Документ взят из кэша поисковой машины. Адрес оригинального документа : http://hea-www.harvard.edu/astrostat/HEAD2008/haiku_tloredo.pdf Дата изменения: Tue Apr 1 07:28:13 2008 Дата индексирования: Tue Oct 2 03:43:57 2012 Кодировка:

Mo deling GRB (and other) p opulations: Some lessons from multilevel mo deling
Tom Loredo and Ira Wasserman, Cornell U.
Goal: Infer properties of a population from measurements of observables with measurement error and selection effects

Multilevel models:

· · ·

Upper level: Draw observables from (adjusted) pop'n dist'n Lower level: Draw measurements from independent data distributions Bayes: Account for measurement error via marginalization marginal likelihood, averages over uncertainties

Ignoring uncertainties corrupts inferences: