Table 4: Estimates of the binary probit model.

Variable Coefficient Std. error -statistic Marginal effects Constant −.5489 .8840 −.621 — GEN .3702 .3063 1.209 .0503 AGE −.0081 .0096 −.843 −.0604 EDU −.2780 .3498 −.795 −.0216 PS −.1837 .3706 −.496 −.0275 MS −.1304 .3040 −.429 −.0192 HS −.1416** .0657 −2.155 −.1103 INCOME .0007** .0004 1.776 .1067 MILKPLACE .326D-07 .406D-07 .803 .0222 PREFREA 3.0417* .3086 9.857 .5515 MPRICE −.9316* .2719 −3.426 −.0858 Log-likelihood −55.924 Akaike I.C. .4395 Restricted Log- −179.742 Schwarz I.C. .5630 Mc Fadden Pseudo-2 .6889 HQIC .4889 2 () 247.635 Ben./Lerman .8912 Significance level .00000 Cramer .7330 Hosmer-Lemeshow 2 25.457 Veall/Zim. .8296 Predicted percentage correction 93.00

Note: (***), (**), (*) denote significance at the 10%, 5%, and 1% levels, respectively.