Abstract
I particularly admire Terry for the breadth of his thinking, the depth of his understanding, the strength and openness with which he expresses his opinions, and the clarity of his writing. All of these qualities are on display in the manuscripts in this section.
You have full access to this open access chapter, Download chapter PDF
Similar content being viewed by others
Keywords
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.
I particularly admire Terry for the breadth of his thinking, the depth of his understanding, the strength and openness with which he expresses his opinions, and the clarity of his writing. All of these qualities are on display in the manuscripts in this section.
But before I comment on those articles, I want to first mention that the best illustrations of Terry’s admirable qualities are his regular commentaries for the IMS Bulletin, begun when he was President of the Institute of Mathematical Statistics and continued by popular demand. (To date, he has written over 70 such commentaries! They are available online at http://bulletin.imstat.org.) Among my favorites: In Praise of Postdocs [10], Keep Gender on the Agenda [8], A Toast to Posters! [11], Books Worth Reading [7] (his five favorite books are Feller Volume I [2], editions 1, 2 and 3, and Feller Volume II [3], editions 1 and 2), and It’s Job-hunting Time! [9]. Terry has strong opinions on a wide range of topics, and we all benefit from his willingness to share them.
Terry’s technical report on probabilities related to the reliability of nuclear reactors [14], commenting on probability statements in the Reactor Safety Study [4], demonstrates Terry’s willingness to delve deeply and thoroughly into a problem and the clarity and insight he then gives us. (The manuscript is especially relevant today, given the problems with nuclear reactors following the tsunami in Japan.) Misuse of the addition and multiplication rules of probability is well known to statisticians. His further point, that probability statements are meaningless without an understanding of how they were derived, is obvious in retrospect yet often overlooked. It is related to a point that Terry has repeatedly emphasized: an estimate is of no value without some statement on its uncertainty (such as a standard error). I can imagine no better (or more amusing) illustration of this issue than the statement from the Chairman of the U.S. National Transportation Safety Board regarding the chance of two jets colliding on the ground, and the response to Terry’s aerogramme. (The word “aerogramme” itself makes me smile.) Also amusing is Terry’s comment [14, p. 24], “The whole exercise can now be recognized as being totally without relevance to the real world.” Ouch!
Terry’s work on salmon populations [16] and bioassays [1] were among his last projects before he began to focus almost exclusively on genetics and genomics. The salmon work is a particularly good illustration of Terry’s approach: know the science and the scientific questions and goals, and let the methods follow. I saw this in action in the bioassay project; our (or really his) normal-Poisson mixture model, which now seems so natural, was the last of a long sequence of preliminary models. My most vivid memory of that effort was our initial error (or really my initial error) in implementing an EM algorithm for the normal-Poisson mixture. The sufficient statistics include not just ∑x i but also ∑x i 2, and so the E-step requires not just the expected values of the x i , but also of the x i 2. That one should follow the log likelihood across iterations, as a diagnostic for the correct implementation of an EM algorithm, was another important lesson.
Terry’s comment [12] on Robinson’s Statistical Science article on BLUP [5] is masterly. The diverse applications of BLUP that Terry provides add great emphasis to the importance of Robinson’s paper. Terry’s comment has been widely cited as being the first articulation of the connection between smoothing splines and BLUPs.
Finally, let me comment on the two encyclopedia entries reprinted in this section, on restricted maximum likelihood [15] and on iterative proportion fitting (IPF) [13], though I can think of little to say except that they are superb examples of the clarity of Terry’s writing (and to note that Terry has an IMS Bulletin commentary [6] on IPF, too!). The breadth and depth of Terry’s thinking often fills me with envy, but the effort he devotes to sharing his ideas allows me to get past the envy and simply appreciate his insights.
References
K. Broman, T. Speed, and M. Tigges. Estimation of antigen-responsive T cell frequencies in PBMC from human subjects. J. Immunol. Methods, 198:119–132, 1996.
W. Feller. An Introduction to Probability Theory and its Applications, volume I. Wiley, New York, 3rd edition, 1968.
W. Feller. An Introduction to Probability Theory and its Applications, volume II. Wiley, New York, 2nd edition, 1971.
N. C. Rasmussen et al. Reactor Safety Study: An assessment of accident risk in U.S. commercial nuclear power plants. Technical report, Nuclear Regulatory Commission, Washington, D.C. (USA), October 1975. Report Number(s): WASH-1400-MR; NUREG-75/014-MR.
G. K. Robinson. That BLUP is a good thing: The estimation of random effects. Stat. Sci., 6:15–51, 1991.
T. Speed. Terence’s stuff: My favourite algorithm. IMS Bulletin, 37 (9): 14, 2008.
T. Speed. Terence’s stuff: Books worth reading. IMS Bulletin, 34 (2):6, 2005.
T. Speed. President’s column: Keep gender on the agenda. IMS Bulletin, 33 (2):4, 2004.
T. Speed. Terence’s stuff: It’s job-hunting time! IMS Bulletin, 38 (1):14, 2009.
T. Speed. President’s column: In praise of postdocs. IMS Bulletin, 32 (6):4, 2003.
T. Speed. Terence’s stuff: A toast to posters! IMS Bulletin, 34 (6):6, 2005.
T. P. Speed. Comment on G. K. Robinson, “That BLUP is a good thing: The estimation of random effects”. Stat. Sci., 6(1):42–44, 1991.
T. P. Speed. Iterative proportional fitting. In P. Armitage and T. Colton, editors, Encyclopedia of Biostatistics. John Wiley & Sons, New York, 1999.
T. P. Speed. Negligible probabilities and nuclear reactor safety: Another misuse of probability? Technical report, Mathematics Department, University of Western Australia, 1977.
T. P. Speed. Restricted maximum likelihood (REML). In S. Kotz, editor, Encylopedia of Statistical Sciences, volume 2, pages 472–481. John Wiley & Sons, New York, 1997.
T. P. Speed. Modelling and managing a salmon population. In V. Barnett and K. T. Turkman, editors, Statistics for the Environment: Water Related Issues, volume 1, pages 267–292. John Wiley & Sons, 1993.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2012 Springer Science+Business Media, LLC
About this chapter
Cite this chapter
Broman, K.W. (2012). Applied Statistics and Exposition. In: Dudoit, S. (eds) Selected Works of Terry Speed. Selected Works in Probability and Statistics. Springer, New York, NY. https://doi.org/10.1007/978-1-4614-1347-9_8
Download citation
DOI: https://doi.org/10.1007/978-1-4614-1347-9_8
Published:
Publisher Name: Springer, New York, NY
Print ISBN: 978-1-4614-1346-2
Online ISBN: 978-1-4614-1347-9
eBook Packages: Mathematics and StatisticsMathematics and Statistics (R0)