Hahahahahaha that explains quite a bit. I wouldn't put any faith whatsoever in these results lol
-
-
Omg lol just reran their meta-analysis using a random-effects model (it appears that they used FI which is totally inappropriate) and the effect completely disappearspic.twitter.com/GomKPheI1q
5 replies 0 retweets 14 likes -
Can you explain that for the non-statisticians out here? Seems like it might be important to know.
1 reply 0 retweets 0 likes -
Replying to @TamarHaspel @KCKlatt and
So the statistical model they used assumes that all of the trials are essentially the same. But the trials were wildly different - they should have used a model that takes into account the fact that (for example) some of the trials were randomized and some not
2 replies 0 retweets 4 likes -
Replying to @GidMK @TamarHaspel and
If you use a more appropriate model, the result goes away But hilariously this isn't even close to the biggest problem. It's just been pointed out to me that the numbers are just...wrong
1 reply 0 retweets 1 like -
Replying to @GidMK @TamarHaspel and
Like, these numbers that they are combining into a model aren't the IMPROVEMENTS that each group saw they are the numbers at the end of the study. So they're not measuring weightloss, they're measuring WEIGHT...pic.twitter.com/pCXVbFz7mL
1 reply 0 retweets 2 likes -
Replying to @GidMK @TamarHaspel and
The Hussain study, for example, is shown as having a mean difference of 2.54 in favour of the control, but actually they found a much greater weight loss for the intervention (keto) group this group just started out heavier!pic.twitter.com/2ZsZ6WToI0
1 reply 0 retweets 0 likes -
I just glanced at it, but the thing that caught my eye was that weight loss was less than one pound. That doesn't necessarily mean other results are equally meh, but it's likely indicative.
2 replies 0 retweets 1 like -
Replying to @TamarHaspel @KCKlatt and
My point is that it's even worse than that. The control group lost 6.5kg, the intervention lost 12kg, but since they only took the post values (rather than pre-post as you should) they compared the 89kg control with 92kg intervention which meant the intervention did worse
1 reply 0 retweets 2 likes -
Ach. The original tweet with the link is deleted so now I don't have it to look at the data but I don't think it matters. Thanks for the analysis. As a journo, I depend on professionals to take good, hard looks a these studies.
2 replies 0 retweets 2 likes
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.
