I estimated bacterial density in dilution series, using the
computerprogram by M.A. Hurley & M.E.Roscoe (journal of applied
bacteriology 1983, vol 55,p 159-164)
For comparing 2 MPN estimates they conduct a t-test referring to an
article by Cochran, W.G.(Estimation of Bacterial Densities by Means of
the 'Most Probable Number'. 1950. Biometrics 6: 105-116).
The nominator for the t-value is Standard Error S.E. calculated by a very
simple way with dilution factor and number of samples per dilution beeing
the only variable > constant S.E. However, the S.E. given by the
computerprogramme by Hurley and Roscoe is not a constant eventhoug my
sample number and dilution factor are the same??.
SO HOW HAS THIS S.E. ARISEN???????
Furthermore according to the article - for estimatiion of the 95%
confidence limits they use the value 1.96 from the normal distribution??
Shouldn't they use the 95% value from the t distibution for the same
reason they use a t-test???
Please reply if you posses any information that will add to my poor
understanding of statistics
Thank You very much
Annette Jensen e-mail: anj at pop.bio.aau.dk