@NiuSanford @5_utr Cool 👍 as you know… the more stats you know the scarier the literature becomes. I was thinking about writing a Stats for the people on this topic looking at the PREMER HA PCI trial. I also would read this great paper by @VickersBiostats
@TomKealy2 Might be worth checking out this paper by @VickersBiostats https://t.co/07NX79trKQ. The recommendation in biostats circles is to basically always use ANCOVA
@GMFHx @HedonicMicrobes I have some concerns about this study. Firstly, they reported a non-significant result (-2.27±0.92kg, P=0.09) as a "slightly decreased body weight" - a red flag for suggesting the effect. Moreover, % change from baseline is not a re
@PloederlM @IoanaA_Cristea @David_NeverDave Indeed, he shows that, if any approach is fatally flawed here, it's change from baseline and percentage change. Reduces power, becomes weird if variances change, etc. The field really should stick to post-treatme
@LoriMichener @ExPhysStudent @CorKinetic @Bill_Vicenzino @chadcookpt @chrisB0Y3R @Henrik_Riel @MarinusWinters It's not strictly the same but it is similar. The Vickers paper might be more appropriate but any ratio (including percent change) is susceptible
RT @JamesSteeleII: @DrStrength4Life This is actually a area where statisticians are pretty much in agreement... ANCOVA is best. @GregTees s…
@DrStrength4Life This is actually a area where statisticians are pretty much in agreement... ANCOVA is best. @GregTees shared some good points in this thread https://t.co/EIZdm1OGrL Also check out some of @VickersBiostats papers e.g.https://t.co/aEnAa5eVi
Ratios & % changes. Atkinson, Batterham, https://t.co/4J4RmYYVgX Lolli et al. Appropriate modelling of “digit ratios”, https://t.co/U2bV5S8pfY Vickers – concerns about percentage changes, https://t.co/RkeHVBk0Tf Atkinson, Batterham – FMD%, https://t
4/7. This is so even though the Grp x time approach can be misapplied in context of parallel group trials, e.g. https://t.co/X4Fo4yJTs6 . Also, using % changes might not get you off the hook, e.g. https://t.co/RkeHVBk0Tf
@felipe_mattioni Worth a read: https://t.co/cQlhY04YQp
RT @BenVanCalster: @EJWagenmakers @stephensenn @f2harrell Not sure whether this was mentioned anywhere already: https://t.co/QpA7K1O7Tq
@EJWagenmakers @stephensenn @f2harrell Not sure whether this was mentioned anywhere already: https://t.co/QpA7K1O7Tq
เปเปอร์นี้เปรียบเทียบ การใช้ outcome ใน 4 รูปแบบ ด้วยการทำ simulation เพื่ออธิบายว่า ทำไมไม่ควรใช้แบบที่ %change ในการวิเคราะห์ข้อมูลทางสถิติ https://t.co/QAZEz8xUqE
RT @DrGBuckingham: Do you have baseline differences between your groups in a learning task? Do you deal with this by analysing % change fro…
Ok #statstwitter a question... For between group pre-post test ANCOVA with baseline covariate has greatest power (https://t.co/yKTSknuUym)... is it fair to assume that the same would be the case for within participant pre-post test with a baseline covaria
RT @DrGBuckingham: Do you have baseline differences between your groups in a learning task? Do you deal with this by analysing % change fro…
RT @DrGBuckingham: Do you have baseline differences between your groups in a learning task? Do you deal with this by analysing % change fro…
RT @DrGBuckingham: Do you have baseline differences between your groups in a learning task? Do you deal with this by analysing % change fro…
RT @DrGBuckingham: Do you have baseline differences between your groups in a learning task? Do you deal with this by analysing % change fro…
RT @DrGBuckingham: Do you have baseline differences between your groups in a learning task? Do you deal with this by analysing % change fro…
RT @DrGBuckingham: Do you have baseline differences between your groups in a learning task? Do you deal with this by analysing % change fro…
RT @DrGBuckingham: Do you have baseline differences between your groups in a learning task? Do you deal with this by analysing % change fro…
RT @DrGBuckingham: Do you have baseline differences between your groups in a learning task? Do you deal with this by analysing % change fro…
RT @DrGBuckingham: Do you have baseline differences between your groups in a learning task? Do you deal with this by analysing % change fro…
Do you have baseline differences between your groups in a learning task? Do you deal with this by analysing % change from baseline? This paper suggests that this method really hurts your power, and you should be running an ANCOVA instead https://t.co/d8W
Interesting argument! #meded #medtwitter
RT @kaz_yos: The use of percentage change from baseline as an outcome in a controlled trial is statistically inefficient: a simulation stud…
RT @kaz_yos: The use of percentage change from baseline as an outcome in a controlled trial is statistically inefficient: a simulation stud…
“percentage change from baseline is statistically inefficient. Perhaps counterintuitively, it does not correct for imbalance between groups at baseline. It may also create a non-normally distributed statistic from normally distributed data.” https://t.co/R
“because it incorporates both baseline and post-treatment scores, it would appear to control for any chance baseline imbalance between groups. However, this is not the case because of regression to the mean: FRACTION will create a bias towards the grou...”
“Theoretical considerations suggest that percentage change from baseline will also fail to protect from bias in the case of baseline imbalance” https://t.co/RH6UeKdKX4
The use of percentage change from baseline as an outcome in a controlled trial is statistically inefficient: a simulation study | BMC Medical Research Methodology | Full Text https://t.co/RH6UeKdKX4
RT @beagoulao: @statsepi He's also the author of the first stats simulation paper @gsmaclennan gave me to read: https://t.co/G3VII4ubHs
@statsepi He's also the author of the first stats simulation paper @gsmaclennan gave me to read: https://t.co/G3VII4ubHs
@RSMetcalfe_PhD @philw80 @Alan_Batterham And this one as a note of caution about using percentage changes: https://t.co/RkeHVBk0Tf
@rrangana1 This paper says ANCOVA has the highest power to detect an effect in this situation https://t.co/d8WQcI8fW4
RT @dennislendrem: Or, don't use percentage change from baseline as an outcome.... https://t.co/T3SMYgwcKh #bmcmedresmethodol
RT @dennislendrem: Or, don't use percentage change from baseline as an outcome.... https://t.co/T3SMYgwcKh #bmcmedresmethodol
This paper has a great choice of references 😉 https://t.co/rc2790xYR5
Or, don't use percentage change from baseline as an outcome.... https://t.co/T3SMYgwcKh #bmcmedresmethodol
@TBandholm @MichaelRathleff ..., is highly sensitive to changes in variance and fails to protect from bias http://t.co/LTd2SaSQqk