talksport Posted January 20, 2008 Share Posted January 20, 2008 Could somebody move this thread to an ISI forum ( The kolkata one) Link to comment
observer1 Posted January 20, 2008 Share Posted January 20, 2008 Oy maan. In addition to the heart pumping cefuroxime etc, I know a bit of stats. I know that a chi-square will be less effective (i.e have a higher p value) if I use 4 and 5 instead of 80 and 100. I was just bored, and trying to do what doctors like you usually do - con the world with bs Link to comment
The Outsider Posted January 20, 2008 Author Share Posted January 20, 2008 The individual row or column values. So in this case Fisher might make sense since the numbers are quite small yet. Link to comment
observer1 Posted January 20, 2008 Share Posted January 20, 2008 A damp squib, now that we are getting serious... Analyze a 2x2 contingency table bucknor in town bucknor not in town Total wins 5 44 49 losses 6 32 38 Total 11 76 87 Fisher's exact test The two-tailed P value equals 0.5230 The association between rows (groups) and columns (outcomes) is considered to be not statistically significant. Link to comment
Dhondy Posted January 20, 2008 Share Posted January 20, 2008 Oy maan. In addition to the heart pumping cefuroxime etc, I know a bit of stats. I know that a chi-square will be less effective (i.e have a higher p value) if I use 4 and 5 instead of 80 and 100. I was just bored, and trying to do what doctors like you usually do - con the world with bs Cho chorry!:ahhhhh: Link to comment
Dhondy Posted January 20, 2008 Share Posted January 20, 2008 So in this case Fisher might make sense since the numbers are quite small yet. Yeah. Ideally, at least one number should be Link to comment
observer1 Posted January 20, 2008 Share Posted January 20, 2008 Could somebody move this thread to an ISI forum ( The kolkata one) craving your indulgence maam, but what is the forum you refer to, and whereabouts can it be found? Link to comment
The Outsider Posted January 20, 2008 Author Share Posted January 20, 2008 is considered to be not statistically significant. As suspected, but thanks for running the numbers.:two_thumbs_up: Link to comment
The Outsider Posted January 20, 2008 Author Share Posted January 20, 2008 Yeah. Ideally, at least one number should be Okay, now you've got me curious. Do the results from the two tests converge at 5? I would expect that to be the case for this demarcation to be chosen. Link to comment
Dhondy Posted January 20, 2008 Share Posted January 20, 2008 Okay' date=' now you've got me curious. Do the results from the two tests converge at 5? I would expect that to be the case for this demarcation to be chosen.[/quote'] Yeah, essentially for very small numbers, you'd go for Fishers, for larger numbers, you'd trust Chi^2. If the lowest value is 5, there would be very little difference in the p values of the two Tests. Link to comment
Dhondy Posted January 20, 2008 Share Posted January 20, 2008 Seeing the stats trend on this thread, there's a nice test I could share with you, Shwetabh, if you are interested. Can be done on a calculator. Link to comment
The Outsider Posted January 20, 2008 Author Share Posted January 20, 2008 Seeing the stats trend on this thread' date=' there's a nice test I could share with you, Shwetabh, if you are interested. Can be done on a calculator.[/quote'] Go ahead. Pretty rusty on it but no harm in giving it a shot. The maximum stats people understand in my new vocation is the mean. Link to comment
talksport Posted January 20, 2008 Share Posted January 20, 2008 craving your indulgence maam' date=' but what is the forum you refer to, and whereabouts can it be found?[/quote'] I was referring to Indian Statistical Institute, Kolkata ....as famous as IITs, IIMs , ISc in India P.S. My Avatar is not me, Its "Mithali Raj" Captain of Indian Cricket Team and a Stylish BatsWoman Link to comment
observer1 Posted January 20, 2008 Share Posted January 20, 2008 considering dhondy's self-reported levels of cerebral differentiation, can I safely presume the secret test on the calculator is random digital (excluding thumb) pressure to see how soon you can get pi (akin to the 'sooner or later, anyone typing on a pc at random will come with a shakespeare work') Link to comment
Dhondy Posted January 20, 2008 Share Posted January 20, 2008 On another thread, Observer pointed out that for comparisons to be meaningful, you need a denominator. The beauty of this test is that you don't need a denominator, as long as they are roughly equal. Say you are comparing two events (say, no of decisive Tests at Perth and MCG) ----a & b. You calculate 'z' as: z= a-b/sq root (a+b) If z is more than 2, the difference is significant (20 Denominators should be roughly equal. Beats many sophisticated tests for its utter simplicity. Link to comment
The Outsider Posted January 20, 2008 Author Share Posted January 20, 2008 A very intuitive test, Dhondy. Something which the casual glance can predict, but when asked to back up with numbers would need at least a calculator. This one can be done in the head. Not sure what you mean by, Event rates should be fairly low. Is it that the number of decisive test matches has to be a low percentage of the total tests as in the above example. That's what I can think of intuitively. Link to comment
Dhondy Posted January 20, 2008 Share Posted January 20, 2008 A very intuitive test, Dhondy. Something which the casual glance can predict, but when asked to back up with numbers would need at least a calculator. This one can be done in the head. Not sure what you mean by, Is it that the number of decisive test matches has to be a low percentage of the total tests as in the above example. That's what I can think of intuitively. Yeah, exactly. Link to comment
Recommended Posts
Create an account or sign in to comment
You need to be a member in order to leave a comment
Create an account
Sign up for a new account in our community. It's easy!
Register a new accountSign in
Already have an account? Sign in here.
Sign In Now