The results from comparing mean differences in per capita
expenditure, poverty rates, and other household characteristics
between adopters and nonadopters appear to indicate that
adopters are better off than nonadopters. However, these comparisons
of mean differences do not account for the effect of
other characteristics of the households and thus may confound
the impact of technology adoption on expenditure and poverty
status with the influence of other characteristics. To investigate
the impact of the adoption of maize technologies on per
capita expenditure and poverty levels, multivariate approaches
that account for selection bias arising from the fact that adopters
and nonadopters may be systematically different are essential.
Smith and Todd (2005) have argued that matching should
be based on variables that influence both treatment assignment
and outcomes and are not affected by the treatment.
The choice of variables in the current study is based on previous
studies on the determinants of adoption of improved
varieties. It is widely documented that adoption and dissemination
of new technologies depends to a large extent on the
household resource endowments, characteristics of the household
head, household locational characteristics, and the nature
of the technology (Abdulai & Huffman, 2005; Feder
et al., 1985). Empirical evidence from studies by Bellon and
Taylor (1993) as well as by Bellon and Risopoulos (2001),
show that micro-environments such as soil type also tends
to influence adoption decisions. However, soil type was not
significantly related to adoption decisions in our estimates
and was therefore dropped from the propensity score specifications.
10