Evaluation of Jackknife and Bootstrap for Defining Confidence Intervals for Pairwise Agreement Measures
Rows refer to the methods by which the CIs were calculated. From top to the bottom: bootstrap percentile method, bootstrap BCa method and jackknife. Each dot represents a simulated population (PFT), with a particular set of parameters, and the average amplitude of the CIs for 1000 samples from the population (CTs). Symbols and colors represent changes in: dimensions of the simulated probability tables, corresponding to the number of clusters in each of the two classifications (left); exponent alpha of the Zipfian distribution determining the distribution of row cluster sizes of the simulated probability tables (middle); sample size or number of elements in the contingency tables (right).