Dear all,
I would like to perform Vuong's (1989) test for model selection to see if the difference in log-likelihoods produced by a zero-truncated normal hurdle model and a log-normal hurdle model are statistically significant, as described in Jeffrey Wooldridge's Econometric Analysis of Cross-section and Panel data (2010), page 701. For this, I need to compute log-likelihood values for each observation in the respective model. How can I do this in Stata? I have searched extensively in Stata manuals, other online sources, and textbooks but could not find an answer. The regression commands I use are churdle linear and churdle exponential for the truncated normal hurdle and the log-normal hurdle model respectively.
I apologize for any formal mistakes I made in asking this question, this is my first post. Please feel free to suggest alterations or require additional information.
Thank you and kind regards,
Paul Witte
I would like to perform Vuong's (1989) test for model selection to see if the difference in log-likelihoods produced by a zero-truncated normal hurdle model and a log-normal hurdle model are statistically significant, as described in Jeffrey Wooldridge's Econometric Analysis of Cross-section and Panel data (2010), page 701. For this, I need to compute log-likelihood values for each observation in the respective model. How can I do this in Stata? I have searched extensively in Stata manuals, other online sources, and textbooks but could not find an answer. The regression commands I use are churdle linear and churdle exponential for the truncated normal hurdle and the log-normal hurdle model respectively.
I apologize for any formal mistakes I made in asking this question, this is my first post. Please feel free to suggest alterations or require additional information.
Thank you and kind regards,
Paul Witte
