A statistcs calculation29 Sep 2020 18:55
VAL have reported a 54.5% success rate on a sample size of 11. Small sample we say. But how much does this matter?
The smallness of the sample can lead to non-significant results and therefore are not much use. A larger sample is always more robust.
In this case, using a simple Binomial model, I have calculated a range of values we can expect from our observation of 6 out of 11 "successes". There is a 95% probabilty that the actual success rate could be anywhere between 25% and 86% . This to me is very encouraging, because, even on our sample of 11 it is likely that a quarter of all patients would respond positively to the treatment, and it may be substanially more.
I look forward to the fuller data analysis, and further progress with 201, 301 and BC201.
P.S. I've added a few more today too.