lllcrc model.lllcrc.boots(boot.list)lllcrc-packagecpi0, gives the estimated number of missing units at
each point; the second, mct, gives the number of
units observed at each point. (Dividing the first vector
by the second gives an estimate rate of missingness)vgam.crc calls this function for each bootstrap
replicate. Many replicates together can be used for
empirical confidence interval estimation. The bootstrap is
much like that described by Zwane 2003, but more like
Method 2 in Norris and Polluck 1996, as we repeat the
process of selecting log-linear terms for each bootstrap
iteration, and the multinomial sampling probabilities are
based on raw local estimates instead of post-log-linear
modeling estimates.