30-2015 , 08:09 PM
Quote:
Originally Posted by 2OutsNoProb
Considerably bigger than 800-1000. This is infinitesimal and unreliable. I mean, the Presidential election will have somewhere between 135m and 145m vote in it.
Not really. If well collected 1000 should actually be fine and the difference between 1k and 5k (or 10k or 50k) would be negligible. This is assuming no bias in the way they poll, but that bias doesn't go away just by increasing the sample size (at least until you run out of old white retired stay at homes to poll)
07-30-2015 , 08:11 PM
I don't know how this is manifesting itself in the polls right now, but isn't polling pretty messed up right now and undergoing a transition since nobody under 50 really uses a landline anymore?
07-30-2015 , 11:19 PM
Quote:
Originally Posted by David Sklansky
Way better. Oh wait you mean Trump.
07-31-2015 , 03:16 AM
What is the chance that we can manage to get her to talk live on her show about us here lol?

I say 60% possible within 2 weeks.

That is the magnitude of the woman then to be able to make a mini segment to correct herself and show the ultimate signature of confidence and maturity.

Nothing more progressive that to always use your mistakes to become better and never be afraid of recognizing them. It is what is missing from politics. It is seen as weakness when it is a big strength instead.

She can easily spend only 1-2 min on this to offer essential education to thousands watching that they can use for future reference in the rest of their lives. She can redeem herself and essentially say that the mistakes made were;

(although partially her guest kind of corrected here along these lines it was never made very clear)

1) Margin of error is estimated always as if the candidate had the maximum possible error that happens at p=50%. So it kind of works on only 2 person close races that they are like 40-60% each.

2) Multiple similar size polls added take us down from the error of each one say 3-4% each (at 95% confidence of the standard 600-1000 people events) to the square root of the number of polls (ie root 5 so like a bit over 2 times less than the 4% down to less than 2% basically. So her 3-4% radius is more like <2% now after adding 5 polls and only for a 50% candidate case)

3) To estimate the error (95% confidence interval) for people that have significantly less than 50% you simply multiply the maximum error ie say 2% here on the 5 polls added with 2 times the root of the result they got. So if one got 2.6% the error is 2% times 2*0.026^(1/2)~0.6% rather than the 2 or 4% she thought.


Then proceed to verify her point is still valid as they remain close to each other even with these errors. The last 3 probably are not able to challenge but the 10 to 13 position people are iffy.

She may also add the general issues polls have with their methodology demographics and different questions and different times taken. (this is why i rounded to 4% the original ones to add an extra error there rather than starting at 3 or 3.5 that looked liek the avg of each)



Have no doubt that this is a service she would be providing to many people that have some education to follow what she is talking about but never bothered to mathematically study the topic (that may include in fact almost >80% of general bachelor degrees people if you want to fair about it - ie they dont know how to adjust properly, only suspect maybe that they should) and who if intelligent enough would have already noticed that there ought to be a problem to have one error for all different percentages and that something else is going on instead. That way they can follow all future polls and elections much better and what the errors say about how close the races are.

All it takes is probably some email to her group and she would do it and even mention 2+2 for fun and maybe a plus for poker issues as a side benefit.

Last edited by masque de Z; 07-31-2015 at 03:22 AM.
07-31-2015 , 07:25 AM
Quote:
Originally Posted by 2OutsNoProb
Considerably bigger than 800-1000. This is infinitesimal and unreliable. I mean, the Presidential election will have somewhere between 135m and 145m vote in it.
A sample size of over a thousand actually is large and reliable as long as it follows good procedures. Saying it is infinitesimal and unreliable is way off.
07-31-2015 , 10:54 AM
Quote:
Originally Posted by masque de Z
What is the chance that we can manage to get her to talk live on her show about us here lol?

I say 60% possible within 2 weeks.
0%

Quote:
That is the magnitude of the woman then to be able to make a mini segment to correct herself and show the ultimate signature of confidence and maturity.
There are two important things to consider here:

1. She was already corrected on her own show by a polling director
2. Her point still stands


It looks like 538 might actually do something on this. Not the Maddow mistake, but the debate criteria which is the real issue. I feel like if there ever was a time to page Nate this would be it.
07-31-2015 , 10:58 AM
Also

Quote:
Originally Posted by 538
Fox has refused to say which polls it will include in its average, as well as how they will round the numbers to draw a clear line between 10th and 11th place.
So yeah, part of the reason Fox hasn't called Maddow out on this is because they still don't know wtf they're doing.
07-31-2015 , 01:39 PM
Quote:
Originally Posted by Onlydo2days
I don't know how this is manifesting itself in the polls right now, but isn't polling pretty messed up right now and undergoing a transition since nobody under 50 really uses a landline anymore?
Was going to bring this up; good point. Many people in their 20s and 30s today buying their own first home or getting an apartment don't even get a landline. We didn't. If they're relying heavily on these, they're getting only the older crowd.
07-31-2015 , 01:40 PM
But if her intention was to educate her audience that hasnt happened yet because people still didnt learn how to estimate errors. Nobody told them that many polls also reduce the error or how.

I noted that she was kind of corrected but if you see the video the correction is not descriptive enough to prove something people can use in the future for their own thoughts.

I think if it was suggested to her to do a small 2 min insert segment on polls, and actually present it as if she took it on herself to understand it better and so she revisited it, it would prove a quality addition to her program because its election season next 15 months.
07-31-2015 , 02:26 PM
Quote:
Originally Posted by 2OutsNoProb
Was going to bring this up; good point. Many people in their 20s and 30s today buying their own first home or getting an apartment don't even get a landline. We didn't. If they're relying heavily on these, they're getting only the older crowd.
I think they're generally getting better about this. The recent CNN poll (that had Trump only +3 btw) has this:
Quote:
This sample includes 615 interviews among landline respondents and 402 interviews among cell phone respondents.
07-31-2015 , 03:58 PM
Quote:
Originally Posted by masque de Z
But if her intention was to educate her audience
I remember this horrible English Literature teacher whose intention was clearly NOT to educate us because she didn't even mention math.

(Maddow's intention is to entertain her audience by saying things that would really annoy conservatives if they were watching)
07-31-2015 , 08:50 PM
Quote:
Originally Posted by TheDuker
I think they're generally getting better about this. The recent CNN poll (that had Trump only +3 btw) has this:
conventional wisdom has it that trump is popular among the youngs, not the olds

Why is conventional wisdom never right?!

Last edited by Low Key; 07-31-2015 at 08:50 PM. Reason: assuming more youngs than olds have cell phones
08-01-2015 , 10:08 AM
Why blame her? The pollsters themselves were the ones who published that the margin of error was 3%-4% without presenting any exceptions.

Some of the candidates deserve to have -1% votes anyway
08-01-2015 , 02:42 PM
Quote:
Originally Posted by Pokerlogist
Why blame her? The pollsters themselves were the ones who published that the margin of error was 3%-4% without presenting any exceptions.
Because if she is not smart enough to at least feel like those words didn't make sense to the point where she should investigate further than she is not that smart. Many millions who have never studied statistics WOULD have sensed something was wrong.
08-01-2015 , 02:49 PM
And yet, by your own estimate, the overwhelming majority of people on the planet wouldn't have realized anything was in error

She should have known there was a problem, but it's not a character flaw, or a statement against her qualifications as a talking head, that she didn't.

Last edited by King_of_NYC; 08-01-2015 at 02:50 PM. Reason: post #5
08-01-2015 , 04:13 PM
Quote:
Originally Posted by King_of_NYC
And yet, by your own estimate, the overwhelming majority of people on the planet wouldn't have realized anything was in error

She should have known there was a problem, but it's not a character flaw, or a statement against her qualifications as a talking head, that she didn't.
I am holding her to a high standard similar to when I criticized Marilyn vos Savant when she made errors. But it does detract from her qualifications to at least some degree.

Except maybe it WAS a character flaw and not dumbness. Just maybe her brain sensed the problem and her emotions chose to ignore it, perhaps subconsciously. Because it allowed her to accuse FOX of something more egregious than their actual crime and her zeal to do that got the best of her.
08-01-2015 , 04:18 PM
Quote:
Originally Posted by David Sklansky
I am holding her to a high standard similar to when I criticized Marilyn vos Savant when she made errors. But it does detract from her qualifications to at least some degree.

Except maybe it WAS a character flaw and not dumbness. Just maybe her brain sensed the problem and her emotions chose to ignore it, perhaps subconsciously. Because it allowed her to accuse FOX of something more egregious than their actual crime and her zeal to do that got the best of her.
Lol DS

Sent from my SM-G920V using 2+2 Forums

      
m