Preprint: Monthly Labour Survey Misconduct since at Least the 1990s (Tanaka S. 2019-03-05)

オンラインメディア『wezzy』記事 (2019-02-07)「「毎月勤労統計調査」は90年代以前から改ざんされていた？ データ改ざんに甘い社会」をベースにした英語論文を公表しました。2001-2003にかけての誤差率の変動を分析したブログ記事 (2019-01-25)「捨てられていたサンプル: 毎月勤労統計調査2001-2003データの検証」の内容も付録としてつけてあります。

TANAKA Sigeto (2019-03-05) Monthly Labour Survey Misconduct since at Least the 1990s: Falsified Statistics in Japan. http://tsigeto.info/19m

This paper is based on a Japanese article published on an online media site wezzy. Its Appendix is based on a Japanese blog article by the author.

Abstract: The Monthly Labour Survey, which is one of the major economic statistics published by the Government of Japan, has been under criticism since January 2019 due to its negligent survey conduct and misinformation regarding its results. This paper approaches this scandal from a viewpoint of how the indicators of the quality of the survey were falsified and misreported. Based on published information regarding sample size and sampling errors, the author outlines three problems. (1) Since at least the 1990s, the survey’s sample size has been reported as larger than it actually was. (2) Since 2002, a significant portion of the sample has been secretly discarded. (3) Since 2004, the sampling error has been underreported by ignoring errors occurring in the strata of large establishments. These problems have escaped public attention as the government and academics are not critical of the falsification of basic information that determines the quality of the survey.

DOI: 10.31235/osf.io/2bf3z (on SocArXiv)