PROJECT ASSESSMENTS - Science for Agriculture and Rural Development in Low-income Countries - page 118

six representative clusters were selected to cover the variation within the semi-arid
areas in Kenya in terms of agro-ecological characteristics, population density and
farming system. Within each cluster one representative village was selected and the
participating farm households were selected during a participatory village 'baraza'
(meeting) in each village.
The diagnostic phase was conducted in the period 2000-2001 and aimed at analys-
ing current nutrient management, determining the magnitude and major sources of
nutrient depletion, analysing financial performance, creating farm household awareness
of nutrient management aspects and jointly with the farm household, arriving at a
research and development agenda.
The participatory learning and action research was implemented in the period
2001-2002, covered on average two cropping seasons and combined various elements
of participatory research methodologies. This included the following steps:

• Group formation

• Sensitization of farmers on soil fertility status

Based on the results of the activities in the diagnostic phase, farmers' meetings
were organized to synthesize the information obtained (soil analysis results, nutrient
flows and balances, financial results) and prioritize the major problems to be
addressed.

• Identification and selection of technology options

In a farmers' meeting, technology options to address the prioritized problems
were identified. The research teams adopted various modes of discussion:
plenary, sub-groups, separate groups for men and women. The researchers also
presented potential technology options during the meeting. All options were
pooled without any order or priority. In a plenary discussion a priority ranking
was made by the farmers through consensus or voting.

• Implementation of on-farm experiments

Jointly with the farmers a research protocol was formulated, comprising a hypo-
thesis, test crop, exact description of treatments, experimental layout, aspects to
be monitored and/or measured, division of responsibilities between farmers and
researchers. In general, a simple experimental layout was designed with 1
replicate per farm (with other group members implementing similar experiments
serving as replicates), plot size varying between 25 and 100 m 2 and at most 4
treatments per experiment.

• Monitoring and data analysis

Records were kept in accordance with the research protocol. Researchers
measured aspects such as nutrient contents of manure, plant density and yield.
Farmers monitored a variety of factors such as date of planting, date of manure
application, emergence date, plant vigour, colour, weed, pest and disease
incidence, prices of inputs and outputs, etc. In many cases farmers were given
record topics to monitor their observations. Unfortunately, no structured
recording and analysis of these farmers' observations through techniques such as
matrix ranking or scoring was conducted.