COMPSTAT 2022: Start Registration
View Submission - COMPSTAT2022
Title: Evaluating complex agency effects on status transitions: Challenges within a Latent Markov Model paradigm Authors:  Marco Doretti - University of Perugia (Italy) [presenting]
Giorgio eduardo Montanari - University of Perugia (Italy)
Francesco Bartolucci - University of Perugia (Italy)
Maria Francesca Marino - University of Florence (Italy)
Abstract: Among their various purposes, Latent Markov models (LMMs) can be useful tools to cluster and/or rank a set of agencies operating on different users, for which some categorical variables measuring an unobserved trait of interest are collected over time. To this end, extensions of the basic LMM have been proposed in order to incorporate agency effects either as fixed or random effects. The focus is on a specific setting where: i) agency evaluation involves effects on transition probabilities only, and ii) these effects have a complex structure that cannot be captured by a single component, sometimes due to the data collection mechanism in use. A suitable example is represented by the assessment of the performance of nursing homes with regard to their ability to avoid residents' health status worsening. Building upon the existing literature employing LMMs in this framework, some issues related to the construction of a proper performance measure (and of a measure of its variability) are analyzed.