Scholarlink

<p>The idea that “statistical significance” somehow endows a result with special properties, revealing “real phenomena” vs. “results due only to chance,” is a myth. Therefore, “null results”—results found to be not “statistically significant”—should be treated like any other results, yielding a best parameter estimate and a measure of precision. The combination of parameter estimate and measure of precision does not tell the researcher anything beyond specifying a range within which the results of hypothetical identical replications of the current study are likely to fall. To make hypothesis test results useful for decision-making, researchers must incorporate —the expected costs of error and the cost of the research itself,—in contrast with the popular approach which ignores costs entirely. Incorporating fallacious “statistical significance” thinking into emerging artificial intelligence (AI) systems will produce new generations of wasteful research spending and misinterpreted results.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Statistical Stockholm syndrome: the statistical significance myth is the problem

  • 期刊论文

  • 发表日期：

    2026年3月24日

查看全文
  • Edward E. Rigdon

摘要

The idea that “statistical significance” somehow endows a result with special properties, revealing “real phenomena” vs. “results due only to chance,” is a myth. Therefore, “null results”—results found to be not “statistically significant”—should be treated like any other results, yielding a best parameter estimate and a measure of precision. The combination of parameter estimate and measure of precision does not tell the researcher anything beyond specifying a range within which the results of hypothetical identical replications of the current study are likely to fall. To make hypothesis test results useful for decision-making, researchers must incorporate —the expected costs of error and the cost of the research itself,—in contrast with the popular approach which ignores costs entirely. Incorporating fallacious “statistical significance” thinking into emerging artificial intelligence (AI) systems will produce new generations of wasteful research spending and misinterpreted results.