Open Conference Systems, CLADAG2023

Font Size: 
Markov switching regression models for longitudinal ordered responses to handle attitude towards response style, unobserved heterogeneity and serial dependence
Sabrina Giordano

Last modified: 2023-06-13


When asked to assess their opinion about attitudes or perceptions on Likert-scale, respondents often endorse the midpoint or extremes of the scale and  agree or disagree regardless of the content. These responding behaviors are known in the psychometric literature as response styles and they generally affect the accuracy of the estimates. The novelty of the proposed approach, in the context of longitudinal ordered categorical data, is in considering simultaneously the temporal dynamics of the responses (observable ordinal variables) and unobservable answering behaviors, possibly influenced by response styles, through a  Markov switching logit model  with two latent components. One component accommodates serial dependence and respondent's  unobserved heterogeneity, the other component determines the responding attitude (due to RS or no-RS). The dependence of the responses on covariates is modelled by a stereotype logit model with parameters varying according to  the two latent components.  A new interpretation of the parameters of the stereotype model is given by defining the allocation sets as intervals of values of the linear predictor that identify the most probable response.  Unobserved heterogeneity, serial dependence and tendency to response style are modelled through our approach on  longitudinal data, collected by the Bank of Italy.