SUMMARY: A study is made of the effect of experimental errors of microbiological tests upon similarity values used in numerical taxonomy. For the coefficient approximations are given for the mean and standard error of similarity values when some test results are erroneous, if the true value of similarity (), the average probability in 1, 0 data of erroneous tests results (), and the number of tests () are known. Formulae are given for estimating from an analysis of variance, and for including or excluding chosen factors that affect test reproducibility.

When is over about 10% the error of similarity values becomes unacceptably large. Test error also increases the existing sampling error. It is generally better to employ many tests even if they are not as reproducible as desired, rather than to use only a few extremely reproducible tests.

Examples are given suggesting that within one laboratory can usually be kept below 5%, but the discrepancies between laboratories may often be much larger. It is advisable to replicate some of the strains in numerical taxonomic studies, so as to afford a check on test reproducibility.


Article metrics loading...

Loading full text...

Full text loading...


Most cited this month Most Cited RSS feed

This is a required field
Please enter a valid email address
Approval was a Success
Invalid data
An Error Occurred
Approval was partially successful, following selected items could not be processed due to error