Target standard error a design implies, for the KL calibration
Source:R/npp_kl_calibration.R
npp_kl_expected_target_se.RdThe standard error of the target estimate expected under the
design, on the scale of the analysis. For a continuous, recurrent-event or
time-to-event endpoint it is the design's sampling standard deviation over
the root sample size per arm, which does not depend on the treatment
effect. For a binary endpoint the standard error depends on the response
rates, so taking it at the scenario's true rates would make the prior a
function of the true treatment effect. It is taken instead at the design's
control rate and the source treatment rate, i.e. at zero treatment drift,
which is also the design the analysis step calibrates on (see
compute_bayesian_ocs()).