r 47 -


 Third, the quantification of anins is admittedly difficult. A number of

possible proxies might be suggested to reflect this minimal standard: (a) for

any sample we might use the percentage of total farm product consumed in the

home to construct a community index and then select an arbitrary cut-off point

(ex. the lower decile or quartile) as the Sms for the community; (b) we might

estimate the fraction of total consumption (food plus non-food) out of farm

production to construct an index and then select an arbitrary cut-off as with

(a) above; or (c) we might merely use the estimated measures of farm family

consumption (farm plus non-farm produced) and then use the negative one
 i/
standard deviation for the sample as the approximate locus of the Sms. -

Each measure would be highly arbitrary but would at least render the Sms opera-

tional.

 The above model as developed represents an extreme case of a subsistence

agricultural producer. Nevertheless, the model is instructive in helping to

explain differential response or adoption in the real world.

 Given a close historical relationship between average annual food output

and a farm family's Sms, the degree of risk aversion or resistance to an

innovation will be reinforced by the following five factors:

 1. the lesser the degree of food/non-food crop diversification on the

 farm;
 2/
 2. the lesser the availability of other food sources; -


1/
 iy preference is for the third simply because it is easier to estimate and
distribution of average consumption figures for rural samples tend to skew to
the left.

 Such food sources may be provided through socio-cultural institutions, viz.
extended family, or ecologically, viz. proximate to easily secured wild/non-
cultivated food sources such as game, roots, berries.