Agriculture Reference
In-Depth Information
in general to manage some amount of acreage. As a result of self-selection
bias, these numbers likely overestimate the percentage of North Dakota
farmers using these two precision technologies. In fact, the average farm
size reported by respondents was 860 ha—much higher than the average
size farm in North Dakota reported in Table 1—which indicates that op-
erators of large farms were also more likely to respond. Table 2 presents
the regression coefficients from a logistic regression model in which farm
size and number of farms workers determine whether a producer uses any
precision technologies. Note that the constant and the effect of farm size
have statistically significant effects on the probability of using these tech-
nologies, while the number of farm workers has no significant impact. We
also tested to see whether income category had significant explanatory
power, which it did not. However, we dropped income from the model
because the direction of causality (if it existed) between income levels
and precision agriculture adoption is ambiguous. That is, income category
may be endogenously determined. Based on the logistic regression model,
the probability of a producer with an average size farm (502 ha for North
Dakota) and two farm workers using Precision agriculture technologies
is P = 1/(exp(−(−2.675 + 0.003 × 502-0.027 × 2)) + 1) = 0.227. In other
words, approximately 22.7% of farmers with average-sized holdings in
North Dakota use precision agriculture technologies of some kind, includ-
ing GPS guidance and/or autosteer. Assuming a symmetric, approximately
normal distribution of farm size in North Dakota with a mean of 502 ha,
22.7% of farm operators have adopted some type of precision agriculture
technology. Thus, the logistic regression model helps attenuate bias from
respondent self-selection of the survey.
TABLE 2:
Logistic regression results relating precision agriculture use to farm size and
number of farm workers
Parameter
Description
Estimate
Standard error
α
Constant
−2.675
a
0.787
β
1
Effect of farm size
0.003
a
0.001
β
2
Effect of number of farm workers
−0.027
0.187
a
Statistical significance at the 99% confidence level or higher.
