statisticalrethinkin..

382 ENDNOTES12. For an autopsy of the experiment, see http://profmattstrassler.com/articles-and-posts/particle-physics-basics/neutrinos/nefaster-than-light/opera-what-went-wrong/. [21]13. See Mulkay and Gilbert 1981 for many examples of “Popperism” from practicing scientists, including famousones. [21]14. For an accessible history of some measurement issues in the development of physics and biology, includingearly experiments on relativity and abiogenesis, I recommend Collins and Pinch 1998. Some scientists have readthis book as an attack on science. However, as the authors clarify in the second edition, this was not their intention.Science makes myths, like all cultures do. at doesn’t necessarily imply that science does not work. Seealso Daston and Galison (2007), which tours concepts of objective measurement, spanning several centuries.[21]15. e first chapter of Sober (2008) contains a similar discussion of modus tollens. Note that the statistical philosophyof Sober’s book is quite different from that of the book you are holding. In particular, Sober is weaklyanti-Bayesian. is is important, because it emphasizes that rejecting modus tollens as a model of statistical inferencehas nothing to do with any debates about Bayesian versus non-Bayesian tools. [21]16. Popper himself had to deal with this kind of theory, because the rise of quantum mechanics in his lifetimepresented rather serious challenges to the notion that measurement was unproblematic. See chapter 9 in hisLogic of Scientific Discovery, for example. [21]17. See the Aerward to the 2nd edition of Collins and Pinch (1998) for examples of textbooks getting it wrongby presenting tidy fables about the definitiveness of evidence. [22]18. A great deal has been written about the sociology of science and the interface of science and public interest.Interested novices might begin with Kitcher (2011), Science in a Democratic Society, which has a very broad topicalscope and so can serve as an introduction to many dilemmas. [22]19. Yes, even procedures that claim to be free of assumptions do have assumptions and are a kind of model. Allsystems of formal representation, including numbers, do not directly reference reality. For example, there ismore than one way to construct “real” numbers in mathematics, and there are important consequences in someapplications. In application, all formal systems are like models. See http://plato.stanford.edu/entries/philosophymathematics/for a short overview of some different stances that can be sustained towards reasoning in mathematicalsystems. [22]20. Saint Augustine, in City of God, famously admonished against trusting in luck, as personified by Fortuna:“How, therefore, is she good, who without discernment comes to both the good and to the bad?” See also theintroduction to Gigerenzer et al. (1990). Rao (1997) presents a page from an old book of random numbers,commenting upon how seemingly useless such a thing would have been in previous eras. [22]21. Most scholars trace frequentism to British logician John Venn (1834–1923), as for example presented in his1876 book. Speaking of the proportion of male births in all births, Venn said, “probability is nothing but thatproportion” (page 84). Venn taught Fisher some of his maths, so this may be where Fisher got his dogmaticopposition to Bayesian probability. [23]22. Fisher (1956). See also Fisher (1955), the first major section of which discusses the same point. [23]23. is last sentence is a rephrasing from Lindley (1971): “A statistician faced with some data oen embedsit in a family of possible data that is just as much a product of his fantasy as is a prior distribution”. Dennis V.Lindley (1923–2013) was a prominent defender of Bayesian data analysis when it had very few defenders. [23]24. It’s hard to find an accessible introduction to image analysis, because it’s a very computational subject. Atthe intermediate level, see Marin and Robert (2007), chapter 8. You can hum over their mathematics and stillacquaint yourself with the different goals and procedures. See also Jaynes (1984) for spirited comments on thehistory of Bayesian image analysis and his pessimistic assessment of non-Bayesian approaches. [24]25. See Gigerenzer et al. 2004. [25]