Masayoshi TAKAHASHI ｜ Nagasaki University – School of Information and Data Sciences

Research Activities

Faculty

Staff Introduction

Masayoshi TAKAHASHI

Position・Degree

Institute of Integrated Science and Technology, Associate Professor
School of Information and Data Sciences, Associate Professor
Ph.D. in Science and Technology
M.A. in Political Science
B.A. in Philosophy

m-takahashinagasaki-u.ac.jp

Researcher number

50781861

Date of arrival

Nagasaki University: Apr.2020-

Areas of Research

Statistical Science, Political Methodology, Causal Inference, Missing Data

Mar.2001

BA in Philosophy, Keio University

Jun.2004

MA in Political Science, California State University, Los Angeles

May.2009

Doctoral candidate (Political Methodology), Michigan State University

Nov.2011

Senior Researcher, National Statistics Center

Apr.2016

Specially-appointed Assistant Professor, IR Office, Tokyo University of Foreign Studies

Sep.2017

Ph.D. in Science and Technology, Seikei University

Oct.2018

Specially-appointed Associate Professor (Lecturer), Center for Innovative Teaching and Learning, Tokyo Institute of Technology

Dec.2018

Professional Survey Statistician

Apr.2019

Associate Professor, Department of Economics, The International University of Kagoshima

Sep.2019

Japan Society of Economic Statistics Award

Research activities

●

Theory and Implementation of Causal Inference in Statistics (Published by Kyoritsu Shuppan)

What should we do in order to increase the number of tourists to Nagasaki? Just as this question, in order to change Factor Y, we need to manipulate another Factor X. In this situation, we say that Y is the outcome and X is the cause, i.e., causality.
The 2021 Nobel prize in economics was awarded to two econometricians, Joshua Angrist and Guido Imbens for their methodological contributions to the analysis of causal relationships. Causal inference has attracted attention in many fields around the world.
This book discusses causal inference in terms of both theory (mathematical mechanism) and implementation (numerical analysis in R) in a coherent manner. The book has become very popular right after publication; thus, reprinted just a few days after publication.

●

Proposal of a novel method for causal inference (New finding in statistics)

Suppose that we want to know the effect of a new drug. It is impossible to simultaneously examine the two outcomes, one with taking the drug and the other without taking the drug. Causal inference is an important goal in many scientific fields; however, we need to think in terms of the difference in these two potential outcomes. Nevertheless, only one outcome of the potential outcomes is observed; thus, causal inference is said to be the problem of missing data.
Multiple imputation has been known as a method to deal with missing data, but little is known about how we can use multiple imputation to estimate the local average treatment effect at the cutoff. Therefore, I propose a new causal inference technique, named multiple imputation regression discontinuity design. To implement the approaches, I have written easy-to-use software, R-package MIRDD.
The finding was published in a statistics journal with an impact factor, Communications in Statistics – Simulation and Computation (Taylor & Francis, https://doi.org/10.1080/03610918.2021.1960374). This method can be applied to many fields, such as the analysis of incumbency advantage, the analysis of the relationship between education and poverty, the analysis of the effect of Covid-19 vaccination.

Figure 1: Observed Data (Left) and Simulation of Potential Outcomes (Right)

Figure 2: Observed Data and Simulation in the Non-Treated Group (Left), and Observed Data and Simulation in the Treated Group (Right)

■Legends:

Vertical line in black: Cutoff

Vertical dashed line in green: Local area

○ in gray: Observations in the non-treated group

△ in blue: Observations in the treated group

○ in red: Simulated values in the non-treated group

△ in red: Simulated values in the treated group

Educational activities

Class

School of Information and Data Sciences：

First-year Seminar, Technical English I/II, Bayesian Statistics, Social and Tourism Informatics Ⅲ, Research Project

Liberal Arts Education：

Introduction to Data Science, Introduction to Statistics

To Staff List›