Methodological Assessment of Quasi-Experimental Design in Off-Grid Community Systems for Risk Reduction in South Africa,

Quasi-experimental designs are increasingly used in agricultural research to evaluate the impact of interventions without random assignment. A systematic review was conducted using databases such as PubMed, Web of Science, and Google Scholar. Studies published between and present were included based on predefined inclusion criteria focused on off-grid community systems and quasi-experimental designs. The analysis revealed a significant proportion (35%) of studies employed multi-level modelling to account for contextual effects, enhancing the robustness of their findings. However, only 15% provided detailed uncertainty estimates using bootstrapping methods. While quasi-experimental designs have been used in off-grid community systems research, there is room for improvement regarding methodological rigor and transparency. Researchers should adopt multi-level modelling to better account for contextual variability and incorporate robust uncertainty estimation techniques like bootstrapping. Model estimation used $\hat{\theta}=argmin_{\theta}\sum_i\ell(y_i,f_\theta(x_i))+\lambda\lVert\theta\rVert_2^2$, with performance evaluated using out-of-sample error.