False discovery rates and P values: the movie. Now on Youtube. http://t.co/Tv6DiXHdzS (simplified version of http://t.co/ytbygLI5dv )
False discovery rates and P values: the movie. Now on Youtube. http://t.co/Tv6DiXHdzS (simplified version of http://t.co/ytbygLI5dv )
False discovery rates and P values: the movie. Now on Youtube. http://t.co/Tv6DiXHdzS (simplified version of http://t.co/ytbygLI5dv )
False discovery rates and P values: the movie. Now on Youtube. http://t.co/Tv6DiXHdzS (simplified version of http://t.co/ytbygLI5dv )
False discovery rates and P values: the movie. Now on Youtube. http://t.co/Tv6DiXHdzS (simplified version of http://t.co/ytbygLI5dv )
False discovery rates and P values: the movie. Now on Youtube. http://t.co/Tv6DiXHdzS (simplified version of http://t.co/ytbygLI5dv )
.@DavidPriceUCL Absolutely But if you want to know why, better read http://t.co/ytbygLI5dv http://t.co/PGv9yfqqsB
Re last RT, see e.g. these from @david_colquhoun http://t.co/OBXkCYFPPj and http://t.co/WuW3x32vmP
Re last RT, see e.g. these from @david_colquhoun http://t.co/OBXkCYFPPj and http://t.co/WuW3x32vmP
3/n @muirgray (3) statistics are execrable "Significance was accepted when P<0.05" see http://t.co/ytbygLI5dv
3/n @muirgray (3) statistics are execrable "Significance was accepted when P<0.05" see http://t.co/ytbygLI5dv
“@HetanShah: Significance tests aren't as significant as you think @david_colquhoun - http://t.co/Vh04mVhhJx”< @tombennett71 @dylanwiliam
“@HetanShah: Significance tests aren't as significant as you think @david_colquhoun - http://t.co/Vh04mVhhJx”< @tombennett71 @dylanwiliam
.@DavidPriceUCL Absolutely But if you want to know why, better read http://t.co/ytbygLI5dv http://t.co/PGv9yfqqsB
Why significance tests aren't as significant as you think @david_colquhoun - http://t.co/qKxQZoavGg
.@DavidPriceUCL Absolutely But if you want to know why, better read http://t.co/ytbygLI5dv http://t.co/PGv9yfqqsB
.@DavidPriceUCL Absolutely But if you want to know why, better read http://t.co/ytbygLI5dv http://t.co/PGv9yfqqsB
.@DavidPriceUCL Absolutely But if you want to know why, better read http://t.co/ytbygLI5dv http://t.co/PGv9yfqqsB
.@DavidPriceUCL Absolutely But if you want to know why, better read http://t.co/ytbygLI5dv http://t.co/PGv9yfqqsB
RT @pash22 An investigation of the false discovery rate and the misinterpretation of p-values http://t.co/SlZMRM3bA3 via @david_colquhoun
An investigation of the false discovery rate and the misinterpretation of p-values http://t.co/w5HjpYmGec via @david_colquhoun
I hope @deevybee 's meeting has something about false discovery rates. A major reason for irreproducibility? http://t.co/ytbygLI5dv
I hope @deevybee 's meeting has something about false discovery rates. A major reason for irreproducibility? http://t.co/ytbygLI5dv
.@stephensenn Not helpful. Impact of http://t.co/ytbygLI5dv far greater than http://t.co/nc9PNHUr8V but latter much better #HEFCEmetrics
fab look at misinterpreting stats. NMS students look at pt3 when consid sensitivity and specificity studies. http://t.co/5j1OBXdui6
@smarfdoc @NPRHealth http://t.co/sLXWhu0n2J & http://t.co/E1eLkJKGtG :) no gluten-effect/dose-response http://t.co/XVvrfNmPoN
fab look at misinterpreting stats. NMS students look at pt3 when consid sensitivity and specificity studies. http://t.co/5j1OBXdui6
fab look at misinterpreting stats. NMS students look at pt3 when consid sensitivity and specificity studies. http://t.co/5j1OBXdui6
fab look at misinterpreting stats. NMS students look at pt3 when consid sensitivity and specificity studies. http://t.co/5j1OBXdui6
Quote: If you use p=0.05 to suggest that you have made a discovery, you will be wrong at least 30% of the time. http://t.co/ysWUfr9LHv T/F?
An investigation of the false discovery rate and the misinterpretation of p-values http://t.co/aj9HYzoIZD RT @tomcrowthersoil #research #sci
An investigation of the false discovery rate and the misinterpretation of p-values http://t.co/aj9HYzoIZD RT @tomcrowthersoil #research #sci
@JPdeRuiter reminds me of a talk I once heard you give http://t.co/BcEbhpT45h
Interesting read on the dangers of p<0.05 and why the false discover rate is often much higher http://t.co/YN1S7qS061
If you use p=0.05 to suggest that you have made a discovery, you will be wrong at least 30% of the time: http://t.co/eMYIc2Zpxh
Great review on the dangers of p values from @david_colquhoun Compulsory reading for all PhD students and post-docs. http://t.co/gHEnwv4JA7
@robinince @ProfBrianCox @bengoldacre @sarapascoe @jrf1968 p=0.05 no longer good enough, it gets even harder for woo http://t.co/OrHD5Uy0qT
An investigation of the false discovery rate and the misinterpretation of p-values http://t.co/XqleopIfOZ
"If you use p=0.05 to suggest that you have made a discovery, you will be wrong at least 30% of the time." http://t.co/C3FUaBNrgd
Interesting read on the dangers of p<0.05 and why the false discover rate is often much higher http://t.co/YN1S7qS061
Interesting read on the dangers of p<0.05 and why the false discover rate is often much higher http://t.co/YN1S7qS061
An investigation of the false discovery rate and the misinterpretation of p-values http://t.co/XqleopIfOZ
“Never, ever use the word ‘significant’ in a paper." @david_colquhoun on the problems with P values. http://t.co/q18WmLAfq5 @royalsociety
An investigation of the false discovery rate and the misinterpretation of p-values http://t.co/XqleopIfOZ
Great review on the dangers of p values from @david_colquhoun Compulsory reading for all PhD students and post-docs. http://t.co/gHEnwv4JA7
“Never, ever use the word ‘significant’ in a paper." @david_colquhoun on the problems with P values. http://t.co/q18WmLAfq5 @royalsociety
"If you use p=0.05 to suggest that you have made a discovery, you will be wrong at least 30% of the time" http://t.co/Bo8DwxYNry
@GdnHigherEd And "Never, ever, use the word ‘significant’ in a paper. It is arbitrary, and... deeply misleading" via http://t.co/w4buTlLA2I
@GdnHigherEd And "Never, ever, use the word ‘significant’ in a paper. It is arbitrary, and... deeply misleading" via http://t.co/w4buTlLA2I
@GdnHigherEd And "Never, ever, use the word ‘significant’ in a paper. It is arbitrary, and... deeply misleading" via http://t.co/w4buTlLA2I
“Never, ever use the word ‘significant’ in a paper." @david_colquhoun on the problems with P values. http://t.co/q18WmLAfq5 @royalsociety
@GdnHigherEd And "Never, ever, use the word ‘significant’ in a paper. It is arbitrary, and... deeply misleading" via http://t.co/w4buTlLA2I
@GdnHigherEd And "Never, ever, use the word ‘significant’ in a paper. It is arbitrary, and... deeply misleading" via http://t.co/w4buTlLA2I
@GdnHigherEd And "Never, ever, use the word ‘significant’ in a paper. It is arbitrary, and... deeply misleading" via http://t.co/w4buTlLA2I
@GdnHigherEd And "Never, ever, use the word ‘significant’ in a paper. It is arbitrary, and... deeply misleading" via http://t.co/w4buTlLA2I
“Never, ever use the word ‘significant’ in a paper." @david_colquhoun on the problems with P values. http://t.co/q18WmLAfq5 @royalsociety
“Never, ever use the word ‘significant’ in a paper." @david_colquhoun on the problems with P values. http://t.co/q18WmLAfq5 @royalsociety
@GdnHigherEd And "Never, ever, use the word ‘significant’ in a paper. It is arbitrary, and... deeply misleading" via http://t.co/w4buTlLA2I
“Never, ever use the word ‘significant’ in a paper." @david_colquhoun on the problems with P values. http://t.co/q18WmLAfq5 @royalsociety
“Never, ever use the word ‘significant’ in a paper." @david_colquhoun on the problems with P values. http://t.co/q18WmLAfq5 @royalsociety
An investigation of the false discovery rate and the misinterpretation of p-values ;Colquhoun, Ioannidis RCT http://t.co/vgzHv2dMwx
“Never, ever use the word ‘significant’ in a paper." @david_colquhoun on the problems with P values. http://t.co/q18WmLAfq5 @royalsociety
“Never, ever use the word ‘significant’ in a paper." @david_colquhoun on the problems with P values. http://t.co/q18WmLAfq5 @royalsociety
Good read on the bullshit that we call significant data. http://t.co/p0ikZnGtf0
Tha last quotation from DC in 1971/ Still true in 2014 http://t.co/ytbygLI5dv @stephensenn #BayesWars
“Never, ever use the word ‘significant’ in a paper." @david_colquhoun on the problems with P values. http://t.co/q18WmLAfq5 @royalsociety
Tha last quotation from DC in 1971/ Still true in 2014 http://t.co/ytbygLI5dv @stephensenn #BayesWars
Tha last quotation from DC in 1971/ Still true in 2014 http://t.co/ytbygLI5dv @stephensenn #BayesWars
“Never, ever use the word ‘significant’ in a paper." @david_colquhoun on the problems with P values. http://t.co/q18WmLAfq5 @royalsociety
@stephensenn @freakonometrics see sections 10,11 http://t.co/ytbygLI5dv The *minimum" false discovery rate for any prior is around 30%
“Never, ever use the word ‘significant’ in a paper." @david_colquhoun on the problems with P values. http://t.co/q18WmLAfq5 @royalsociety
Lies, damn lies, and p values: Great paper on the problems of statistics by @david_colquhoun http://t.co/frlTphFaeS @royalsociety
“Never, ever use the word ‘significant’ in a paper." @david_colquhoun on the problems with P values. http://t.co/q18WmLAfq5 @royalsociety
“Never, ever use the word ‘significant’ in a paper." @david_colquhoun on the problems with P values. http://t.co/q18WmLAfq5 @royalsociety
“Never, ever use the word ‘significant’ in a paper." @david_colquhoun on the problems with P values. http://t.co/q18WmLAfq5 @royalsociety
“Never, ever use the word ‘significant’ in a paper." @david_colquhoun on the problems with P values. http://t.co/q18WmLAfq5 @royalsociety
RT @freakonometrics: An investigation of the false discovery rate and the misinterpretation of p-values" http://t.co/583mQBMV6P #ecostatsCo
“@freakonometrics: "investigating false discovery rate, misinterpretation of p-values" http://t.co/68L9zfLATQ by @david_colquhoun” #p_value
has simply been an abysmal failure" Sellke et al 2001 http://t.co/ytbygLI5dv @freakonometrics
MT @glbryson The dangers of underpowered studies even when statistical significance achieved. http://t.co/3mcE5D1tfF
@kevin2kane @inductivestep But clinical usefulness is all that matters! (and never use the word significance! http://t.co/ytbygLI5dv )
@kevin2kane @inductivestep both homeop & SSRIs have neg and pos RCTs if you call P=0.04 "positive" -I don't http://t.co/ytbygLI5dv
@kevin2kane @inductivestep it does in some -usual abuse of P values http://t.co/ytbygLI5dv The negative RCTs for SSRIs were hidden by GSK
@inductivestep looks like fun. In my R script in http://t.co/ytbygLI5dv you specify limits for P. eg 0 to 0,05 OR 0.045 to 0.05
@inductivestep @DavidPriceUCL @stephensenn prob with table is it doesn't take into account P =0.05 vs P<0.05 Para 10 http://t.co/ytbygLI5dv
@inductivestep he's pulling my leg -just because I said " never *ever" use the word significant" @DavidPriceUCL http://t.co/ytbygLI5dv
@DavidPriceUCL @UnhealthyEcon Never *ever* use the word significant http://t.co/ytbygLI5dv
More on p-value controversies (since Ioannidis 2005) : http://t.co/UaYVL6l2u9 #hypothesistesting Contrast with Oreskes on climate change...
How (not) to abuse p-values http://t.co/YtWAByqrsl
@AlexanderEtz statistically, false negatives will be rare if prevalence of correct hyoptheses is low -P(real) in http://t.co/ytbygLI5dv
How (not) to abuse p-values http://t.co/YtWAByqrsl
@lakens @chrisdc77 @rallen81 what about false discovery rates? http://t.co/ytbygLI5dv Surely that contributes?
False discovery rate is a lot higher than most think and too often p-values are misinterpreted http://t.co/xIzjhKK9H2
If you use p=0.05 to suggest that you have made a discovery, you will be wrong at least 30% of the time. Read this. http://t.co/pT7XCWQ9ih
If you use p=0.05 to suggest that you have made a discovery, you will be wrong at least 30% of the time. Read this. http://t.co/pT7XCWQ9ih
If you use p=0.05 to suggest that you have made a discovery, you will be wrong at least 30% of the time. Read this. http://t.co/pT7XCWQ9ih
If you use p=0.05 to suggest that you have made a discovery, you will be wrong at least 30% of the time. Read this. http://t.co/pT7XCWQ9ih
.@pgseye thanks -I corrected that in comments (now vanished). Also corrected in arXiv version which I can update http://t.co/relFUWulfj
Want clarity on significance? Read this: An investigation of the false discovery rate... http://t.co/yw50cTpMgx
If you use p=0.05 to suggest that you have made a discovery, you will be wrong at least 30% of the time. Read this. http://t.co/pT7XCWQ9ih