Totally. All those stupid affluent New Yorkers. They are probably spending time on their boats instead of spending all weekend troubleshooting their old Macs! God, if they'd just have put in 8 hours working on it, they could have saved 2 grand and gotten a decent meal at Alain!
My point is, I don't think this behavior makes stupid "stupid" ipso facto. It just means they have more money than you (or me), which might mean they are not so stupid.
Why would Google "owe" any of us anything? Couldn't they just do whatever they want, I don't think I ever signed a contract with them specifying what behavior they are bound to?
Can you find a brand of laptop where the same Google searches don't turn up the same results of customers that feel ripped off because theirs happened to be one of the ones that failed?
All laptop brands have a (hopefully) small percentage that fail. Searching google will turn up people complaining about this. In my opinion, it's more important how the company responds to the issue.
Hi, as a statistician, thank you. I have been meaning to reply to the OP about 'statistics not being able to do prediction' or some nonsense like that all morning, but you beat me to it.
No! Citing Grimmett and Stirzaker [Probability and Random Processes, 7], "An event A is called null if P(A) = 0. If P(A) = 1, we say that A occurs almost surely. Null events should not be confused with the impossible event, the empty set. Null events are happening all around us, even though they have zero probability; after all, what is the chance that a dart strikes any given point of the target at which it is thrown? That is, the impossible event is null, but null events need not be impossible." (emphasis mine).
I have no idea if this was taken into account, and I don't have time to check. But Dell (almost?) always has something like 25-40% off coupons for machines over X dollars. Anyone buying a computer can find these with a simple google search.
Please read what I wrote more carefully. You are calculating a probability of making a Type I error under our assumptions. My claim was the the *expected value* of the number of Type I errors under our assumptions is 1, which is of course correct, since our assumptions lead us to a binomial distribution with n = 20, p =.05, and E[X] = n*p = 1 for a binomial random variate.
I agree, it's not clear from the article whether this was comparison was the prespecified one or not. I couldn't track down the source article to check either (it didn't appear to be in the May 2007 issue of BJP). To give them the benefit of the doubt, it sounds like this was the whole point of their trial, but who knows? We won't until we can find the original publication. It sounded like from the article that the effect size was very small anyway, although significant.
Because the 95% confidence level means you are allowing a 5% chance of finding a correlation where one doesn't actually exist. This is called a Type I error in hypothesis testing. If you do 20 tests, each with a 5% chance of a type I error, the expected number of times that you incorrectly reject the null hypothesis is 1.
Well said. It seems the desire to interact is now so great that he's resorted to "Ask Slashdot", a sure sign of "going too far" if there ever was one! I personally think the best thing she said was "I'm glad to benefit from it".
If you believe Steven Levitt's stats in Freakonomics, I do.
From page 22,
Homicides per 100,000 people (England) 13th/14th c. 23.0 15 NA 16 7 17 5 18 1.5 19 1.7 1900-1949 0.8 1950-1994 0.7
There is a very similar trend in the four other areas mentioned on page 22 of that book. These statistics, according to Levitt, were compiled by the criminologist Manuel Eisner.
Well I of course agree with your sentiment, calling the results "meaningless" isn't quite right either. They have some meaning, even if they are based on a small sample. For instance, if all 8 patients died the day of their first dose of vaccine, would you call that meaningless? If you read the article, all the actual investigators understand this is only a Phase I study, but that's where everything starts off. I agree, so many things that seem promising do not work out when done in a large trial, so excitement should be tempered until then. Your criticisms of this study would be valid if it were a Phase II/III trial, but this one clearly wasn't designed to show efficacy.
Slashdot Mirror
I was about to point out the exact same thing. And I took the LSAT a few years ago, you'll do fine, don't worry.
Totally. All those stupid affluent New Yorkers. They are probably spending time on their boats instead of spending all weekend troubleshooting their old Macs! God, if they'd just have put in 8 hours working on it, they could have saved 2 grand and gotten a decent meal at Alain!
My point is, I don't think this behavior makes stupid "stupid" ipso facto. It just means they have more money than you (or me), which might mean they are not so stupid.
Why would Google "owe" any of us anything? Couldn't they just do whatever they want, I don't think I ever signed a contract with them specifying what behavior they are bound to?
It took me a while to get the hang of rocket jumping, too. Keep at it!
Can you find a brand of laptop where the same Google searches don't turn up the same results of customers that feel ripped off because theirs happened to be one of the ones that failed?
All laptop brands have a (hopefully) small percentage that fail. Searching google will turn up people complaining about this. In my opinion, it's more important how the company responds to the issue.
Pointing out MS FUD is like taking home the drunkest, ugliest girl in the bar. Yah, you did it, but no one is impressed.
I'm not sure people share their porn the way they share their music.
Sounds like you need new friends.
So is it your (my?) moral duty to stop all unecessary predation in 'nature'? If not, then why should we do it in the human animal case?
Hi, as a statistician, thank you. I have been meaning to reply to the OP about 'statistics not being able to do prediction' or some nonsense like that all morning, but you beat me to it.
No! Citing Grimmett and Stirzaker [Probability and Random Processes, 7], "An event A is called null if P(A) = 0. If P(A) = 1, we say that A occurs almost surely. Null events should not be confused with the impossible event, the empty set. Null events are happening all around us, even though they have zero probability; after all, what is the chance that a dart strikes any given point of the target at which it is thrown? That is, the impossible event is null, but null events need not be impossible." (emphasis mine).
t slab/Stuff/topic5.php
Also, see http://instruct1.cit.cornell.edu:8000/courses/sta
seconded
I have no idea if this was taken into account, and I don't have time to check. But Dell (almost?) always has something like 25-40% off coupons for machines over X dollars. Anyone buying a computer can find these with a simple google search.
Please read what I wrote more carefully. You are calculating a probability of making a Type I error under our assumptions. My claim was the the *expected value* of the number of Type I errors under our assumptions is 1, which is of course correct, since our assumptions lead us to a binomial distribution with n = 20, p = .05, and E[X] = n*p = 1 for a binomial random variate.
I agree, it's not clear from the article whether this was comparison was the prespecified one or not. I couldn't track down the source article to check either (it didn't appear to be in the May 2007 issue of BJP). To give them the benefit of the doubt, it sounds like this was the whole point of their trial, but who knows? We won't until we can find the original publication. It sounded like from the article that the effect size was very small anyway, although significant.
Because the 95% confidence level means you are allowing a 5% chance of finding a correlation where one doesn't actually exist. This is called a Type I error in hypothesis testing. If you do 20 tests, each with a 5% chance of a type I error, the expected number of times that you incorrectly reject the null hypothesis is 1.
Let's just say that's not all we're better at.
Well said. It seems the desire to interact is now so great that he's resorted to "Ask Slashdot", a sure sign of "going too far" if there ever was one! I personally think the best thing she said was "I'm glad to benefit from it".
From the bottom of the first tutorial...
Cassandra said...
this is awesome Dan thanks!! you OBVIOUSLY have way too much time on your hands...lol but i'm glad to benefit from it!
In other words, DENIED! Sorry, man, we've all been there.
Doesn't having a MySpace account make you a suspected sex offender, ipso facto?
But are they pilotless drones?
I did not RTFA, but I did read the summary. I did not hear his argument, I heard his conclusion repeated with more words.
If it was really better someone would have already invented this "new" method.
What a bizarre claim! You're implying that there has been no progress ever, and furthermore, there can be no progress ever!
This legislation deserves to be modded (-1, Flamebait).
You have murder stats going back over 800 years?
I'll assume you meant 2001.
If you believe Steven Levitt's stats in Freakonomics, I do.
From page 22,
Homicides per 100,000 people (England)
13th/14th c. 23.0
15 NA
16 7
17 5
18 1.5
19 1.7
1900-1949 0.8
1950-1994 0.7
There is a very similar trend in the four other areas mentioned on page 22 of that book. These statistics, according to Levitt, were compiled by the criminologist Manuel Eisner.
Well I of course agree with your sentiment, calling the results "meaningless" isn't quite right either. They have some meaning, even if they are based on a small sample. For instance, if all 8 patients died the day of their first dose of vaccine, would you call that meaningless? If you read the article, all the actual investigators understand this is only a Phase I study, but that's where everything starts off. I agree, so many things that seem promising do not work out when done in a large trial, so excitement should be tempered until then. Your criticisms of this study would be valid if it were a Phase II/III trial, but this one clearly wasn't designed to show efficacy.