Rebuilding and repairing damaged physical infrastructure is a primary source of disaster aid spending following major earthquakes. Although aid distribution is monitored, it is not well understood how economic support and technical assistance affect reconstruction behavior. This study develops and evaluates a Bayesian item response theory modeling framework for estimating the probability of reconstructive action from household-level survey data. Household responses on reconstruction status, aid received, and willingness to commit additional resources from Inter-Agency Common Feedback Project surveys (n = 5,913) collected eleven, twelve, and fourteen months after the Gorkha, Nepal, earthquake are used to estimate the probability of reconstructive action. Results show differences in marginal reconstruction probabilities ranging from 2 to 78 percent across varying combinations of aid receipt and household willingness to commit additional resources. Estimated reconstruction probabilities are lowest for households with low willingness to commit additional resources and households that have not received a reconstruction-related engineering consultation. All model results showed strong variability with geographic location. These findings provide detailed quantitative estimates of earthquake recovery that have not previously been available and offer a promising methodology for using future postdisaster household-level survey data.

在大地震之后，重建和修复受损的物理基础设施是灾难援助支出的主要来源。尽管对援助的分配进行了监测，但人们对经济支持和技术援助如何影响重建行为还没有很好的了解。这项研究开发并评估了贝叶斯项目响应理论建模框架，用于从家庭水平的调查数据中估计采取重建行动的可能性。家庭对重建状况，收到的援助以及愿意从机构间共同反馈项目调查中投入额外资源的回应（n 尼泊尔地震之后的11、12和14个月收集了5913条地震），以评估重建行动的可能性。结果显示，在各种不同的援助接收方式和家庭愿意投入额外资源的组合中，边际重建概率的差异介于2％到78％之间。对于不愿投入更多资源的家庭和尚未接受与重建相关的工程咨询的家庭，估计的重建概率最低。所有模型结果均显示地理位置差异很大。这些发现提供了以前没有的详细的地震恢复定量估计，并为使用未来的灾后家庭级调查数据提供了有希望的方法。