Journal Design Engineering Masthead
African Structural Engineering | 25 February 2026

Replication and Methodological Evaluation of Power-Distribution System Diagnostics for Yield Improvement in Kenya

A Quasi-Experimental Design
W, a, n, j, i, k, u, M, w, a, n, g, i, ,, K, a, m, a, u, K, a, r, i, u, k, i, ,, A, m, i, n, a, O, c, h, i, e, n, g
Replication StudyQuasi-Experimental DesignAgricultural YieldMethodological Evaluation
Quasi-experimental design tested diagnostic protocol for power-distribution systems.
Yield improvement point estimate: 0.15 tonnes/ha (95% CI: -0.08, 0.38).
Results sensitive to error clustering and baseline soil quality imbalance.
Methodology requires stricter pre-intervention matching on agronomic variables.

Abstract

{ "background": "Power-distribution system diagnostics are critical for improving agricultural yield in regions with unreliable electricity. Previous studies, predominantly simulation-based, have claimed significant benefits from such interventions, but their methodological rigour and field applicability in sub-Saharan contexts require verification.", "purpose and objectives": "This study aimed to replicate and methodologically evaluate a quasi-experimental design for assessing the impact of power-distribution equipment diagnostics on maize yield. The core objective was to test the robustness of the original diagnostic protocol and its reported yield improvements under controlled field conditions.", "methodology": "A quasi-experimental design was implemented across multiple sites. Treatment groups received the full diagnostic and corrective intervention for power-distribution infrastructure, while control groups received standard maintenance. Yield was measured as tonnes per hectare. The impact was estimated using a difference-in-differences model: $Y{it} = \\beta0 + \\beta1 (Treati \\times Postt) + \\gamma X{it} + \\epsilon_{it}$, where robust standard errors were clustered at the substation level.", "findings": "The replication found a positive but statistically non-significant yield effect. The point estimate for yield improvement was 0.15 tonnes/ha (95% CI: -0.08, 0.38), contrasting with the originally reported significant gain of 0.35 tonnes/ha. Methodological evaluation identified sensitivity of results to the clustering of standard errors and baseline imbalance in soil quality.", "conclusion": "The purported yield benefits of the diagnostic protocol were not robustly replicated under the study's conditions. The quasi-experimental design, while field-applicable, requires stricter pre-intervention matching on agronomic variables to isolate the effect of electrical diagnostics.", "recommendations": "Future applications of this diagnostic methodology should incorporate pre-intervention soil and water data into the experimental design. Power-distribution interventions should be integrated with broader precision agriculture practices to discern their standalone value.", "key words": "replication study, quasi-experiment, power