Diagnosing Disruption in Data-Scarce Environments: An Integrated RII-TISM-BBN Framework for Public Housing Construction Management in Myanmar

Public housing construction projects are inherently vulnerable to chronic delays, cost overruns, and quality deficiencies, vulnerabilities critically amplified in Myanmar, where ongoing political and economic instability, compounded by the COVID-19 pandemic, has produced a data-scarce, high-volatility environment in which traditional deterministic risk models are ineffective. To address this gap, this study develops an integrated three-stage decision-support framework combining Relative Importance Index (RII), Total Interpretive Structural Modelling (TISM), and Bayesian Belief Networks (BBNs). RII first screens 51 candidate disruption factors down to a critical subset using survey data from 210 industry respondents. TISM then structures these factors into a causal hierarchy through focus group consensus with eight domain experts, circumventing the need for historical databases that are largely absent or unreliable in this context. This causal structure is parameterized into a BBN using discrete survey data and the Expectation-Maximization algorithm, enabling probabilistic inference under severe data scarcity. Through 693 exhaustive single-, dual-, and triple-node scenario simulations, the framework identifies eight high-impact conditional what-if scenarios. Under the highest-ranked of these (T3), the probability of favourable performance for project delays, poor quality, and cost overruns rises from baseline levels of 47.3%, 70.4%, and 57.4% to 83.0%, 78.9%, and 82.6%, respectively. Findings from this study can help practitioners diagnose the current state of public housing construction performance and prioritize corrective actions accordingly.