Port by Andrew Leifer [email protected] 24 January 2018
In their paper: Krishnamoorthy, K and Thomson, J. (2004) A more powerful test for comparing two Poisson means. Journal of Statistical Planning and Inference, 119, 249-267]
Krishnamoorthy and Thomson describe an improved test to determine if two Poissonion means are significantly different, now referred to as the "E-test." Krishnamoorthy released a fortran implementation here:
Original Code http://www.ucs.louisiana.edu/~kxk4695/ and http://www.ucs.louisiana.edu/~kxk4695/statcalc/pois2pval.for
Here is a quick and dirty port of their fortran code to MATLAB. Note I kept all of the original fortran structure and variable names and style.
The matlab function testPoissonSignificance() is described below:
function pvalue=testPoissonSignificance(k1,k2,n1,n2,d,iside)
% function pvalue=testPoissonSignificance(k1,k2,n1,n2,d,iside)
%
% k1, k2 = sample counts (must be integer)
% n1, n2 = sample size (must be integer)
% d = value of mean1-mean2 under H0 (default is zero)
% iside = 1 for right tail-test or 2 for two-tail test (default)
To use the example from Przyborowski and Wilenski (1940): a factory spec's their supply of clover seeds to have no dodder seed contaminates in a 100g sample. Upon receiving shipment, the purchaser takes a 100g sample and finds 3 dodder seeds. The purchaser wishes to determine if the difference could not be due to chance. k1=0, k2=3, n1=n2=1, d=0, and iside=2 for the two tail test.
>> pvalue=testPoissonSignificance(0,3,1,1,0,2)
pvalue =
0.0884