Integration of qualitative and quantitative information by Bayesian SEM for the survey of small scale communities

2023 Fiscal Year Final Research Report

• Project/Area Number: 21K18743
• Research Category: Grant-in-Aid for Challenging Research (Exploratory)
• Allocation Type: Multi-year Fund
• Review Section: Medium-sized Section 22:Civil engineering and related fields
• Research Institution: The University of Tokyo
• Principal Investigator: HONDA RIKI 東京大学, 大学院新領域創成科学研究科, 教授 (60301248)
• Co-Investigator(Kenkyū-buntansha): 小谷 仁務 京都大学, 工学研究科, 助教 (30814404) ; 田中 尚人 熊本大学, 熊本創生推進機構, 准教授 (60311742)
• Project Period (FY): 2021-07-09 – 2024-03-31
• Keywords: ベイジアンSEM / 小規模コミュニティ / スモールデータ / WAIC / WBIC

Outline of Final Research Achievements

This study proposed a framework that integrates quantitative and qualitative information in the analysis of community surveys with small sample sizes. It uses Bayesian SEM to derive convincing by integrating quantitative analysis with qualitative information obtained from interviews and other sources.
We developed a procedure for setting prior information, the use of WBIC as a method for evaluating the validity of the prior distribution. We conducted an actual community survey and confirmed the applicability of the method to a community survey with a small sample size. Some issues were also found, arising from the small sample size in the calculation of WBIC used in the validity evaluation of the prior distribution.

Free Research Field

防災工学

Academic Significance and Societal Importance of the Research Achievements

本研究では，コミュニティ調査の分析においてベイズ理論を適用して定式化されたベイジアンSEM（構造方程式モデリング）を利用する手法を提案した．これは，量的情報に質的情報を統合した分析を可能とすることに加え，データ数が少ない場合の分析を安定化させる効果もあるため，サンプル数が限られる小規模コミュニティへの調査などへの利用可能性が高まることが期待される．