目次

信頼区間

定義

$$ 95\% CI=\text{標本平均} \pm 1.96SE $$

$$ 95\% CI=exp\left( \text{標本平均} \pm 1.96SE\right) $$

解釈1

解釈2

棄却されない=判定を保留する推定値の集合

適合性区間 compatible intervals

この名称を用いる利点1

A major advantage of using confidence intervals as compatibility intervals is that it directs our attention to the concrete hypotheses (effect size values) included in the interval, rather than encouraging blurry statements like ‘uncertainty is high’ or ‘low’ (Greenland, 2019b). A focus on the spread of the values contained in the interval helps avoiding falsely declaring ‘no difference’ or ‘no effect’, because the hypothesis of no effect can (and should) be discussed as but one of the many reasonable possibilities inside the interval.

この名称を用いる利点2

As a further important benefit, describing the interval helps to avoid putting too much emphasis on the point estimate (the observed association). Although the data are most compatible with the effect size given by the point estimate, the interval will usually show that, under the same background model, the data are also reasonably compatible with many other effect sizes.

二値的判断の限界

区間推定が不可能な場合

参考

1)
解釈1に基づくなら割合とも取れるが
2) , 3)
【CC-BY】Amrhein, V., & Greenland, S. (2022). Discuss practical importance of results based on interval estimates and p-value functions, not only on point estimates and null p-values. Journal of Information Technology, 37(3), 316–320. DOI