文章基本信息

标题：Understanding significance and p-values
本地全文：下载
作者：Shrikant I Bangdiwala
期刊名称：Nepal Journal of Epidemiology
电子版ISSN：2091-0800
出版年度：2016
卷号：6
期号：1
页码：522-24
DOI：10.3126/nje.v6i1.14732
语种：English
出版社：International Nepal Epidemiological Association (INEA)
摘要：Correction 7th May 2016: p.523, Column 1, second paragraph, line 10 - 'the quantiles at the extremes - say 5%, 1%, which were useful' was changed TO 'the quantiles at the extremes - say 10%, 5%, 2%, and 1%, which were useful'. p.523, Column 2, second paragraph, line 3 - 'this is misconstrued with 'statistical significance' changed TO 'mistaken with 'statistical significance'. Since the p-value is a single index, following the ASA’s statement, we strongly support that it cannot and should not be considered as the sole basis for scientific reasoning. Given the misuses and misconceptions concerning p-values, the recommendation is to present the estimate of the effect, provide a measure of uncertainty of the estimation (e.g. confidence interval), and interpret the results in terms of scientific importance.
其他摘要：Correction 7th May 2016: p.523, Column 1, second paragraph, line 10 - 'the quantiles at the extremes - say 5%, 1%, which were useful' was changed TO 'the quantiles at the extremes - say 10%, 5%, 2%, and 1%, which were useful'. p.523, Column 2, second paragraph, line 3 - 'this is misconstrued with 'statistical significance' changed TO 'mistaken with 'statistical significance'. Since the p-value is a single index, following the ASA’s statement, we strongly support that it cannot and should not be considered as the sole basis for scientific reasoning. Given the misuses and misconceptions concerning p-values, the recommendation is to present the estimate of the effect, provide a measure of uncertainty of the estimation (e.g. confidence interval), and interpret the results in terms of scientific importance.
关键词：p value;significance;American Statistical Association