60 The Global Food System
can diffuse to others is the subject of the next section. This is followed by an
examination of the hypothesis that local-level development of LEIT can make
significant contributions to strengthening human and social capital. The paper
concludes with some considerations regarding the effective promotion of LEIT.
The Study
The research reported here is based on three case studies (in Honduras, Kenya and
Sri Lanka); a more complete description is available in Tripp (2005). The study sites
were chosen following several criteria. The cases all examine relatively large projects
that introduced examples of LEIT to many farming communities; this avoids pos-
sible biases related to the analysis of pilot project experiences. The cases were chosen
from among well-managed LEIT projects to avoid confounding technology per-
formance with problems in project management. In addition, the cases were chosen
on the basis of evidence that some of the examples of LEIT had been taken up by a
significant number of farmers. Finally, the research was organized to examine the
experience of participating farmers (and their neighbours) at least five years after the
termination of project activity. Assessment of technology adoption is often difficult
in the immediate aftermath of a project, and the research was particularly interested
in understanding further technology adaptation, diffusion, or abandonment, as
well as seeking opportunities to examine the evolution of any project-related human
and social capital. Table 2.1 presents a summary of the three cases.
The field research followed a similar pattern in the three sites, although differ-
ences in project organization were responsible for some variation in research pro-
tocols. Once the focus projects had been chosen, initial field visits were devoted to
selecting appropriate sites, identifying participant communities whose experience
seemed typical of the project, and finding comparable non-project communities to
serve as controls. Researchers held informal discussions with farmers in the area
and conducted small focus group exercises to gain a better appreciation of the
important issues. Discussions were also held with local officials and others, and all
relevant secondary data were examined. Researchers then selected random samples
for a formal survey. In each case the samples included: (i) project participants, (ii)
non-participants from the same community, (iii) farmers from nearby ‘control’
communities. The total sample sizes were 178 (Honduras), 128 (Kenya), and 210
(Sri Lanka). The researchers were present for all the interviews, although trained
enumerators were used in Honduras and Sri Lanka. The interviews focused on
general farming practices and conditions, not on the project per se. The interviews
included details of current farming practices, those of the recent past and sources
of information. The survey data were analysed and appropriate statistical tests were
utilized. Lack of space in this paper precludes discussing detailed examples, but the
reader should assume that any relationship described below (e.g. the correlation of
a particular practice with a particular farmer characteristic) exhibits statistical sig-
nificance (unless otherwise stated).