The independent variables used in the statistical analyses were designed to test a
range of farm survival determinants that are discussed both by other researchers
and by Dodge Countians. The dependent variable is the dichotomous debt-to-asset
ratio indicator described in chapter 8. The independent variables fall under four
general headings: farming skill (including experience), resource access, standard
of living, and management style. Coding to achieve a useful measure of each of
these elements was carried out in an interactive process. With observational and
open-ended interview data as well as survey data, the anthropologist can visu-
alize each case; because I had seen the equipment, walked the land, visited the
homes, and talked at length with each farm family, I could assess each coding of
the data for its usefulness. If, for example, a diverse range of farms were identified
as having "high use of hired labor," I first verified the data and then considered
other, perhaps more accurate, ways to capture this aspect of an entrepreneurial
management style.

Table A-1 reports the results of univariate analysis for the association of 22
independent variables with farm survival. The first two farming skill variables are
described in chapter 4 and in appendix 1. The overall skill coding dichotomized
the cases into two groups (higher skill versus lower skill) based on the combined
ten variables discussed in appendix 1. The crop skill measure coded as "lower
skill" any farm with low peanut yield in its five-year average of production, so
long as it had no high yields in any of the other 3 crops. Experience (years of
farming) was also dichotomized at the midpoint (13.5 years in 1982).

Measurements of the resource access variables were described in chapter 4.
The good soil deficit variable (see appendix 1) was used to measure land quality.

-261-