Table 3.
Determinants of digital training participation: instrument Variables.
| IV – 1 | IV – 2 | IV – 3 | Interaction term (1*2*3) | |
|---|---|---|---|---|
| Digital training participation | 1.729*** (0.351) | 3.042* (1.697) | 1.636** (0.885) | 2.395*** (0.716) |
| Cragg-Donald Wald F Statistic | 27.65** | 44.73*** | 29.28*** | 35.72*** |
| Wu-Hausman F test | 11.37*** | 17.91*** | 8.52** | 13.71*** |
| Anderson canonical correlation LM statistic | 249.7** | 218.6*** | 164.4*** | 281.4*** |
| VIF | 1.37 | 2.25 | 1.83 | 1.41 |
| Breusch-Pagan test statistic |
5.28 (P-value 0.071) |
4.84 (P-value 0.063) |
6.26 (P-value 0.151) |
5.19 (P-value 0.107) |
| Hansen J statistic |
4.35 (P-value 0.29) |
3.29 (P-value 0.27) |
3.94 (P-value 0.35) |
2.46 (P-value 0.31) |
N = 723, endogenous variable (digital training participation), three IVs (1) access to internet connection (ATC) (2) distance to training center, and (3) farm radius, and additional explanatory variables included: household head age, education of household head, access to market, livestock ownership, off-farm income, access to credit, skill development, tenancy status, farming experience, marital status. The Cragg-Donald Wald F Statistic measures the relevance of the instruments used (F-stat > 10 = ***p < 0.01). *p < 0.1; **p < 0.05; ***p < 0.01. A Variance Inflation Factor (VIF) greater than 10 indicates potential multicollinearity. The Hansen J test assesses over-identification, with a statistic greater than 0.15 suggesting that we fail to reject the null hypothesis, supporting the validity of the instruments. The Breusch-Pagan test measures heteroscedasticity, and a p-value greater than 0.05 indicates no evidence of heteroscedasticity.