- #3,256
- 19,303
- 24,987
<WWGD said:Congrats, @russ_watters on the Eagles' victory last night.
<WWGD said:Congrats, @russ_watters on the Eagles' victory last night.
Wow, I thought (US) Football was not very popular outside US ( I am assuming you are not that much of an outlier in that regard)fresh_42 said:<> I first had been happy that Gronkowski was in the line-up. Then I saw how the Eagles' defense dealt with it and that the Patriots' pass defense was basically not existent. All in the first quarter, and the half-time result wasn't very promising either. But I had a little hope for the 4th. However, I fell asleep during Timberlake only to awake afterwards seeing my hopes had been all in vain. </
>
They had even live reporters in the stadium, and not the usual idiots, but actually experts.WWGD said:Wow, I thought (US) Football was not very popular outside US ( I am assuming you are not that much of an outlier in that regard)
"The rolling English country road... laid by the rolling English country drunkard..."fresh_42 said:Just found a Random Road ... What must have happened to call it this way?
WWGD said:Another example of how/that Math is useful in daily life. Probabilities. Someone was going through Yelp in order to find a professional for hire. I went over the search with her, I noticed the highest ranking one, which she was considering, had some 150 ratings, and all of them 5 stars out of 5. I argued, using the Binomial with even p=0.95 , the probability of 150 ratings _all_ being 5 out of 5 was less than one in a million http://stattrek.com/online-calculator/binomial.aspx
Yes, too good to be true . First, must have p=0.95 , then must satisfy people at a 100% rate. Even when someone is extremely good there are those who will nitpick for one reason or another. Rare that none of some 150 people had something to complain about. EDIT: I tried to consider the caveat that those who are satisfied are more likely to comment than those that are not, but this does not seem to bear out in general in Yelp or other review sites.StoneTemplePython said:What was the verdict then? Too good to be true / rigged? (I didn't see a prior here...)
I missed the molten cheese ...Borg said:I'm trying to decide if that is deadpan humor or deathpan humor.![]()
I miss it too. And I heard it misses us. And it misses the mrs's.fresh_42 said:I missed the molten cheese ...
Here's a famous book titled: Why men can't listen and women can't park. There are a lot of things to be said for why there are differences. The problems start, if those differences are evaluated which is always biased. Anyway, I know if I say "A" and a woman says "B", I'll put my money on "B", whatever it is.WWGD said:Interesting article and dataset on proving that !Q in males has larger variability than in females. A ratio of variances passes the F-test ( Ratio of Variances) even at the 0.01 significance level, though result is not politically-correct nowadays because some use it to explain why there are more male CEOs -- tho it may also explain why some 93%+ of prison population is male -- the left tail is fatter in males , just like the right tail. . The site is not politically neutral, but it sticks to the data, as I saw it. .http://www.aei.org/publication/statistical-tests-shows-greater-male-variance/ Hope it is acceptable to post this; let me know otherwise.
Watch your nature shows.fresh_42 said:Why men can't listen and women can't park.
WWGD said:Interesting article and dataset on proving that !Q in males has larger variability than in females. A ratio of variances passes the F-test ( Ratio of Variances) even at the 0.01 significance level, though result is not politically-correct nowadays because some use it to explain why there are more male CEOs -- tho it may also explain why some 93%+ of prison population is male -- the left tail is fatter in males , just like the right tail. . The site is not politically neutral, but it sticks to the data, as I saw it. .http://www.aei.org/publication/statistical-tests-shows-greater-male-variance/ Hope it is acceptable to post this; let me know otherwise.
AEI said:meaning that there is only a 1-in-a-1000 chance that we would find these results purely by chance, and a 99.9% chance that we have established a statistical difference in variances.
True there are subtleties, and very few , specially in today's charged climate, to abandon their preconceptions.StoneTemplePython said:I don't care much for AEI -- way too partisan and not very insightful. For example:
Interpreting classical statistics correctly is perilous, and I'm pretty sure that this is wrong.
It also depends on what part of the distribution you are interested in -- in particular, consider the extremes.
(a) It's been fairly well documented that severe intellectual disabilities are in the neighborhood of 4x - 6x more likely in males. (I can foot to some stuff from The Economist I think.) That alone is enough to spike variance if the means are comparable and we are in fact evaluating the entire distribution (and remember we are talking about squared deviations so variance weights extreme things more). (b) A more interesting test would look at whether the distributions are actually well approximated as symmetric (and in particular whether this holds at the extremes which may not be so easy).
There is a ton of subtlety involved and I've met just about no one who is able to think through these things dispassionately, deal with subtleties and guard against ideological and self-serving biases. As a result, I have a hunch that this is not appropriate for the forum.
It's worth recalling that the final straw in Larry Summer's presidency at Harvard was speculating on variance in intellect and its potential impact in physics. Edge.org had a very good discussion on superforecasting which at one point remarked that this is viewed merely as hypothesis generation by those small few that qualify as superforecasters and most everyone else went berserk after hearing it.
Come to think of it there is a lot of good stuff in that thing on superforecasting, that is probably a lot more fruitful and interesting to read through:
https://www.edge.org/event/edge-master-class-2015-philip-tetlock-a-short-course-in-superforecasting
(I actually think this 5 part discussion may be better than the book.)
Still, though, if you accept the data, the variability claim holds. Doesn't it? Do you think "left" variability would explain all the difference in variances, even at the 0.01% level? Or would you like to see the original data to examine for "right" variability ( i.e., variability on higher values)StoneTemplePython said:I don't care much for AEI -- way too partisan and not very insightful. For example:
Interpreting classical statistics correctly is perilous, and I'm pretty sure that this is wrong.
It also depends on what part of the distribution you are interested in -- in particular, consider the extremes.
(a) It's been fairly well documented that severe intellectual disabilities are in the neighborhood of 4x - 6x more likely in males. (I can foot to some stuff from The Economist I think.) That alone is enough to spike variance if the means are comparable and we are in fact evaluating the entire distribution (and remember we are talking about squared deviations so variance weights extreme things more). (b) A more interesting test would look at whether the distributions are actually well approximated as symmetric (and in particular whether this holds at the extremes which may not be so easy).
There is a ton of subtlety involved and I've met just about no one who is able to think through these things dispassionately, deal with subtleties and guard against ideological and self-serving biases. As a result, I have a hunch that this is not appropriate for the forum.
It's worth recalling that the final straw in Larry Summer's presidency at Harvard was speculating on variance in intellect and its potential impact in physics. Edge.org had a very good discussion on superforecasting which at one point remarked that this is viewed merely as hypothesis generation by those small few that qualify as superforecasters and most everyone else went berserk after hearing it.
Come to think of it there is a lot of good stuff in that thing on superforecasting, that is probably a lot more fruitful and interesting to read through:
https://www.edge.org/event/edge-master-class-2015-philip-tetlock-a-short-course-in-superforecasting
(I actually think this 5 part discussion may be better than the book.)
WWGD said:Still, though, if you accept the data, the variability claim holds. Doesn't it? Do you think "left" variability would explain all the difference in variances, even at the 0.01% level? Or would you like to see the original data to examine for "right" variability ( i.e., variability on higher values)
StoneTemplePython said:The issue is in part that normal approximations have curious breakdowns in the real world. Reference financial data (returns are approximately log-normal except extremes) and even human heights (which can be approximated as normal by sex, except in each case there are far too many very tall and very short people). For financial data there is even some reason to believe that variance may in fact be infinite.
In general variance is quite sensitive to extreme events. (You could even phrase this as a ruler problem -- does high variance tell you a lot about extreme events, or does the existence of too many extreme events tell you a lot about the quality of variance estimates and normal approximation?) It's tricky because a ##\approx 10 \%## change in variance doesn't change all that much near the center, but it has a massive impact on the tails of a normal distribution.
It could be the the left tail explains just about all of the variance difference. It may not be the case. There are also some other issues akin to pre-registering or data snooping. (I.e. you actually need people to agree to a methodology before examining the data.)
I don't see much to be gained from this line of inquiry, for the reasons I outlined above, so I will drop now.
And the question is? Playing Jeopardy?fresh_42 said:The answer is 101010.
WWGD said:And the question is? Playing Jeopardy?
What do you get when you multiply 110 by 1001?WWGD said:And the question is?fresh_42 said:The answer is 101010.
00101011 OR 11010100, that is the question.fresh_42 said:The answer is 101010.
Square-free numbers ( and containing neither 2,5 as factors) seem "primerer" (more likely to be prime) than non-square-free ones, is my impression. A nice random result: 9-digit numbers without repeated digits =10!- 9! ; 10! rearrangements of {0,1,...,9} minus all arrangements starting with 0. But I wonder what is the prime density _restricted to odd numbers_ . ? EDIT: Asympotically, of course, otherwise we get a(n) (almost) doublingfresh_42 said:I find that besides the astonishing symmetry the decomposition into primes is funny, too: ##101010 = 2\cdot3\cdot5\cdot7\cdot13\cdot37## - no powers, all primes below ##10## included, and ##37## for the symmetry. "Here I am with a brain the size of a planet and they ask me to..." ... do some numerology. I hate numerology.
Psinter said:To think that the doctor put me to sleep in like 5 or 10 minutes, performed the procedure, and when I woke up I remembered nothing. Literally nothing.