Concepts of Hypothesis Testing
Concepts of Hypothesis Testing
假说检验的目的是利用样本来测试一个或者多个群体的参数值
Steps for Testing a Hypothesis
1.设定虚无假说(null hypothesis,\(H_0\))和对立假说(alternative hypothesis,\(H_1/H_a\))
2.指定显著水准(level of significance,\(\alpha\))
3.决定适当的检定统计量(test statistic)
4.决定弃却域(rejection region)
5.下结论--推翻\(H_0\)/不推翻\(H_0\)(fail to reject \(H_0\) )
说明
1.对立假说(alternative hypothesis,\(H_1/H_a\)):研究者收集证据想要支持的假说
根据不同的\(H_1\),我们可以将检定分为:
1.单尾检定(One-sided Test)
对立假设里面有‘</>’出现,‘<'出现在\(\rightarrow\)左尾假设
型一误差:\(H_0\)对当错,在我们证明的结论是\(H_0\)可以推翻时(即h1成立)我们就犯了第一类错误(\(\alpha\) 型一误差)
型二误差:\(H_0\)错当对,在我们证明的结论是\(H_0\)不能推翻(即h1不成立)时我们就犯了第二类错误(\(\beta\) 型二误差)
单个正态总体的均值与方差
检定\(\mu\)
- 单样本\(U\)检验法:
\(X_{1}, X_{2}, \cdots, X_{n}\) 是从正态总体 \(\quad N\left(\mu, \sigma_{0}^{2}\right) \quad\) 中抽取的简单随机样本
已知 \(\sigma_{0}^{2},\) 检验假设 \(\quad \boldsymbol{H}_{0}: \boldsymbol{\mu}=\boldsymbol{\mu}_{0}, \quad \boldsymbol{H}_{1}: \mu \neq \boldsymbol{\mu}_{0}\)
原假设成立时,检验统计量为: \(\quad U=\frac{\bar{X}-\mu_{0}}{\sigma_{0} / \sqrt{n}} \sim N(0,1)\)
拒绝域为: \(\quad|u|>u_{\frac{\alpha}{2}}\) - 单样本\(t\)检验法:
\(X_{1}, X_{2}, \cdots, X_{n}\) 是从正态总体 \(N\left(\mu, \sigma^{2}\right)\) 中抽取的简单随机样本.
\(\sigma^{2}\) 未知,检验假设 \(\quad \boldsymbol{H}_{0}: \boldsymbol{\mu}=\boldsymbol{\mu}_{0}, \quad \boldsymbol{H}_{1}: \mu \neq \boldsymbol{\mu}_{0}\)
原假设成立时,检验统计量为: \(\quad T=\frac{\bar{X}-\mu_{0}}{S / \sqrt{n}} \sim t(n-1)\)
拒绝域为: \(\quad|t|>t_{\frac{\alpha}{2}}(n-1)\)
检定\(\delta^2\)
\(\chi^{2}\) 检验法
\(X_{1}, X_{2}, \cdots, X_{n}\) 是从正态总体 \(N\left(\mu, \sigma^{2}\right)\) 中抽取的简单随机样本.
检验假设: \(\quad H_{0}: \sigma^{2}=\sigma_{0}^{2}, \quad H_{1}: \sigma^{2} \neq \sigma_{0}^{2}\)
- \(\mu\) 已知
原假设成立时,检验统计量为: \(\quad \chi^{2}=\sum_{i=1}^{n}\left(\frac{X_{i}-\mu}{\sigma_{0}}\right)^{2} \sim \chi^{2}(n)\) - \(\mu\) 未知
检验假设: \(\quad H_{0}: \sigma^{2}=\sigma_{0}^{2}, \quad H_{1}: \sigma^{2} \neq \sigma_{0}^{2}\)
原假设成立时,检验统计量为:
\[\chi^{2}=(n-1) \frac{\mathrm{S}^{2}}{\sigma_{0}^{2}} \sim \chi^{2}(n-1)
\]