Distributional assumptions in mixed logit models: implications for welfare estimates in South China tiger conservation efforts

The mixed logit model is widely used in choice experiments to capture random variations in individual tastes; however, the predefined distributional forms significantly affect welfare estimates. This study addresses the lack of research on the effects of four common distributions (normal, log-normal, triangular and uniform) on welfare estimates, attribute significance and model fit, particularly in the context of endangered species conservation. The results highlight visitors’ willingness to contribute financially to the South China tiger conservation, while also emphasizing the importance of identifying the optimal distribution in advance to avoid potentially biased welfare estimates and subsequent erroneous policy recommendations. Notably, except for the log-normal model, the other distributions yielded consistent estimates regarding attribute significance and preference. Interestingly, while the log-normal model fitted well, it yielded fewer significant attributes and lower welfare estimates. The triangular distribution performed best in terms of log-likelihood, pseudo-R2 and Akaike information criterion. This finding challenges the common use of normal distributions for mixed logit models and addresses the importance of developing mixed logit models with different distributions. Moreover, effective budget allocations based on visitors’ preferences for South China tiger conservation are recommended for policymakers.

Text 121035.pdf - Published Version Restricted to Repository staff only Available under License Creative Commons Attribution. Download (1MB) Official URL or Download Paper: https://www.hrpub.org/journals/article_info.php?ai...

Actions (login required)

View Item