Other >> EducationSubtle Artistry in Cogent Statistical Inferencesby Tyler Hicks
Submitted : Fall 2013 The purpose of this paper is to engage in a point of contention between Bayesian and Classical statisticians about the role of artistry in statistical inference. Artistry is a kind of human creativity that goes beyond formal algorithms. This paper can best be read as a long argument for the thesis that both Bayesians and Classical statisticians must each, in their own way, resort to artistry when making statistical inferences. Bayesians must artfully construct priors to make inferences; Classical statisticians must artfully select sampling distributions when making inferences. It is well known that prior selection is a Bayesian art rather than an exact science. This paper, however, mathematically documents why selection of sampling distributions in applied research settings can also be seen as requiring a little bit of artistic imagination and luck – a reality not well known among researchers in social science and applied disciplines. Related Links:
|