Difference in difference method to impact Evaluation: a case study of fruiting Africa Project

Abstract

Development programs designed to address particular global challenges are facing increasing pressure to demonstrate their impact on the targeted com munities. The method mostly used to measure impact of development programs is by comparing the changes in outcomes of the program participants over time commonly know as before-and-after comparison. However, in some cases, the treatment and control groups are usually heterogeneous at baseline, making the difference in difference (DiD) method the most appropriate as it accounts for the changes that would have occurred in the absence of the program. The aim of the project was to evaluate if ’Fruiting Africa Project (FAP)’, an agroforestry project implemented by World Agroforestry Centre (ICRAF), made a difference in the livelihoods of beneficiaries. A sample of 300 households were randomly selected from the baseline sample of 600 households and questions of farm fruit tree diversity and abundances, food and nutrition security indicators, and knowledge on fruit tree grafting techniques. The Difference in Difference (DiD) method showed that there was significant change in the trend for the number of total and exotic fruit tree abundances from baseline to endline between the control and treatment groups for both Western and Lower Eastern Kenya. However, indigenous fruit tree abundances only had significant change in trend from baseline to endline only in Lower Eastern but not in Western Kenya. Total, exotic and indigenous fruit tree diversities had significant change in trend between control and treatment groups from baseline to endline in Lower Eastern, but not in Western Kenya. The percentage of respondents who had grafted a fruit tree significantly changed between control and treatment from baseline and endline in Western, however, not the same case for Lower Eastern Kenya. Regarding the dietary diversity variables, there was no significant change in dietary diversities between control and treat ment groups from baseline to endline. In summary, the impact of the projects’ interventions were different across the different sites and the findings in this study could contribute towards developing better programs that enhance the livelihoods of smallholder farmers.