Last Words on Anova?

Chapter
Part of the Selected Works in Probability and Statistics book series (SWPS)

Abstract

Many people like to say the last words in an academic debate, and I am no exception. I have tried to do this on a few occasions, only to discover that when I came to say my piece, everyone had left the room. The analysis of variance is a case in point, and my comments on Tukey’s contributions to anova explain the problem. If – as I believe to be true – people don’t care much these days what Tukey thought about anova, they are going to care even less what I think. This is not said with any sense of bitterness. Indeed I regard myself as something of a student of fads, fashions and trends in statistics, so why should I expect otherwise? Nevertheless, I’m very happy to see these articles reprinted, as their easier availability may kindle the interest of someone, somewhere, sometime in what I still believe to be an important part of (the history of) our subject.

Keywords

Covariance Alan 

Many people like to say the last words in an academic debate, and I am no exception. I have tried to do this on a few occasions, only to discover that when I came to say my piece, everyone had left the room. The analysis of variance is a case in point, and my comments on Tukey’s contributions to anova explain the problem. If – as I believe to be true – people don’t care much these days what Tukey thought about anova, they are going to care even less what I think. This is not said with any sense of bitterness. Indeed I regard myself as something of a student of fads, fashions and trends in statistics, so why should I expect otherwise? Nevertheless, I’m very happy to see these articles reprinted, as their easier availability may kindle the interest of someone, somewhere, sometime in what I still believe to be an important part of (the history of) our subject.

My main stimulus for work in this area came from the papers of six people: R.A. Fisher, Frank Yates, and John A. Nelder from the U.K, indeed all from Rothamsted, Alan T. James and Graham Wilkinson from Adelaide, Australia, and John W. Tukey from the U.S.A. Unpublished lecture notes by James were extremely helpful in getting me going. The anova program within GENSTAT, initially created by Wilkinson based on research by James, Wilkinson, James & Wilkinson and Nelder, was enormously influential. It was (and remains) truly brilliant in conception and execution, and I wanted to understand it. For a long time I was interested in – one might say obsessed with – the symmetries underlying much of anova, and that is reflected in some of the papers reprinted (thank you Rosemary Bailey)! But also I wanted to understand how users of anova saw things, including gory details such as the combination of information, the analysis of covariance and dealing with missing values, all topics with wonderful histories. I made one attempt to put it all together for general consumption, but that got rejected, and so I moved on to other things. As explained above, it is not clear how many people now care. I hope you enjoy the papers. There are several more if you do.
Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window Open image in new window

References

  1. 1.
    R. A. Bailey and T. P. Speed. On a class of association schemes derived from lattices of equivalence relations. In P. Schultz, C. E. Praeger, and R. Sullivan, editors, Algebraic Structures and their Applications. Marcel Dekker, New York, 1982.Google Scholar
  2. 2.
    R. A. Bailey and T. P. Speed. A note on rectangular lattices. Ann. Stat., 14:874–895, 1986.MathSciNetMATHCrossRefGoogle Scholar
  3. 3.
    T. P. Speed and R. A. Bailey. Factorial dispersion models. Int. Stat. Rev., 55:261–277, 1987.MathSciNetMATHCrossRefGoogle Scholar
  4. 4.
    R. A. Bailey, C. E. Praeger, C. A. Rowley, and T. P. Speed. Generalized wreath products of permutation groups. Proc. Lond. Math. Soc. (3), 47: 69–82, 1983.Google Scholar
  5. 5.
    A. Houtman and T. P. Speed. Balance in designed experiments with orthogonal block structure. Ann. Stat., 11:1069–1085, 1983.MathSciNetMATHGoogle Scholar
  6. 6.
    A. Houtman and T. P. Speed. The analysis of multi-stratum designed experiments with incomplete data. Aust. J. Stat., 26:227–246, 1984.MathSciNetMATHCrossRefGoogle Scholar
  7. 7.
    T. P. Speed. John W. Tukey’s contributions to analysis of variance. Ann. Stat., 30(6):1649–1665, 2002.Google Scholar
  8. 8.
    T. P. Speed. ANOVA models with random effects: An approach via symmetry. In J. Gani and M. B. Priestley, editors, Essays in Time Series and Allied Processes : Papers in Honour of E. J. Hannan, pages 355–368. Applied Probability Trust, Sheffield, UK, 1986.Google Scholar
  9. 9.
    T. P. Speed. What is an analysis of variance? Ann. Stat., 15:885–910, 1987.MathSciNetMATHCrossRefGoogle Scholar

Copyright information

© Springer Science+Business Media, LLC 2012

Authors and Affiliations

  1. 1.Department of StatisticsUniversity of CaliforniaBerkeleyUSA
  2. 2.Division of BioinformaticsWalter and Eliza Hall InstituteParkvilleAustralia

Personalised recommendations