Table 2.
Efficiency of Beta Approximation Compared with Permutation Test
|
No. of Binomial Replicates Needed to Achieve Accuracyof Current Study Method when |
||||
|
n=1,000a |
n=10,000a |
|||
| P | 95% CIb | No. of Binomial Trials with Same CI Length | 95% CIb | No. of Binomial Trials with Same CI Length |
| .2 | .0397 | 1,562 | .0128 | 15,032 |
| .1 | .0297 | 1,570 | .00959 | 15,045 |
| .05 | .0201 | 1,809 | .00650 | 17,034 |
| .01 | .00667 | 3,420 | .002155 | 32,745 |
| .001 | .00106 | 13,538 | .000340 | 132,778 |
a
Number of random deviates from extreme-value distribution.
b
Length of 95% CI for the P value, estimated by parametric bootstrap.