Statistics and Chemometrics for Analytical Chemistry

A recent visit to one of BP’s European technology centers produced a request to develop statistics training for the people in the analytical lab. Of course, I first searched to see what books were available. I have always liked the book by Caulcott and Boddy (1983), but it is rather dated. The most recent review in Technometrics of a similar book was that for Meloun et al. (1992) (see LaLonde 1995). There also was a review of the book by Meier and Zünd (1993) (see Hillyer 1995). A second edition of that book was published in 2000. It is a large book, nearly 500 pages, with a hefty price, $126.00 from Amazon.com. This more modest-sized and lower-priced book actually would be a reasonable starter book for a typical analytical chemist. This book was first published in 1984, but it has had suitable updating. This edition uses Excel and MINITAB for statistical computing. It also is the first of the editions to incorporate chemometrics topics. The book now has had a new final chapter added to introduce the multivariate methods of chemometrics. The 11 different tools that are presented run the gamut from principal components and cluster analyses to partial least squares regression and neural networks. The topics are all discussed without any use of matrix algebra and little use of equations. As for the other seven chapters, they start with the basics of “Statistics for Repeated Measurements” in Chapter 2 and progress through significance tests, measurement quality, calibrations using regression methods, nonparametric methods, robust methods, experimental design, and optimization. There are excellent discussions of a number of important indigenous topics relevant to the analytical laboratory, including lognormal distributions, propagation of errors, components of variability, collaborative trials, limits of detection, comparisons from calibration curves, standard additions, and cusum charts. The authors, apparently a husband and wife pairing of an analytical chemist and a statistician, have an excellent understanding of the technical level appropriate for the people in the lab. There are relatively few equations but a considerable number of plots and displays of results. Hopefully there will soon be a report for the second edition of the text of Meier and Zünd (1993), but it is hard to imagine a much better book than this one for use as the first course with a group of analytical chemists. There are excellent sets of references, both from the statistics and analytical chemistry literature, that provide sources for further study by anyone so inclined.