MethodsOur analysis is comprised of a series of cross-sectional and lagged-panel OLS regression models that look at both the number of civic agriculture farms and the number of commodity agriculture farms. The first set of cross-sectional models regresses the number of civic and commodity agriculture farms on a set of social, economic, and agriculture structural variables circa 1992. The second set of lagged-panel regressions uses the number of direct sales farms and the number of large commercial farms in 1997 and regresses the 1992 independent measures on them.In a lagged-panel design, independent variables are measured at points earlier in time than the dependent measures. Additionally, lagged panel models also include dependent variables measured at earlier points in time as regressors as well. This type of analysis is more rigorous than simple cross-sectional analysis, since it will provide an assessment of which variables are associated with changes in the number of civic agriculture farms selling directly to the public and the number of commodity-oriented farms over the 1992-1997 time period.Because our regression analyses are based on a total population of counties, and thus none of lhc observed differences are due to random sampling error, we use the T statistic not as a measure of statistical significance but rather as an indicator of the strength and consistency of the relationship between the independent and dependent variables.
đang được dịch, vui lòng đợi..