empirical_pval

p-values for sample statistics based on estimated empirical null distribution

Syntax

   pval = empirical_pval(x,XNULL,ptails,ksmooth)

Arguments

See also Common variable names and data structures.

input

   x          matrix of statistic values
   XNULL      matrix of null sample statistics
   ptails     Pareto tails lower and upper probabilities (default: no Pareto tails)
   ksmooth    use kernel smoothing to estimate cdf (default: no smoothing)

output

   pval       matrix of p-values

Description

Returns p-values pval for empirical null distribution estimated from sample statistics in XNULL (derived e.g. from a permutation test), evaluated at values in x. The first dimension of XNULL must index samples, while the other dimensions must match the shape of x. NaN s are ignored. See empirical_cdf for details of other parameters.

Important: To test p-values for statistical significance you should correct for multiple null hypotheses; see routine significance.

See also

empirical_cdf | empirical_cdfi | empirical_confint | empirical_cval | significance | tsdata_to_mvgc_permtest | mvgc_demo_permtest | significance