@ R2ABS1.CMD -- Random A-B probit model with two random
parameters: catch and FCA, and deterministic multiplicative
differentials for each of the parameters for fish species or
different FCA levels. Simulation is used for approximation.
Uses 1,000 draws rather than 500. Sept. 28, 1999 @
@ Standard deviations of catch parameters vary in proportion
to their means @
new;
library maxlik;
clear xdif;
output file = R2ABS1.OUT reset;
"R2ABS1.CMD. This is the weighted random A-B probit model with two random
variables, the catch and FCA parameter. For different species and FCA levels,
a deterministic multiplicative component is used. Therefore,
standard deviations of catch parameters vary in proportion
to their means. Simulation is used to approximate the joint probability,
using 1,000 draws rather than 500. Sept. 28, 1999";
/* Model Selection: choice of explanatory variables */
let kvec = 1 2 3 4 5 6 7 8 9 10 11 12 13;
/* 1-4 are the catch rates for four species: perch, trout/salmon, walleye, bass;
5-12 are the FCA dummies for categories 2-9; 13 is the launch
fee */
/* Choose the starting values */
@ if strtflag eq 0; start with adjusted NRAB parameters @
@ if strtflag eq 1; start with nonrandom probit @
strtflag = 1;
@ Choose the number of draws for the simulation @
draws=1000;
/* Z is an n x 9 matrix. Its rows are the individuals.
The first eight columns are the results of the paired comparisons (1 or 2).
If the respondent did not choose, then a 0 is entered.
The ninth row contains the version number (1-10). Z is used to form
Xdif, or the differences in characteristics. */
loadm Z[647, 9] = Z.dat;
n = rows(Z);
Xdif = zeros(8*n,rows(kvec));
/* X is a 160 x 13 matrix of characteristics of the choices. Its rows are
arranged by version (1-10), then choice pair (1-8), then choice (A or B).
Its columns are the four fish catch rates; the FCA dummies (eight
dummies for categories 2-9); and the launch fee.
There are 16 lines for each version. The first two lines of the 16 are
pair 1 alternative A and pair 1 alternative B; the next two lines are
for pair 2, the next two are for pair 3, and so on... */
loadm X[160, 13] = X.dat;
"
Estimation uses X.dat.";
/* The following loop creates the matrix Xdif, dimension 8n x 13,
** where n is the number of individuals, and there are 13 explanatory
** variables in the model. */
ll=0; i=0;
do while i