Elsevier, 2013, ISBN 9780124169715
reviewed by Dave Koch and Emily Ryan (Emory University, Atlanta, GA)
The word “statistics” evokes a variety of reactions. Most of them, in my (DK) experience, have been negative. While the majority of us involved in science, chemistry, and particularly clinical work realize that statistics are valuable tools, we struggle to utilize these resources properly and with confidence. Perhaps we don’t use statistics as often as we should, or employ the right statistic for the application. The availability of spreadsheet programs containing built-in statistical calculations has, ironically, exacerbated this problem. Availability produces deployment, but not necessarily understanding. Use of statistics has increased in the past 20 years, but so has the lingering doubt that the answers produced are correct. It comes down to the question of are the right statistics being applied at the right time, and in the correct fashion. A wealth of information about statistical formulas and their application is available from numerous sources today, but who has time to sort through all of these sources to locate the explanation that is needed at the right moment?
Into this quandary steps this nifty handbook by Dr. Anders Kallner. Dr. Kallner is a chemist from the Karolinska Institute in Sweden who has enjoyed a productive career in clinical chemistry. He has taken active roles in many international committees, including several CLSI evaluation protocols and other efforts in metrology. Thus, he is well qualified to write a handbook on the use of laboratory statistics. Two sentences in his introduction summarize his objective better than we could describe: “This present ‘compendium’ is for those who like me are engaged in practical laboratory work and do not have a major in statistical analysis and feel somewhat uncomfortable with the statistical jargon….An idea with this compendium is to have most of the statistical procedures used in the laboratory collected in one source and described in a standardized but not compressed format.” What we can add is to report in our review that Dr. Kallner has largely succeeded in his goal of producing a convenient, usable resource about laboratory statistics.
In fact, this book would replace all of our searching for “How exactly is that statistic calculated?” Laboratory Statistics includes virtually every basic statistic a clinical scientist should need to comprehend in order to deal with most data sets, from exploring characteristics of the data to various simple linear regressions. While the handbook is not for the uninformed, the explanations are succinct and to the point. For nearly every statistic described, Dr. Kallner adds practical, real-world situations to illustrate how they are applied. If you have a reasonable grasp on the theory and methods of statistics already, this book will allow you to perform robust statistical analysis with a simple spreadsheet.
A couple of things we are disappointed about is that the book didn’t cover any multivariate analyses (two-way ANOVA, linear regression with interaction terms), but it covers all the two-variable statistics you would need no matter how non-Gaussian your data is. Also, while the book includes a number of useful figures, most of these figures are too small. As we reviewed them, we kept wishing we could “zoom in” on some of these, but of course we were holding the handbook, not reviewing it in electronic fashion.
Finally, there is a short “bonus” section at the end on metrology where the reader will find a list of definitions of terms about accuracy and other concepts relevant to the correct application of statistics. Dr. Kallner appropriately understands that users of laboratory statistics should also agree on our definitions of metrological concepts so we properly apply statistics in our work.