JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.. Biometrika Trust is collaborating with JSTOR to digitize, preserve and extend access to Biometrika.
SUMMARYIn multivariate calibration the relationship between a q-variate response vector Y and p explanatory variables X are estimated from training data in order to predict an unknown X, denoted by e, from further observed responses. When q > p both the profile likelihood for e and Bayesian inference for e depend on a prediction inconsistency diagnostic which highlights those response vectors used for prediction which are internally inconsistent in the prediction of e. When several further response vectors together display systematic anomalies one is led to questioning the estimated model. The information in prediction data about changes in parameters is investigated under various assumptions. The results indicate systematic anomalies in prediction data may be detected in a variety of ways, but corrected only under strong assumptions so that recalibration might be needed.