91

number of acres cultivates (p=0.12) is worth noting. The uncertainty coefficient reduces

the log likelihood by 59% (r2= 0.59) at a p-value >0.01.

 I then ran a logistic regression with the three TPB variables, attitude, subjective

norms and modern identity, which showed differences between groups (Appendix F).

The logistic regression shows that none of the three TPB variables are significant

predictors of behavior (a = 0.05). The uncertainty coefficient reduces the log likelihood

by 18% (r2= 0.18) at a p-value= 0.10. It worth indicating that the p-value is 0.10 for this

model.

 I did not run a logistic regression with any of the diffusion innovation variables

because none of them showed a difference between user groups. The variables provide no

explanation or predictive value between independent and dependent variables.

 This hypothesis is rejected based on the logistic regressions ran for selected

socioeconomic characteristics and TPB variables. The socioeconomic characteristics

were more significant predictors of farmer behavior than the TPB variables (a = 0.05).