[R] Coverage Probability

[Ben Bolker]

Muhammad  Kashif <mkashif <at> uaf.edu.pk> writes:

> 
> Dears
> Can anyone help me to solve the issue.
> 
> By using" boot" and "boot.ci" package in R we can construct bootstrap
confidence intervals. How we
> calculate the coverage probability of these intervals.

  Calculating coverage probability for any but the simplest
cases requires simulations.  You need to simulate a particular
scenario; for each simulation, use your estimation and confidence-interval
calculation procedure; and then score the particular simulation
as 1 (estimated CI included the true parameter value) or 0
(true parameter value fell outside the estimated CI). Across
a large number of simulations, the proportion of ones is
an estimate of the coverage.

This is discussed more here:

http://ms.mcmaster.ca/~bolker/emdbook/chap5A.pdf