Using the K-S test for hypothesis testing of spontaneous or minuature
synaptic events has always confused me.
1) Why is it so widely used? It was my understanding that the KS test
was actually invalid unless one of the distributions was completely
known. Does anyone know who set this trend?
2) Does anyone know how to do power calculations on KS tests? Several
times I have run KS tests, comparing two control sections of data. So
long as you get the number of events high enough (~200-500), you can
get completely functionally unimportant differences to be
statistically very significant (0.1% difference in frequency =
3) Despite these qualms, I would really like to institute the KS test
in a script I am writing. Has any one attempted anything similar, or
have some K-S test psuedocode, or anything like that?