Construction-management models are often judged by in-sample explanatory statistics, although high explained variance does not guarantee equally strong performance on held-out observations. This study evaluates the predictive and managerial usefulness of a bridge-construction efficiency model estimated from 348 respondents with direct bridge-construction experience in Saudi Arabia. The explanatory model was first assessed using PLS-SEM and then examined using PLSpredict, the cross-validated predictive ability test (CVPAT) and importance-performance map analysis (IPMA). The structural model explained 85.5% of Bridge Construction Efficiency, 25.1% of Stakeholder Collaboration and 24.8% of Stakeholder Communication. Q²predict was 0.220 for collaboration, 0.217 for communication and 0.066 for efficiency. CVPAT showed lower overall PLS loss than both the indicator-average and linear-model benchmarks; for Bridge Construction Efficiency, PLS loss was lower than the linear-model benchmark (p < 0.001) but not statistically lower than the indicator-average benchmark (p = 0.072). IPMA identified Stakeholder Communication (0.556) and Stakeholder Collaboration (0.454) as the highest nominal importance values for efficiency, with similar performance scores near 44. Sample-sensitivity analysis using 432 responses produced stable explained variance and stakeholder paths. The findings show that explanation, prediction and managerial priority provide different information. The model has high in-sample explanatory power, positive but uneven out-of-sample relevance, and a clear stakeholder-process priority domain. The paper illustrates how construction-management studies can separate explanatory inference from predictive and decision-oriented assessment.

Moustafa Metwally, Intan Rohani Endut, "From Explanation to Prediction in Bridge Construction Efficiency: A PLS-SEM, PLSpredict, CVPAT and IPMA Study" International Journal of Scientific Research and Engineering Development, V9(5): Page(655-658) September - October 2026. ISSN: 2581-7175. www.ijsred.com. Published by Scientific and Academic Research Publishing.