Dear Statalist,
I want to replicate the results of the Monte Carlo experiments reported in Chang and Lusk (2011) (Chang and Lusk. (2011) Mixed logit models: Accuracy and software choice j. Appl. Econ. 26: 167–172) and I want to generate data with characteristics described in section 4.1 of the paper. In addition to the instruction given in section 4.1, I want to add the following conditions.
(a) I want to assume that the parameters estimates (Item1-Item5 from the attached dataset) are correlated. For this, I believe, I need to specify variance covariance matrix that tells Stata how this variables are correlated.
(b) I want to assume the Price and Item1-Item5 (again referring to the attached dataset) variables are drawn from log normal, Johnson’s Sb, gamma, and triangular distributions.
(c) Price and Item1-Item 5 variables are drawn from more flexible distributions with logit function which shapes is represented by polynomials, splines, steps functions as described by Train’s 2016 paper (Train, K. (2016) Mixed logit with a flexible mixing distribution. The Journal of Choice Modelling 19 (40–53).
I have attached the dataset used by Chang and Lusk (2011). However, I don’t have access to the do file that generate the dataset and I’m not sure how to implement my conditions (a)-(c).
Thank you for any help you can.
Kind Regards
0 Response to How to generate dataset to test different mixed logit specifications using Monte Carlo experiment
Post a Comment