The Design of Experiments

by Ronald A. Fisher

Chapters

1. Introduction
2. The principles of experimentation, illustrated by a psycho-physical experiment
3. A historical experiment on growth rate
4. An agricultural experiment in randomised blocks
5. The latin square
6. The factorial design in experimentation
7. Confounding
8. Special cases of partial confounding
9. The increase of precision by concomitant measurements. Statistical Control
10. The generalisation of null hypotheses. Fiducial probability
11. The measurement of amount of information in general

Genres: Science, Stem, Mathematics, Nonfiction

Hardcover

First published January 1, 1935

About the author

Sir Ronald Aylmer Fisher was an English statistician, evolutionary biologist, geneticist, and eugenicist.

Among other things, Fisher is well known for his contributions to statistics by creating ANOVA (analysis of variance), Fisher's exact test and Fisher's equation.

Reviews

[loafingcactus · Rating 4 out of 5]

Charmingly written in the old gentleman's style, which is of especial note since this book marks the delineation between the physician as a gentleman relying on his superior intuition and sense and the physician as a technitian creating inputs and enacting outputs from scientific method.

[Chrissy · Rating 3 out of 5]

It's difficult to treat this statistical hallmark fairly from a modern perspective, and particularly from the perspective of a social scientist. Fisher's work, while undeniably fundamental to current statistical techniques in psychology, lay firmly in the applied realms of genetics and agriculture. Whereas it is possible to read his treatment of Latin Squares in plots of land, for example, as generalizable to the design of factorial behavioural experiments, holy hell is it tedious: it demands not only a translation of antiquated scientific prose to modern equivalents but also a translation of agricultural terminology to that of participant groups. His approach is, moreover, deeply mathematical. I skipped over much of the final chapter which, near as I could tell, dealt with algebraic proofs of the sufficiency and biases of parameter estimates-- topics that I had a hard time following in more accessible terms in my courses, lacking as I do a strong mathematical background.
Having both taken and taught the topics covered in the book, I know that most students in my field have a difficult time with this dry approach to statistics, preferring instead to have the concepts bootstrapped to their research programs. Nowadays, very few experimentalists are statisticians, and vice versa.

Nevertheless, the book certainly has value as an alternative perspective for those already familiar with the techniques. Fisher illustrates the fundamentals of the exact z test, single sample and paired t tests, confidence intervals, factorial tests, chi-square tests, and others. The way he explores them is vastly different from how I learned them, but the comparison was a worthwhile exercise. My existing knowledge allowed me to better understand his approaches to the same concepts, and to expand my understanding of their theoretical foundations. Old tricks, new angles. And a few fun jabs at Neyman and Peason, to boot!

I would not recommend this book to anyone without a strong grasp of the Fisherian tradition of data analysis. Those few of you well-equipped, on the other hand, may find it worth your while.