Sample-size calculator for logistic regression

This is an R program that calculates the sample size required for data that will be analyzed by logistic regression. It is written by Jaymie Strecker and implements formulas by Hsieh, Bloch, and Larsen. You can read more about these formulas in Appendix B and Section 4.4.8 of my dissertation.

Files to download:

• Sample-size calculator
• Tests of the calculator implementation against examples from Hsieh et al. (optional)
• Example data file (optional)

To see usage information, run the program with no arguments to R (nothing after --args). For example:

R --no-save --slave --args < logistic_sample_size.R

To calculate the sample size for the example data file (above), run the program like this:

R --no-save --slave --args data_for_sample_size2.R < logistic_sample_size.R

This is the output you should get:

Dependent variable: Det

Sample size for each independent at each effect size:
                      0.318 0.477 0.953
LineCov                 544   242    61
NormEvents             1201   534   134
NormE2Cov              1198   533   133
MutType                 399   399   399
NormMaxEventSuccs       664   295    74
NormMinLinesCovBefore   580   258    65
NormMinLinesCovWith     593   263    66
AllRepeat               926   926   926


Maximum sample size at each effect size:
    0.318 0.477 0.953
max  1201   926   926

If you have any questions or notice any bugs, please e-mail me at the address given on my home page.