Daniel McFadden and Kenneth Train, "Mixed MNL Models for Discrete Response", Journal of Applied Econometrics, Vol. 15, No. 5, 2000, pp. 447-470. The file xmat.txt contains the data. The file dotable2.g is a Gauss command file that takes the data in xmat.txt and creates the parameter estimates given in Table 2 of the paper. Both files, which are in DOS format, are zipped in the file mt-data.zip. Details of these two files are given below. XMAT.TXT Xmat.txt is an ascii file consisting of 181,506 lines of data with 4 numbers in each line. Each choice experiment is an observation, and there are 39 lines of data for each choice experiment. The file therefore contains data on 4654 observations (4654 = 181,506 / 39) and 156 pieces of information for each observation (where 156 = 39 * 4). This file with 181,506 lines and 4 numbers per line should be read, or reconfigured, into a matrix of dimension 4654x156, such that each row contains all the information about one choice experiment. In each experiment, the respondent was presented with 6 alternative vehicles. The columns contain the attributes of the six vehicles and the identification of which vehicle the respondent chose. Each attribute consists of 6 columns, giving the value of the attribute for each of the six vehicles. The variables contained in each set of 6 columns are given below. Precise definitions of the variables are given in Table 1 of the paper. Variable Columns: 1 1-6 Price divided by ln(income) 2 7-12 Range 3 13-18 Acceleration 4 19-24 Top speed 5 25-30 Pollution 6 31-36 Size 7 37-42 "Big enough" 8 43-48 Luggage space 9 49-54 Operating cost 10 55-60 Station availability 11 61-66 Sports utility vehicle 12 67-72 Sports car 13 73-78 Station wagon 14 79-84 Truck 15 85-90 Van 16 91-96 Constant for EV 17 97-102 Commute < 5 x EV 18 103-108 College x EV 19 109-114 Constant for CNG 20 115-120 Constant for methanol 21 121-126 College x methanol 22 127-132 These data are not used. 23 133-138 Dependent variable. 1 for chosen vehicle, 0 otherwise 24 139-144 Non-EV dummy 25 145-150 Non-CNG dummy 26 151-156 These data are not used. DOTABLE2.G This text file contains the Gauss code that reads the data from xmat.txt and calculates the mixed logit parameter estimates given in Table 2 of the paper. A reader who has Gauss installed, along with the Gauss maxlik routine, can run the program to reproduce the results. The only modification that is needed is to change the file paths appropriately in specifying the output file (output file=path/filename) and in loading the input data (load[4564,(6*26)]=path/xmat.txt.) This Gauss code is the cross-section version of the mixed logit estimation routines contained at K. Train's website: http://elsa.berkeley.edu/~train, with the model and data specified as needed for the model in Table 2. The website contains a manual that describes how to modify the code for other specifications and other data.