Sampsize Home Page

Why sampsize?

I wrote sampsize mostly because I needed a small utility to easily compute sample size for surveys, particularly surveys of events with low probability of occurrence, and sample or power of comparative studies. Most commercial statistical software have a sample size routine, but none quite like this one. Also, I needed a unix programme that could be easily interfaced with a web page through a cgi script, to fill the needs of researchers in my research Institute. Sampsize is written in C.
A certification script is included in the package.

Web interface to Sampsize calculator

Files

	Download sampsize files
	Documentation html, download pdf file of the manual
	Sampsize 0.6.0 compiles well on the following platforms: i386, Debian (3.0) Sparc - R220, Sun Solaris (8) PPC - G4, Mac OS X (10.2 Server) Sparc Ultra60, Linux 2.2 (Debian 3.0) Alpha, Linux 2.2 (Debian 3.0) J2EE IBM Websphere, Linux 2.4 The self-test script was created on a 32-bit i386 machine. It will fail to certify the accuracy of sampsize when the executable is built on 64-bit machines. The script should then display minor differences in some results, but this is of no practical consequence.

Features

Sample size for prevalence surveys
Finite population correction
Sample size to detect a minimum number of events
Exact binomial confidence intervals
Sample size and power for one-sample comparison of percentage to hypothesized value
Sample size and power for two-sample comparison of percentages
Sample size and power for one-sample comparison of mean to hypothesized value
Sample size and power for two-sample comparison of means
Correction for cluster sampling designs
Sample size and power for unmatched and 1:1 matched case-control studies
Minimum/Maximum detectable odds-ratio in case-control studies
Optional Continuity Correction
Sample size of non-inferiority trials with comparison of means and percentages

Sampsize Home Page

Why sampsize?

Files

Features

Project page