You are here: Home > Where we work > Tanzania > Hypothesis

Hypothesis

As the Sesame Marketing Project was already demonstrating increases in production and revenue for farmers undertaking training, the pilot was designed to focus on the knowledge impacts of ICT-based training. For the purposes of this pilot, it was assumed that if the knowledge could be imparted and retained, improved productivity would follow. Additionally, we were not attempting to directly compare the relative effectiveness of the two approaches, but rather to find out whether the knowledge gains seen from the conventional approach could be mirrored without the use of fixed demo plots, and their particular constraints. So, the evaluation was designed to test the following hypothesis: 

Methodology

Farm Africa worked with the Cambridge Malaysian Education and Development Trust and the Malaysian Commonwealth Studies Centre to design and implement a small pilot using tablet computers instead of demonstration plots in two of the project villages. Ten CFs and two government extension agents were trained in operating the tablets, which were loaded with locally-produced videos explaining best practice for each stage of the production cycle. Between November 2013 and April 2014, tablets were given to 10 CFs to take around to sesame farmers within their community as ‘portable demo plots’. The farmers viewed training modules relevant to key milestones in the agricultural season, testing their understanding with inbuilt learning questions. Each participating farmer was visited several times as new modules were developed, giving them the chance to go back and repeat sessions, as desired. 

The ten contact farmers collectively reached 499 sesame farmers. Of these, 49 farmers were interviewed to gather data on (i) their knowledge of key aspects of sesame cultivation, and (ii) their experiences of using the tablets.[1] The evaluation combined household surveys and interviews with the project crops officer[2] and two of the most effective CFs.[3] The surveys were developed in the local language (Kiswahili) and administered by trained enumerators using the same tablet computers. To minimise the likelihood of respondents discussing the answers to the knowledge-testing questions, all household data collection was completed in a single day. 

Initially, a ‘difference-in-difference’[4] approach was planned; using a comparison group trained through demo plots to compare against those receiving tablet-based training. The two comparison villages were selected on the basis of being similar to the tablet villages, in terms of key socio-economic, geographic and ecological parameters. Farmers in the comparison group answered knowledge questions prior to undertaking training and again some months after training was complete. These were administered by independent enumerators using paper surveys.

For the tablet training group, the same knowledge questions were uploaded to the tablets and answered before viewing the first training module. However, this approach proved insufficient for establishing a baseline, as the modules were not always completed by individual farmers operating the tablets alone. In many cases, the CFs worked with the adopter farmers to complete the modules, such that the knowledge questions were in fact a collective effort and it was not possible to isolate the baseline knowledge of individual farmers.

Instead, to estimate knowledge gains in the tablet group, participating farmers were visited by independent enumerators on a one-on-one basis. While it was not possible to retrospectively reconstruct baseline knowledge levels (we cannot ‘unlearn’), the baseline for the comparison group was judged to provide a reasonable approximation, since they had been selected on the basis of socio-economic and geographic similarity. Furthermore, none of the sites selected had received prior training on improved sesame production practices and none had access in the past to improved seed varieties. The sites were also isolated from each other, so no lateral transmission of knowledge was likely to have occurred.


1. "Lindi white," "Naliendele" and "Ziada" cultivars

2. To test the potential of the approach in general, rather than the efficacy of the initial pilot itself, we focused our limited resources on interviewing farmers trained by the best performing contact farmers. First, the subset of farmers trained by the top 3 performing CFs was selected, giving a pool of 143 farmers. From this group, 49 were randomly drawn for interview. While this approach clearly does not allow us to draw any conclusions about actual knowledge gain of the total
population of farmers reached by all contact farmers trained in the use of tablets, it does provide an assessment of the potential of this approach.

3.Farm Africa staff member

4. Selected on the basis of having reached the most farmers regularly at appropriate points in the agricultural season.

5.This refers to comparing the difference between the changes in a key variable experienced by two groups – e.g. comparing knowledge gains, rather than absolute knowledge levels.

Next section
Previous section
Back to main page

Stay up to date with the latest news and projects