Quantifying irrigation system performance using PLS-SEM: a statistical approach

ABSTRACT Irrigation performance assessments have traditionally relied on variables selected without clear justification, often resulting in subjective judgments due to the absence of robust and coherent analytical frameworks. This study addresses these limitations by applying partial least squares structural equation modeling to data collected from 296 respondents in the Mahakali Irrigation System–II (MIS-II), Nepal. The analysis generated a streamlined framework comprising 6 constructs and 12 indicators, refined from an initial set of 9 constructs and 24 indicators. Service delivery (β = 0.93) emerged as the most influential determinant of irrigation performance, followed by spatial (β = 0.54), physical (β = 0.46), groundwater (β = –0.24), and financial (β = 0.10) factors, while institutional, environmental, and soil variables were excluded due to insufficient statistical significance. The resulting framework enabled quantitative assessment and benchmarking, revealing an overall performance level of 30.63% for MIS-II, indicating poor system performance. Although variable selection remains sensitive to spatial and temporal contexts, the findings offer actionable insights for targeted interventions, while underscoring the need for further research to enhance the generalizability of the developed framework.