In model 1 of Example 7.5, try a scale mixture version for the random effects \(b_{i, 1: 2}\) with degrees of freedom an unknown (equivalent to a bivariate Student \(t\) ). The analysis may be sensitive to the prior adopted for the degrees of freedom (e.g. a uniform between 4 and 100, an exponential with unknown mean, etc.). Are there any clear outliers (scale factors clearly under one)?

Data from Example 7.5

Transcribed Image Text:

To illustrate longitudinal approaches for ordinal responses, consider data on the impacts of drug treatments on symptom severity among n = 324 schizophrenic patients, collected as part of the NIMH Schizophrenic Collaborative Study (Hedecker and Gibbons, 1994). There were originally four treatments: chloropromazine, fluphenazine, thioridazine, and placebo. Since previous analysis revealed similar effects of the three anti-psychotic drugs, the treatment is