Analysis of Two - Part Random Effects Model for Semi-Ordinal Longitudinal Response

We have proposed a new variable called semi-ordinal variable. This variable exhibits a combination of responses, with a portion of them taking on ordinal values and the remaining values following a continuous distribution, often with truncation. In longitudinal analyses, variable of this type may be described by a pair of regression models, for example, one approach to analyze this variable is to use ordinal regression for the ordinal component and a conditional linear model for the continuous component. We have proposed a two - part ranom effects model for longitudinal semi-ordinal responses. A full likelihood-based approach that allows yielding maximum likelihood estimates of the model parameters is used. To illustrate the utility of the proposed model, we apply a new method for analysis outlier data. Also, the model is applied to using data from a large data set excerpted from the British Household Panel Survey (BHPS) is analyzed.