(Total Views: 742)
Posted On: 09/23/2020 12:08:28 AM
Post# of 154111
Re: (Small) stat penalty to see the data - most interesting tidbit tonight - as many have observed. The following thoughts are entirely IMO:
I like the breakdown into two positive categories of response from the DSMC. The "call off for ethical concerns" sure sounded like it was rare - it felt to me like we'd need p<.01 (or better) to get that designation.
The "if current trends continue, you will be significant" could be interpreted to mean p<.17 (z=-1.37) - as then doubling the population shortens the confidence interval 30% (=1/sqrt(2)) to get z=-1.96.
As to the stat cost of better information - p<.045 vs. p<.050 sounds cheap to me - just publicizing very good results would presumably speed recruitment - perhaps CYDY could negotiate an extra 10 participants to offset the stat hit.
I like the breakdown into two positive categories of response from the DSMC. The "call off for ethical concerns" sure sounded like it was rare - it felt to me like we'd need p<.01 (or better) to get that designation.
The "if current trends continue, you will be significant" could be interpreted to mean p<.17 (z=-1.37) - as then doubling the population shortens the confidence interval 30% (=1/sqrt(2)) to get z=-1.96.
As to the stat cost of better information - p<.045 vs. p<.050 sounds cheap to me - just publicizing very good results would presumably speed recruitment - perhaps CYDY could negotiate an extra 10 participants to offset the stat hit.
(6)
(0)
