Thank you so much! Your example really helps me understands The example I was reading from book null hypothesis is : honey cough does. It help cough. So I have one groups treated with honey and their outcome. One groups not treated with honey and their outcome 1. Since we assume honey cough does not work. We have 2* data because the treated group will have same outcome when they are untreated vice versa 2. I calculated the observed outcome. Which is just outcome* treated- outcome * untreated 3. Then form a table with the all assume sharp null hypothesis to form a table of outcome for all combinations 4. Find the number of combinations that have absolute outcome greater than observed 5. Pvalue= number in steps4= seeing effect as big as actual observed I guess I just really don’t get why people do fisher exact test because I understand the formula to get it but not why. If P value is small, just like the lady gets the correct combination. It means the effect of honey is not by chance, basically rejecting null hypothesis
Hey, I really liked the video, but as a fellow youtuber I'd recommend using some sort of denoise tool in your video editor. It helps with that kind of bass line that you have. Cheers!
Great video. I want to mention one caveat about the p-value. At 0.05, we can expect around 30% non-reproducibility (weak evidence). A lower p-value would be ideal. Fisherian statistics are frequentist methods. Bayesian methods take into account prior probabilities.
In practical terms, without any math, for almost any judge, the Lady guessing right the four cups with tea-first would be very convincing. While a 3 right, 1 wrong, wouldn´t be enough. I don´t know why Fisher was happy being wrong at most 1 out of 20, pvalue=0.05 or less, but surely he subconsciously and subjectively inferred it from his experiments with chickens, and then objectively observing that being right 19 out of 20 warranted a profit in the long term, while keeping over-optimism at check.
great explanation i have a question the p value here isn't suppose to be under the two tailed t-test distribution since the p is equal to something equallty rare and extremer so 2/70 ???
The video fails to indicate how to calculate the fisher.test. It brings the probability distribution from Wikipedia for the particular Tea Lady situation, but what about any other situation? In the case of the TeaLady the calculation is : (4C0) * (4C4) / (8C4) = (1)*(1) / 70 = 0.01428.... where C means combinations - Contingency Table -------------------------- | a | b | a+b | c | d | c+d -------------------------- a+c | b+d | a+b+c+d - Fisher.Test probability = ( (a+b)Ca * (c+d)Cc ) / ( (a+b+c+d)C(a+c) ) -
Please can i interpret the result of the Fisher’s Exact Test as Fisher-Freeman-Halton Test, for a 3*5 contingency table? I saw throught the internet that the result labeled as Fisher's Exact test in the output is in fact the Fisher-Freeman-Halton Test
Is the p-value correctly calculated or in other videos it is shown different. p-value should be the sum of random chance of generated data or something else that is equal or rarer. so all TTTT as one possible another possibility all MMMM then it should be (1/70) + (1/70) = 0.014 + 0.014 = 0.028. is this not the way p-value is calculated?
Hey , When Fisher first introduced this test , he introduced it to be 1-sided ( one-tailed ) , Hence he didn't consider another equally rare Combination which could have participated in P-value ...
it's a combination: there are 8 cups of tea tested and the lady guesses 4 out of the 8 to be with milk poured in first (she knows that the over 4 must therefore be tea poured in first). look up "combination" in mathematics online if you want the formula :)
that solid bass line tho
+Pes Ces yeah I know 🙁
You can use the free Audacity software to eliminate that background noise. Very easy to do.
🤣🤣
Your explanation was very clear and very easy to follow. And, it helped me understand this test a lot better. Thank you!
Thank you so much! Your example really helps me understands
The example I was reading from book null hypothesis is : honey cough does. It help cough. So I have one groups treated with honey and their outcome. One groups not treated with honey and their outcome
1. Since we assume honey cough does not work. We have 2* data because the treated group will have same outcome when they are untreated vice versa
2. I calculated the observed outcome. Which is just outcome* treated- outcome * untreated
3. Then form a table with the all assume sharp null hypothesis to form a table of outcome for all combinations
4. Find the number of combinations that have absolute outcome greater than observed
5. Pvalue= number in steps4= seeing effect as big as actual observed
I guess I just really don’t get why people do fisher exact test because I understand the formula to get it but not why.
If P value is small, just like the lady gets the correct combination. It means the effect of honey is not by chance, basically rejecting null hypothesis
Hey, I really liked the video, but as a fellow youtuber I'd recommend using some sort of denoise tool in your video editor. It helps with that kind of bass line that you have.
Cheers!
it's actually kind of relaxing haha i'm wondering if it's not doing on purpose
The sound system in my car doesn’t like the bass line😅
Great video. I want to mention one caveat about the p-value. At 0.05, we can expect around 30% non-reproducibility (weak evidence). A lower p-value would be ideal. Fisherian statistics are frequentist methods. Bayesian methods take into account prior probabilities.
In practical terms, without any math, for almost any judge, the Lady guessing right the four cups with tea-first would be very convincing. While a 3 right, 1 wrong, wouldn´t be enough. I don´t know why Fisher was happy being wrong at most 1 out of 20, pvalue=0.05 or less, but surely he subconsciously and subjectively inferred it from his experiments with chickens, and then objectively observing that being right 19 out of 20 warranted a profit in the long term, while keeping over-optimism at check.
great explanation i have a question the p value here isn't suppose to be under the two tailed t-test distribution since the p is equal to something equallty rare and extremer so 2/70 ???
Very nice and clear explanation
Nice Explanation.
Why do you multiply the possible number of ways of correctly choosing the cups by itself? (1x1, 4x4, 6x6, 4x4, 1x1).
The video fails to indicate how to calculate the fisher.test. It brings the probability distribution from Wikipedia for the particular Tea Lady situation, but what about any other situation? In the case of the TeaLady the calculation is :
(4C0) * (4C4) / (8C4) = (1)*(1) / 70 = 0.01428.... where C means combinations
-
Contingency Table
--------------------------
| a | b | a+b
| c | d | c+d
--------------------------
a+c | b+d | a+b+c+d
-
Fisher.Test probability = ( (a+b)Ca * (c+d)Cc ) / ( (a+b+c+d)C(a+c) )
-
Thanks, I was going insane thinking I was missing something
Please can i interpret the result of the Fisher’s Exact Test as Fisher-Freeman-Halton Test, for a 3*5 contingency table? I saw throught the internet that the result labeled as Fisher's Exact test in the output is in fact the Fisher-Freeman-Halton Test
Excellent explanation. Thank you.
crystal clear explanation.
this is awesome! TY!
very helpful, thanks
Is the p-value correctly calculated or in other videos it is shown different. p-value should be the sum of random chance of generated data or something else that is equal or rarer. so all TTTT as one possible another possibility all MMMM then it should be (1/70) + (1/70) = 0.014 + 0.014 = 0.028. is this not the way p-value is calculated?
Hey , When Fisher first introduced this test , he introduced it to be 1-sided ( one-tailed ) , Hence he didn't consider another equally rare Combination which could have participated in P-value ...
what's with that heavy humming bass noise 😰
which pack that 808 from tho
ha ha :(
really cool video
I love you!!!
nice video.
I really like the voice as well, so pretty :)
thkssss !
thank you :)!
Nice bass
audio frequency is hitting the head!
my handwriting is kinda the same^^
please I didn't understand why there was no. 70?
it's a combination: there are 8 cups of tea tested and the lady guesses 4
out of the 8 to be with milk poured in first (she knows that the over 4
must therefore be tea poured in first). look up "combination" in
mathematics online if you want the formula :)
+math et al thanks dear for the notice.
8!/(4!*(8-4)!)=70 , Combination N elements K classes without repeating !, n=8, k=4. N! / (K! * (N - K)!)
This DC offset :(