TESTING FOR OUTLIERS BASED ON BAYES RULE
Giampaolo E. D’Errico · CINECA IRIS Institutional Research Information System (IRIS Istituto Nazionale di Ricerca Metrologica) · 2009
A Bayesian treatment of conjectured outlying observations is developed, using the computational device of inverse probability. The device’s performance is discussed in term of posterior probability of missed or false detections. The key role of prior probability is shown through a numerical example.