The information below is not nearly enough information to work from. We need
to know what the data look like! This could be a feature of the statistical
test, the code, or the particular data set you have in hand.
Terry Therneau
--begin included message ---
hie carried out a comparison of survival curves using the log rank test in R
with the values of rho ranging from -2 to 2 in increments of 0.5 the question
is at rho=-1.5 the power of the test seems to fluctuate or differ from the trend
shown by the other values of rho is the statistical reason for this
difference.ie if the power is increasing as i move from 0 to -2 at rho=-1.5 the
power either greatly increases/decreases then at rho = -2 it follows the general
trend
if any one can help then thanks although this might not be an R question.