v1.01 (4 April 2002) -------------------- * Eliminated global variable NonParBootInds. Now function BootstrapReplications returns nested list {{...boot repl.},{...inds.}} in case of nonparametric mode and {{...boot repl.},{}} in case of parametric mode, instead of the list {...boot repl.} in v1.0. * Output from BootstrapReplications can be read with {boot1,nbind}=BootstrapReplications[] with user variable nbind containing indices previously held in NonParBootInds. v1.1 (22 April 2002) -------------------- * Removed function AchievedSignificanceLevel. * Rewrote function BootstrapResamples. Previously, number of resamples was calculated incorrectly because an equation has two roots and only the larger one has to be taken. Advice from Prof. Stephen M.S. Lee on this is warmly appreciated. Also, the function now can calculate numbers of resamples for one-sided and two-sided percentile intervals. * Augmented function BootstrapCI. It can now calculate both one-sided and two-sided intervals. If asked an extreme percentile interval, it calculates the number of resamples and checks the dimension of input bootstrapped list. v1.11 (23 April 2002) --------------------- * Changed default option for ConfidenceInterval to TwoSidedAsymmetrical. * Corrected a bug in BootstrapCI that would not recognize TwoSidedAsymmetrical. * Now can pass options to RootSearch through BootstrapResamples and BootstrapCI. * Updated Bootstrap.nb; most important change is in section describing option ConfidenceLevel. Pointed out there that for two-sided equitailed interval with coverage alpha, one should specify ConfidenceLevel->(1+alpha)/2. v1.12 (29 April 2002) --------------------- * Modified the return value of JackknifeReplications in case it failes to calculate: it now returns a list {$Failed, }, where is index of the jackknife sample that causes calculation failure (i.e., index of the element in the data list omission of which causes failure). * Fixed a bug that caused supposedly incorrect calculation of variance of endpoints of BCa intervals: BootstrapReplications was calculating indices of points in a sorted sample data list, but VarianceNonParametricBCaCIEndpoints was using these indices on an unsorted sample data list.