The estimated parameters from the logit and probit models are presented in Table 2. Table 3 reports the estimated marginal effects for the likelihood to sample pawpaw and sorghum syrup products. The estimated parameters are compared between the logit and probit models for paw-paw and sorghum in Tables 2 and 3.
Table 2. Results of Logit and Probit Models for the Likelihood to Sample for the Products
| Dependent Variable | Pawpaw | Sorghum | ||
| Estimator | Logit | Probit | Logit | Probit |
|---|---|---|---|---|
| Female | -0.689** | -0.422** | -0.999*** | -0.606*** |
| Age | -0.013*** | -0.008*** | -0.012*** | -0.007*** |
| Age*Female | 0.014*** | 0.008*** | 0.017*** | 0.010*** |
| White | 0.292*** | 0.181*** | 0.365*** | 0.224*** |
| Have kids | 0.064 | 0.040 | -0.026 | -0.016 |
| Education | 0.453** | 0.284** | 0.039** | 0.023** |
| Education2 | -0.014* | -0.009* | - | - |
| Income | 0.001 | 0.0006 | 0.0005 | 0.0003 |
| Rural | 0.156 | 0.096 | 0.219** | 0.131** |
| MO | -0.080 | -0.050 | 0.335** | 0.206** |
| KY | 0.295* | 0.180* | 0.694*** | 0.418*** |
| TN | -0.039 | -0.024 | 0.582*** | 0.357*** |
| WV | 0.166 | 0.100 | 0.475** | 0.290** |
| VA | -0.160 | -0.098 | 0.025 | 0.015 |
| IL | -0.059 | -0.036 | 0.094 | 0.057 |
| IN | 0.099 | 0.062 | 0.130 | 0.081 |
| constant | -2.668 | -1.675 | 0.165 | 0.102 |
| Log Likelihood | -2241.898 | -2241.924 | -2203.740 | -2203.945 |
| LR χ2 | 37.570 | 37.520 | 71.400 | 70.990 |
| Pseudo R2 | 0.008 | 0.008 | 0.015 | 0.015 |
| N. of observations | 3406 | 3406 | 3406 | 3406 |
| Correctly predict | 62.33% | 62.33% | 64.06% | 64.12% |
| Homoscedasticity test | yes | yes | yes | yes |
Note: Asterisks indicate levels of significance: * = 0.10, ** = 0.05, and **** = 0.01.
Many estimated coefficients of the demographic characteristic variables are significant for both products. Race (white), gender (male), education, and residence (rural – for sorghum) have positive effects, while age has a negative effect.
The results showed stronger state-by-state differences for sorghum products than for pawpaw products with particularly higher likelihood to sample products in the deep southern states. Consumers in Kentucky were 7% more likely to have higher sampling interest compared to those in Ohio for pawpaw products; other state effects were not significant. Consumers in other states were more likely to have higher sampling interest than Ohio for sorghum products, including Kentucky (16%), Tennessee (13%), West Virginia (11%), and Missouri (8%). No differences were observed for consumers in Virginia, Illinois, or Indiana.