sampsizecalc

Simple sample size calculator

Function file: sampsizecalc
N = sampsizecalc (TESTTYPE, EFFSZ)
N = sampsizecalc (TESTTYPE, EFFSZ, POW)
N = sampsizecalc (TESTTYPE, EFFSZ, POW, ALPHA)
N = sampsizecalc (TESTTYPE, EFFSZ, POW, ALPHA, TAILS)

Perform sample size calculations by power analysis. 

N = sampsizecalc (TESTTYPE, EFFSZ) returns the required sample size to 
reach the significance level (alpha) of 0.05 in a two-tailed version of 
the test specified in TESTTYPE for the specified effect size, EFFSZ, with 
a power of 0.8 (i.e. a type II error rate of 1 - 0.8 = 0.2)

TESTTYPE can be:

't2' (default) : two-sample unpaired t-test

't' : paired t-test or one-sample t-test

'z2' (default) : two-sample unpaired z-test (Normal approximation)

'z' : paired z-test or one-sample z-test (Normal approximation)

'r' : significance test for correlation

EFFSZ can be numeric value corresponding to the standardized effect
size: Cohen's d or h (when TESTTYPE is 't2', 't', 'z' or 'z'), or 
Pearson's correlation coefficient (when TESTTYPE is 'r'). For
convenience, EFFSZ can also be one of the following strings:

'small' : which is 0.2 for Cohen's d (or h), or 0.1 for Pearson's r.

'medium' : which is 0.5 for Cohen's d (or h), or 0.3 for Pearson's r.

'large' : which is 0.8 for Cohen's d (or h), or 0.5 for Pearson's r.

N = sampsizecalc (TESTTYPE, EFFSZ, POW) also
sets the desired power of the test. The power corresponds to 1 - beta, where
beta is the type II error rate (i.e. the probability of not rejecting the
null hypothesis when it is actually false). (Default is 0.8)


N = sampsizecalc (TESTTYPE, EFFSZ, POW, ALPHA)
also sets the desired significance level, ALPHA, of the test. ALPHA
corresponds to the type I error rate (i.e. the probability of rejecting the
null hypothesis when it is actually true). (Default is 0.05)

HINT: If the test is expected to be among a family of tests, divide ALPHA
by the number of tests so that the sample size calculations will maintain the
desired power after correction for multiple comparisons.

N = sampsizecalc (TESTTYPE, EFFSZ, POW, ALPHA, TAILS)
also sets whether the test is one-sided or two-sided (Default is 2)

Author: Andrew Penn