Michael P. Keane and Nada Wasi, "Comparing Alternative Models of Heterogeneity in Consumer Choice Behavior," Journal of Applied Econometrics, Vol. 28, No. 6, 2013, pp. 1018-1045. This paper estimates several consumer choice models on 10 stated choice experiment datasets. There are two zip files. The file Data_KW2012.zip (2.00 MB compressed; 69.5 MB uncompressed) contains 10 datasets. The file Program_KW2012.zip (396 KB compressed; 640 KB uncompressed) contains Matlab programs. All data files are stored in ASCII(.txt) files using the same format. Column 1: respondents' id Column 2: scenario or (choice occasion) number (unique within each respondent) Column 3: alternative number (unique within each choice occasion) Column 4: dependent variable (1 if the alternative in that row is chosen; 0 otherwise) Column 5-last column: choice attributes The number of total rows in each data file is the number of alternatives x the number of choice occasions per respondent x the number of respondents. Online Appendix A, Table A1 describes the general characteristics of each dataset (i.e., number of attributes, number of choices, number of choice occasions), while Table A2 gives details about the design of each experiment (i.e., how the attributes of the choice alternatives in column 5-last column are coded). The three datasets involving medical decision making (i.e., decisions about genetic and cervical cancer tests) were collected by Centre for Health Economics Research and Evaluation (CHERE), University of Technology Sydney. The other seven datasets were collected by Centre for the Study of Choice (CenSoC), University of Technology Sydney. Any researcher interested in using some or all of these datasets for publications should also ask CHERE/CenSoC for permission. CHERE email: mail@chere.uts.edu.au. CenSoC email: censoc@uts.edu.au. The file Program_KW2012.zip contains two separate folders. The gmnl2009b folder contains the programs to estimate scale heterogeneity, mixed logit and G-MNL models. A readme and demo about the programs are provided in that folder. The gmx2n2008b folder contains the programs to estimate latent class and mixture-of-normal mixed logit models.