Premium
The relevance of linear regression for assessing the performance of microbiological counting techniques
Author(s) -
Sacree Margaret,
Kilsby D.C.
Publication year - 1988
Publication title -
letters in applied microbiology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.698
H-Index - 110
eISSN - 1472-765X
pISSN - 0266-8254
DOI - 10.1111/j.1472-765x.1988.tb01239.x
Subject(s) - relevance (law) , library science , operations research , political science , computer science , engineering , law
The usefulness of an automated or rapid microbiological method to count bacteria is normally assessed by comparison of the method's results with those obtained using a standard plating technique to produce a linear relationship between the two measurements. If the correlation coefficient obtained for this regression is high, then the resulting regression line is often used as a calibration curve for counting. Such an approach can fail to identify imprecisions in the counting techniques especially when the calibration curve is derived over a large counting range. The problems encountered are illustrated by reference to results using a conductance method, and an alternative experimental approach towards assessing the performance of such techniques is proposed.