RT @david_colquhoun: The tyranny of P=0.05 is hard to break. It's disastrous Results in many false positives https://t.co/B9Kqtayspo https…
RT @tadhg50: An investigation of the false discovery rate and the misinterpretation of p-values | Open Science @SameiHuda https://t.co/qFN…
An investigation of the false discovery rate and the misinterpretation of p-values | Open Science @SameiHuda https://t.co/qFNM7b0M6U
RT @david_colquhoun: The tyranny of P=0.05 is hard to break. It's disastrous Results in many false positives https://t.co/B9Kqtayspo https…
RT @david_colquhoun: The tyranny of P=0.05 is hard to break. It's disastrous Results in many false positives https://t.co/B9Kqtayspo https…
RT @david_colquhoun: The tyranny of P=0.05 is hard to break. It's disastrous Results in many false positives https://t.co/B9Kqtayspo https…
The tyranny of P=0.05 is hard to break. It's disastrous Results in many false positives https://t.co/B9Kqtayspo https://t.co/rNJd82WZWQ
An investigation of the false discovery rate and the misinterpretation of p-values | Open Science #yayStats https://t.co/dUjrqZFEiq
@SiMarwood In any case, P values so marginal that false positive rate likely to be huge https://t.co/B9Kqtayspo @patient
@IanLaneSci agreed. But neither P values nor confidence intervals tell you much about the false positive rate https://t.co/B9Kqtayspo
RT @BroadhurstDavid: @dbkell @NancyXuanHe @metabolomics I have always liked this one from David Colquhoun https://t.co/kaoYaHMxrb
RT @beala: "If you use p=0.05 to suggest that you have made a discovery, you will be wrong at least 30% of the time." https://t.co/I0BJhW0W…
RT @beala: "If you use p=0.05 to suggest that you have made a discovery, you will be wrong at least 30% of the time." https://t.co/I0BJhW0W…
RT @beala: "If you use p=0.05 to suggest that you have made a discovery, you will be wrong at least 30% of the time." https://t.co/I0BJhW0W…
"If you use p=0.05 to suggest that you have made a discovery, you will be wrong at least 30% of the time." https://t.co/I0BJhW0Wpr
RT @david_colquhoun: I think you forgot the false positive rate in single unbiased tests https://t.co/B9Kqtayspo https://t.co/h2JTESqXZi
RT @david_colquhoun: I think you forgot the false positive rate in single unbiased tests https://t.co/B9Kqtayspo https://t.co/h2JTESqXZi
I think you forgot the false positive rate in single unbiased tests https://t.co/B9Kqtayspo https://t.co/h2JTESqXZi
@cluckin change the words used to describe P values https://t.co/FSTLju0kdp
@deevybee there is no escaping use of a lower P value (apart from replicating yourself) https://t.co/FSTLju0kdp
@deevybee actually the false positive rate is independent of power (but depends strongly on prior) https://t.co/B9Kqtayspo #medscilife
RT @BroadhurstDavid: @dbkell @NancyXuanHe @metabolomics I have always liked this one from David Colquhoun https://t.co/kaoYaHMxrb
RT @BroadhurstDavid: @dbkell @NancyXuanHe @metabolomics I have always liked this one from David Colquhoun https://t.co/kaoYaHMxrb
@dbkell @NancyXuanHe @metabolomics I have always liked this one from David Colquhoun https://t.co/kaoYaHMxrb
RT @david_colquhoun: I hope somebody deals with the misinterpretation of P values https://t.co/B9Kqtayspo https://t.co/TuNcdojwzE
RT @david_colquhoun: I hope somebody deals with the misinterpretation of P values https://t.co/B9Kqtayspo https://t.co/TuNcdojwzE
I hope somebody deals with the misinterpretation of P values https://t.co/B9Kqtayspo https://t.co/TuNcdojwzE
@DerekGriffin86 Oh dear. It's abundantly clear that Cummings doesn't understand P values https://t.co/B9Kqtayspo
RT @BrendanEich: @sapinker previously: https://t.co/1LmLVCPzu7
An investigation of the false discovery rate and the misinterpretation of p-values https://t.co/2y1n8c2mid
@nntaleb distribution of p-values. https://t.co/qbx8vy9ba2
@sapinker previously: https://t.co/1LmLVCPzu7
"If you use p=0.05 to suggest that you have made a discovery, you will be wrong at least 30% of the time" https://t.co/pnMW0qOS0r
Now is the time to read or re-read @david_colquhoun on the danger of using p values as a signal of a discovery https://t.co/CaQk0gPuDp
RT @david_colquhoun: Amazed to see my P value paper has passed 15,000 pdf downloads https://t.co/tNBlxLKq4G
RT @david_colquhoun: Amazed to see my P value paper has passed 15,000 pdf downloads https://t.co/tNBlxLKq4G
Amazed to see my P value paper has passed 15,000 pdf downloads https://t.co/tNBlxLKq4G
RT @janzilinsky: Also: An investigation of the false discovery rate and the misinterpretation of p-values - the paper is ungated! https://t…
On p 328 he assumes prior prob of psyc theories being "true" is 10%. Colquhoun assumes same rate in his simulations https://t.co/m5VcFNxjiu
@JFattaccioli ce tw n'ets pas passé https://t.co/q3nfIAD6oD lui il sait ..maybe!
@JFattaccioli ben comme bcp de biologiste je sais pas....surtout avec n=3:-) sinon lio il sait https://t.co/q3nfIAD6oD @david_colquhoun
P values prevent you from making a fool of yourself, they do not to make unpublishable results publishable. https://t.co/KlLrqUZV0c
RT @janzilinsky: Also: An investigation of the false discovery rate and the misinterpretation of p-values - the paper is ungated! https://t…
Also: An investigation of the false discovery rate and the misinterpretation of p-values - the paper is ungated! https://t.co/JJJHWKSU4S
RT @david_colquhoun: Very obvious that it was a victim of #squabblingstatiticians This might be clearer? https://t.co/B9Kqtayspo https://t…
Very obvious that it was a victim of #squabblingstatiticians This might be clearer? https://t.co/B9Kqtayspo https://t.co/f4rtwcwSYa
Even citations can't measure quality, as my recent experience shows dramatically https://t.co/tNBlxLKq4G https://t.co/cpa2g5y0ql
Use https://t.co/iVVwyptdxJ as a chaser on previous tweet.
"The false discovery rate and the misinterpretation of p-values".. Perspectives for future thought! https://t.co/LXtBX0ACzX
RT @BraveNewClimate: Investigation of false discovery rate & misinterpretation of p-values https://t.co/fLl4ju8VrW Even if you (unwisely) u…
RT @BraveNewClimate: Investigation of false discovery rate & misinterpretation of p-values https://t.co/fLl4ju8VrW Even if you (unwisely) u…
@stephensenn not "when pressed". I said it explicitly in the original paper! https://t.co/FSTLju0kdp @hildabast
@lakens and even that wording is overoptimistic for implausible hypotheses https://t.co/STKoQl6y5p (see the Fig) @theNASEM @hildabast
@lakens NO!. They should all be published -just use the right words eg https://t.co/FSTLju0kdp @theNASEM @hildabast
@sole1810 the question that matters, I think, is what's the probablility that you have a false postive https://t.co/B9Kqtayspo
"If you use p=0.05 to suggest that you have made a discovery, you will be wrong at least 30% of the time." https://t.co/bOqOaB2pOz
RT @BraveNewClimate: Investigation of false discovery rate & misinterpretation of p-values https://t.co/fLl4ju8VrW Even if you (unwisely) u…
RT @BraveNewClimate: Investigation of false discovery rate & misinterpretation of p-values https://t.co/fLl4ju8VrW Even if you (unwisely) u…
Investigation of false discovery rate & misinterpretation of p-values https://t.co/fLl4ju8VrW Even if you (unwisely) use p-values, take care
but it misses out on false positive rate https://t.co/B9Kqtayspo https://t.co/PaVmXRK5XT
@chuhnk Very marginal P values (see https://t.co/B9Kqtayspo ) and no multiple comparison correction, as far as I can see
.@CarmelHannan The real answer is in sec10, but perhaps that could be omitted? https://t.co/fvwZ4THNj8 @BrendanTHalpin @eoinefla @MarkWard_
.@CarmelHannan well Figs 1 & 2 in https://t.co/X1NEc1p2Lj don't need maths. And you can say interpret P as at https://t.co/Hb7so5LL6d
@SocSciMcAndrew @BrendanTHalpin @eoinefla @MarkWard__ lots of good examples by @david_colquhoun here https://t.co/iJK1wHVe5d
@vgr This shows why "false discovery rate" is >> p-value, one root of "reproduciblity crisis" in science. https://t.co/4e5MY0Ssdo
Direct link to paper, which is also a very fun read: https://t.co/DcPg3nqxWn
Thought-provoking paper, via @F1000, on how p<0.05 *dramatically* underestimates FDR in most experimental settings. https://t.co/9mD55p95hu
@SoofiAziz anf if you want to know why a small number of trials seem to show an effect, read this https://t.co/X1NEc1p2Lj @mc_hankins
@MrMMarsh @TomChivers It's easy to understand screening test prob https://t.co/KohlxTibTd Significance test prob is similar
@MrMMarsh @TomChivers The most recent comment may clarify it? https://t.co/tHfgrLDQiA
@TomChivers - it's only 5% if NH is true. To see how often you'll be wrong you need to consider also when it's false https://t.co/X1NEc1p2Lj
Good read. Easily digestible:::: An investigation of the false discovery rate and the misinterpretation of p-values https://t.co/WJBkEiqgAm
T-shirt slogan: p ≤ 0.05 ≠ significant. #pvalue #statistics #science https://t.co/qHL33xBrJ9
An investigation of the false discovery rate and the misinterpretation of p-values https://t.co/qfQftzFI3Q
RT @IJNSJournal: An investigation of the false discovery rate and the misinterpretation of p-values... Read, digest and act! https://t.co/m…
An investigation of the false discovery rate and the misinterpretation of p-values... Read, digest and act! https://t.co/mS6UFmlCph
@stephensenn and last week's comment shows why it makes no sense to ignore prevalence https://t.co/tHfgrLDQiA @inductivestep @toates_19
@toates_19 @toates_19 The prob is, there is no unambiguous alternative. You have toget P, but interpret it thus https://t.co/Hb7so5LL6d
Thanks. Server crashed halfway through. but this might help https://t.co/tHfgrLDQiA https://t.co/VqqTTx6nar
Nice but I don't like the way it uses "significant" https://t.co/YAXqofDLg7 https://t.co/JBGbJRRAAV
.@statsmethods I don't entirely agree, cos P values are the only thing you can calc unambiguously Must change words https://t.co/Hb7so5LL6d
After 3 days, ny comment about P values has eventually appeared https://t.co/tHfgrLDQiA
RT @david_colquhoun: And that's before you even get to thinking about the false postive rate https://t.co/X1NEc1p2Lj https://t.co/423xmH…
And that's before you even get to thinking about the false postive rate https://t.co/X1NEc1p2Lj https://t.co/423xmHhhOT
RT @iamdavecampbell: Easy reading: ’An investigation of the false discovery rate and the misinterpretation of p-values’: https://t.co/SFfhC…
RT @iamdavecampbell: Easy reading: ’An investigation of the false discovery rate and the misinterpretation of p-values’: https://t.co/SFfhC…
Easy reading: ’An investigation of the false discovery rate and the misinterpretation of p-values’: https://t.co/SFfhCzLbBj
RT @david_colquhoun: OMG that must be one of the best yet -obvioulsy hasn't read https://t.co/X1NEc1p2Lj https://t.co/U6CUFy0JqH
RT @david_colquhoun: OMG that must be one of the best yet -obvioulsy hasn't read https://t.co/X1NEc1p2Lj https://t.co/U6CUFy0JqH
OMG that must be one of the best yet -obvioulsy hasn't read https://t.co/X1NEc1p2Lj https://t.co/U6CUFy0JqH
Excellent presentation of the argument. Note that the statistics are not the problem. TLDR: Die NHST, die! https://t.co/xksXOBpd9S
@skyking171717 @CaloriesProper and another: https://t.co/ZtZ95Jv1lQ
RT @david_colquhoun: Yesterday my P value paper passed 100,000 full text views, despite being unoriginal. So much for #metrics https://t.co…
Yesterday my P value paper passed 100,000 full text views, despite being unoriginal. So much for #metrics https://t.co/lFsDIx7TqI
@TomChivers Better make sure you've read https://t.co/fvwZ4THNj8 first. Come to talk on 27th? https://t.co/cgzVHKTYYl
RT @raferh: Best description I've found so far on why p-value != false discovery rate: https://t.co/0jzWIuX8Sm
Best description I've found so far on why p-value != false discovery rate: https://t.co/0jzWIuX8Sm
@juancommander @theWinnower I like to point to @david_colquhoun paper published in @royalsociety open science: https://t.co/vCfC1fzfYk