DETERMINANTS OF SMOKELESS TOBACCO PREVALENCE IN A DEVELOPING COUNTRY: A TRIVARIATE ORDERED PROBIT APPROACH

A variant of bivariate models, such as the hurdle or double-hurdle method, is often utilized in extant studies to examine the consumption of addictive goods. The zero observations in such models are attributed to either (a) non-participation or (b) non-consumption upon participation. However, non-participation in addictive goods itself consists of two separate processes: The decision not to start addictive goods use and the decision to quit after starting addictive goods use. Kasteridis et al. [1] highlighted that integrating information from such separate decisions can lead to efficiency gains in statistical analyses and proposed a trivariate model. This paper explores the determinants of smokeless tobacco prevalence in a developing country using the trivariate ordered probit model proposed by Kasteridis et al. [1]. Using data from Nepal, the statistical findings highlight that involvement in other health risk behaviors is positively correlated with the decision to start, the decision not to quit, and the frequency of smokeless tobacco use in Nepal. The findings can provide insight into the effectiveness of policy tools that aim to reduce tobacco prevalence. Received: May 13, 2024Revised: June 6, 2024Accepted: June 25, 2024