Still Going: Recent Debates on the Goldschmidt Hypothesis 269This study examines the relationship between farm structure and well-being. We rem-
edy problems in previous research by carefully selecting our farm structure variables,
incorporating indicators of nonfarm sectors, taking into account the extent of each
county’s farm dependence, and using data on all nonmetropolitan counties.If these contributions are not original, what does the Barnes and Blevins (1992)
article add to the literature? First, they controlled for agricultural dependency in a
somewhat different way than previous studies. They used national data and per-
formed analyses for non-metropolitan counties grouped into three levels of depend-
ency. A second difference is their choice of independent variables for non-farm
economic structure; they used an earnings rather than an employment measure.
The contributions of the Barnes and Blevins (1992) article thus appear to be
largely empirical. It is certainly useful to question the methodology and measures
of previous studies – and if the article had been premised on these aims, we would
likely have no qualms. However, overlooking the inroads made in the Goldschmidt
literature makes the study vulnerable to other problems.
First, unlike other recent studies, Barnes and Blevins (1992) cannot test
whether changes in farming are related to poorer well-being over time. Their anal-
ysis was cross-sectional rather than longitudinal. Since their focus seemed to be
large-scale, hired-labour-dependent farming, this is particularly critical. There is
some evidence that the negative effects of such farming, while small, are observable
over time not cross-sectionally (Lobao, 1990; see also Gilles and Dalecki, 1988).
Second, Barnes and Blevins (1992) failed to control for key non-farm variables,
such as ethnicity, unemployment levels and education. These variables are well-
known correlates of poverty and median family income and tend to be associated
with economic structure (e.g. industrialized farming areas tend to have higher
non-white populations). Relatedly, they do not control for the region of the coun-
try nor acknowledge how this fits into their findings. For example, Barnes and
Blevins (1992) found that the percentage of the population hired as farm labour
and large farms are inversely related to poverty for farming-dependent counties.
Such counties, however, are located mainly in the Great Plains and Corn Belt
(Reimund and Brooks, 1990) where analysts have found less detrimental and
sometimes positive impacts of industrialized farming (Flora and Flora, 1988;
Lobao and Schulman, 1991; van Es et al, 1988). Midwestern social structural
attributes and the greater presence of resident owners (rather than absentee land-
lords) of large farms have been given as possible explanations for these findings,
which conflict with the traditional Goldschmidt hypothesis (Lobao and Schul-
man, 1991; Swanson, 1990).
At face value, the methodology used by Barnes and Blevins’ (1992) study is
inadequate for assessing the relationship between farm structure and community
well-being. Moreover, one would expect that the analysis would be tilted toward
finding no negative impacts of large-scale or hired-labour-dependent farming,
given its cross-sectional approach, regional bias, and lack of additional control
variables. Now perhaps Barnes and Blevins did address these problems in an earlier