Is that supposed to be a serious study!? If so, it's utterly bonkers. Eg, we don't know anything about respective training loads before the 2 races. Or anything about a myriad of other potentially confounding variables.
-
-
Replying to @PhilosophyExp @lo_stats and
2 minute read. https://www.sportsci.org/2016/WCPASabstracts/ID-1699.pdf# …
2 replies 0 retweets 5 likes -
Replying to @FondOfBeetles @PhilosophyExp and
I don't think I've ever seen such high p values quoted in a published study! p=1 means data is no more than random chance. p<0.05 is valid result. The Harper study is recording p values of 0.84 and 0.68!! https://pdfs.semanticscholar.org/1e6a/bd2c1e03ba88e9ac8da94ea1d69ff3f4878a.pdf?_ga=2.254440527.659551599.1550520323-1192624875.1550520323 …pic.twitter.com/Twp4KfJA72
2 replies 5 retweets 14 likes -
Replying to @fairplaywomen @FondOfBeetles and
Yes, but they're trying to make the point there isn't a significant difference. That's a win for them.
1 reply 0 retweets 1 like -
Replying to @PhilosophyExp @fairplaywomen and
If a reasonable scientist was there, they should of asked what was the design's ability to detect meaningful differences. The answer would of not been convincing.
2 replies 0 retweets 2 likes -
Replying to @lo_stats @fairplaywomen and
Yes, of course. But in terms of their crappy design, null hypothesis is no difference between age-graded scores pre & post-transition (because post score is age-graded score for women). High p-values means you fail to reject null hypothesis. That's why they're being quoted.
1 reply 0 retweets 0 likes -
Replying to @PhilosophyExp @fairplaywomen and
The scientifically responsible strategy for them would of been to 1) perform an equivalence test (but the n requirement would be too large), or 2) show that if there was a true important difference they would likely have significant p-values. They went with neither strategy.
1 reply 0 retweets 1 like -
Replying to @lo_stats @fairplaywomen and
Pre-test, post-test design in real world conditions, given a gap of a year, is always going to be nightmarisly difficult to control. I'm not sure if could be done.
1 reply 0 retweets 0 likes -
Replying to @PhilosophyExp @fairplaywomen and
Not correctly with that sample size and that design. Personally I would of considered a cohort design with a biological women control group matched by age and maybe some other factors.
2 replies 0 retweets 1 like -
Replying to @lo_stats @PhilosophyExp and
But if you are presenting studies like the one they did, you probably don't have the know-how to do what I just proposed.
1 reply 0 retweets 0 likes
Replying to @lo_stats @fairplaywomen and
Plus you might get the wrong result...
1:25 PM - 7 Mar 2019
0 replies
0 retweets
0 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.