Modified Poisson Regression Analysis of Grouped and Right-Censored Counts

Abstract Grouped and right-censored (GRC) counts are widely used in criminology, demography, epidemiology, marketing, sociology, psychology other related disciplines to study behavioural event frequencies, especially when sensitive research topics or individuals with possibly lower cognitive capacities at stake. Yet, the co-existence of grouping right-censoring poses major difficulties regression analysis. To implement generalised linear GRC counts, we derive modified Poisson estimators their asymptotic properties, develop a hybrid line search algorithm for parameter inference, demonstrate finite-sample performance these via simulation, evaluate its empirical applicability based on survey data drug use America. This method has clear methodological advantage over ordered logistic model analysing counts.