Enlargement of the scope of data analysis by extending nonparametric tests

2021 Fiscal Year Final Research Report

• Project/Area Number: 19K06856
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Review Section: Basic Section 45040:Ecology and environment-related
• Research Institution: Kyushu University
• Principal Investigator: Kasuya Eiiti 九州大学, 理学研究院, 准教授 (00161050)
• Project Period (FY): 2019-04-01 – 2022-03-31
• Keywords: データ解析 / ノンパラメトリック検定 / ノンパラメトリック法 / 確率分布 / 不等分散 / データの分布 / 高次のモーメント

Outline of Final Research Achievements

For a wide range of data analysis, nonparametric statistical tests have been applied. These tests make a kind of assumptions on the distribution of the data. However, this has been often overlooked. Mann-Whitney U test (i.e., Wilcoxon rank sum test), which is a test for comparison of locations between 2 treatments in unmatched data, does not allow unequal variances, Several nonparametric tests that allowed unequal variances were postulated. Whether these nonparametric tests cover actual problems in data analysis is examined. The region covered by these methods and that not covered by them are clarified. Factors affecting this result is also clarified.

Free Research Field

生態学

Academic Significance and Societal Importance of the Research Achievements

データ解析の広い範囲において、母集団分布が既存のよく使用される確率分布（代表的には正規分布など）にしたがうことは想定し難く、その場合、ノンパラメトリックな検定が広く使われている。だが、ノンパラメトリックな検定といえどもどんな場合にでも使えるわけではない。これまでのノンパラメトリックな検定でカバーされている範囲されていない範囲を検討し、明らかにした。