Abstract
{ "background": "Manufacturing systems in developing economies face unique challenges in process optimisation, with a paucity of robust, context-specific methodologies for evaluating yield improvements. Existing approaches often rely on retrospective data or theoretical models not validated in real-world, resource-constrained settings.", "purpose and objectives": "This article presents a novel methodological framework for designing and implementing randomised field trials (RFTs) to causally identify yield-optimising interventions in active manufacturing plants. The objective is to provide a rigorous, step-by-step protocol for engineers to generate high-quality evidence on process efficacy.", "methodology": "The proposed RFT methodology employs a clustered, stepped-wedge design where production lines are randomised to receive a sequence of technical interventions. Yield is measured as the proportion of output meeting specification against raw material input. The core analysis uses a generalised linear mixed model: $\\logit(P(Y{ijt}=1)) = \\beta0 + \\beta1 T{ijt} + ui + vj + \\epsilon{ijt}$, where $Y{ijt}$ is the yield binary outcome for unit $j$ in plant $i$ at time $t$, $T{ijt}$ is the treatment indicator, and $ui$, $v_j$ are random effects. Inference is based on cluster-robust standard errors.", "findings": "As a methodology article, this paper presents no empirical results from a completed trial. However, a pilot application of the framework indicated that the stepped-wedge design successfully managed plant operational constraints, with a priori power analysis suggesting a minimum detectable effect size of a 7-percentage-point yield increase with 80% power.", "conclusion": "The structured RFT methodology provides a viable and rigorous alternative to observational studies for evaluating engineering interventions in real manufacturing environments, balancing internal validity with practical feasibility.", "recommendations": "Researchers and industrial engineers should adopt this RFT framework to strengthen the evidence base for process improvements. Particular attention must be paid to