开发者:上海品职教育科技有限公司 隐私政策详情

应用版本:4.2.11(IOS)|3.2.5(安卓)APP下载

梦梦 · 2024年05月13日

P value为什么是概率密度函数的右尾

NO.PZ2020010304000054

问题如下:

Suppose that for a linear regression, an estimation of slope is stated as having a one-sided p-value of 0.04. What does this mean?

解释:

The p-value tells us the “probability of observing a test statistic that is more extreme than the one computed from the observed data when the null hypothesis is true.” In other words, if the estimate of the slope is true, then in a randomized trial the probability of getting a larger new slope estimate is 4%.Assuming a normal distribution, this will correspond to the value of x that solves the following equation:

1 - Φ(x) = 0.04

Using the excel function NORMSINV gives x = 1.75.



老师,p value和a不都应该是在左边吗?概率累积函数P value左侧的面积才是p value对应的概率呀?

2 个答案

pzqa39 · 2024年05月16日

嗨,爱思考的PZer你好:


答案里面那句话可以翻译成:p 值告诉我们“在零假设为真时,观察到比从观察数据计算出的检验统计量更极端的检验统计量的概率。” 换句话说,如果斜率估计值为真,那么在随机试验中得到一个更大新斜率估计值的概率是 4%。


其实这道题默认是检验斜率大于0,因为答案先说了是“零假设”,又提到的得到一个【更大】新斜率的估计值的概率是4%,这个更大就说明观察的是右侧。感觉这里的题干确实有点没说清楚,应该给一下检验的是什么,这样就清楚是右侧了。

----------------------------------------------
就算太阳没有迎着我们而来,我们正在朝着它而去,加油!

pzqa39 · 2024年05月14日

嗨,爱思考的PZer你好:


实际上P值的位置(左侧还是右侧)取决于你正在进行的假设检验类型。P值确实代表了在原假设为真的情况下,观察到的数据或更极端数据出现的概率,但这个“更极端”既可以指左侧也可以指右侧,具体取决于检验的方向。

右侧检验(Right-tailed test):

  • 当我们检验的是数据点是否大于某个值(比如,新药的效果是否显著优于安慰剂),或者检验的是总体均值是否大于一个特定值时,我们关注的是分布的右侧尾部。此时,P值就是从检验统计量的观测值到分布的右端(正无穷)的累积概率,即概率密度函数(PDF)右侧尾部的面积。

左侧检验(Left-tailed test):

  • 相反,如果我们想知道数据点是否小于某个值(比如,某种疗法是否显著减少了不良反应的发生率),或者检验总体均值是否小于一个特定值,这时我们关注的是分布的左侧尾部。P值将是从负无穷到检验统计量观测值的累积概率,即PDF左侧尾部的面积。

双侧检验(Two-tailed test):

  • 如果我们不预先设定方向,只是想检验观测到的差异是否显著,无论它是大于还是小于某个值,那么P值将是对称分布在两侧的面积。通常是将从观测值到正无穷(右侧)和从负无穷到观测值(左侧)的累积概率相加,然后除以2,或者直接找到两边相对应的更极端单侧P值并取较小的那个。


----------------------------------------------
就算太阳没有迎着我们而来,我们正在朝着它而去,加油!

梦梦 · 2024年05月16日

1、老师我还是不明白,从哪里能看出来拒绝域是分位点右边的面积,而不是左边的面积。

梦梦 · 2024年05月16日

2、probability of observing a test statistic that is more extreme than the one computed from the observed data when the null hypothesis is true.” In other words, if the estimate of the slope is true, then in a randomized trial the probability of getting a larger new slope estimate is 4%.中文是?

  • 2

    回答
  • 1

    关注
  • 136

    浏览
相关问题

NO.PZ2020010304000054 问题如下 Suppose thfor a lineregression, estimation of slope is statehaving a one-sip-value of 0.04. Whes this mean? The p-value tells us the “probability of observing a test statistic this more extreme ththe one computefrom the observeta when the null hypothesis is true.” In other wor, if the estimate of the slope is true, then in a ranmizetrithe probability of getting a larger new slope estimate is 4%.Assuming a normstribution, this will corresponto the value of x thsolves the following equation:1 - Φ(x) = 0.04 Using the excel function NORMSINV gives x = 1.75. 不太懂题目的考点,也不明白解析。

2024-02-07 22:34 1 · 回答

NO.PZ2020010304000054 问题如下 Suppose thfor a lineregression, estimation of slope is statehaving a one-sip-value of 0.04. Whes this mean? The p-value tells us the “probability of observing a test statistic this more extreme ththe one computefrom the observeta when the null hypothesis is true.” In other wor, if the estimate of the slope is true, then in a ranmizetrithe probability of getting a larger new slope estimate is 4%.Assuming a normstribution, this will corresponto the value of x thsolves the following equation:1 - Φ(x) = 0.04 Using the excel function NORMSINV gives x = 1.75. 还是不懂1.75是怎么得出来的?

2023-03-04 20:58 2 · 回答

NO.PZ2020010304000054问题如下Suppose thfor a lineregression, estimation of slope is statehaving a one-sip-value of 0.04. Whes this mean?The p-value tells us the “probability of observing a test statistic this more extreme ththe one computefrom the observeta when the null hypothesis is true.” In other wor, if the estimate of the slope is true, then in a ranmizetrithe probability of getting a larger new slope estimate is 4%.Assuming a normstribution, this will corresponto the value of x thsolves the following equation:1 - Φ(x) = 0.04 Using the excel function NORMSINV gives x = 1.75. 老师我理解p value是最小拒绝原假设的α,这意思是α再变小就会拒绝吗,然后为啥是右尾哦,有点没反应过来

2022-05-03 06:04 1 · 回答

请问这道题是匹配哪个section的知识点啊

2020-02-24 15:34 1 · 回答