问题如下图:
选项:
A.
B.
C.
解释:
请问老师们
为什么不是
样本均值-2.58*standard deviation
刘哲源老师讲的没看懂,谢谢!
星星_品职助教 · 2019年10月29日
同学你好,
其实可以首先从记忆的角度出发,前面均值是“样本均值”,后面的标准差自然就应该是“样本均值的标准差”,也就是标准误了。
但为什么样本均值也会有自己的标准差,这是一个很重要的知识点,涉及到中心极限定理的理解。第一就是要理解样本均值本身就是一个随机变量。这是因为样本均值是由被抽出来的样本决定的,但是每次哪个样本被抽出来却是随机的。
例如总体有1万个人,如果只抽1个样本,里面有100个人(sample size)的话,那么被抽出来的是哪100个人其实是不确定的。如果抽200次,那么前后就会有200个样本均值,每个样本均值取值都是由具体那一次被抽出来的200人决定的。所以样本均值本身也是随机变量的概念。于是就有自己的分布,均值(也就是“样本均值X bar”的均值,和自己的标准差(样本均值X bar的标准差,也就叫做标准误))。
中心极限定理说的是从一个总体中抽样,如果样本容量足够大(n>30),则样本均值作为一个随机变量就服从正态分布,样本均值的均值(注意此处理解)就为总体均值,样本均值的方差为总体方差/样本容量,即样本均值的标准差(也就是标准误)为总体标准差/根号下n。
有了中心极限定理之后,哪怕只抽一次样,也可以确定抽的这一次样所得到的样本均值也服从以上的正态分布。
而如果放宽了假设,当总体方差未知时,就可以用只抽一次样被抽出来的那个样本的标准差(注意样本的标准差和样本均值的标准差是两个概念,后者才是标准误)来代替总体的标准差。也就是样本均值的标准差可以被代替为 样本标准差S / 根号n。
所以最后不是 样本均值 - 2.58 * 样本标准差,而应该是:样本均值 - 2.58 * 样本均值的标准差 (或直接写为标准误)。
这是一个很重要的知识点,何老师在基础班里讲的很详细,如果这里不理解的话,一定要回头再听一遍基础班的中心极限定理部分。
最后,从考试的角度,如果看到一道题里有样本均值,样本标准差,然后还给了个样本的个数,那么大概率就是考察标准误了,硬凑公式也可以把题目做出来的,加油。
NO.PZ2017092702000114 问题如下 For a sample size of 65 with a meof 31 taken from a normally stributepopulation with a varianof 529, a 99% confinintervfor the population mewill have a lower limit closest to: A.23.64. B.25.41. C.30.09. A is correct.To solve, use the structure of Confininterv= Point estimate ± Reliability factor × Stanrerror, which, for a normally stributepopulation with known variance, is representethe following formula:X‾±zα/2σn\overline X\pm z_{\alpha/2}\frac\sigma{\sqrt n}X±zα/2nσ For a 99% confininterval, use z0.005 = 2.58. Also, σ = 529\sqrt{529}529 = 23. Therefore, the lower limit =31−2.58×2365=23.639831-2.58\times\frac{23}{\sqrt{65}}=23.639831−2.58×6523=23.6398 我们需要用到【置信区间结构】的计算公式解决本题the structure of Confininterv= Point estimate ± Reliability factor × Stanrerror即X‾±zα/2σn\overline X\pm z_{\alpha/2}\frac\sigma{\sqrt n}X±zα/2nσ当置信区间=99%的时候,Z0.005=2.58,且 σ = 529\sqrt{529}529 = 23. 所以,下限为 =31−2.58∗2365=23.6398=31-2.58*\frac{23}{\sqrt{65}}=23.6398=31−2.58∗6523=23.6398 老师,这种题型的计算思考步骤是什么呢,总是容易搞混,看题找不到方向?谢谢老师
NO.PZ2017092702000114 问题如下 For a sample size of 65 with a meof 31 taken from a normally stributepopulation with a varianof 529, a 99% confinintervfor the population mewill have a lower limit closest to: A.23.64. B.25.41. C.30.09. A is correct.To solve, use the structure of Confininterv= Point estimate ± Reliability factor × Stanrerror, which, for a normally stributepopulation with known variance, is representethe following formula:X‾±zα/2σn\overline X\pm z_{\alpha/2}\frac\sigma{\sqrt n}X±zα/2nσ For a 99% confininterval, use z0.005 = 2.58. Also, σ = 529\sqrt{529}529 = 23. Therefore, the lower limit =31−2.58×2365=23.639831-2.58\times\frac{23}{\sqrt{65}}=23.639831−2.58×6523=23.6398 我们需要用到【置信区间结构】的计算公式解决本题the structure of Confininterv= Point estimate ± Reliability factor × Stanrerror即X‾±zα/2σn\overline X\pm z_{\alpha/2}\frac\sigma{\sqrt n}X±zα/2nσ当置信区间=99%的时候,Z0.005=2.58,且 σ = 529\sqrt{529}529 = 23. 所以,下限为 =31−2.58∗2365=23.6398=31-2.58*\frac{23}{\sqrt{65}}=23.6398=31−2.58∗6523=23.6398 请问为什么 No.PZ2017092702000113 (选择题),这道题的置信区间计算需要使用“分母根号下n-1”,即需要用自由度求解,但是这道题不需要呢?请问老师区别在哪里?谢谢老师
NO.PZ2017092702000114 问题如下 For a sample size of 65 with a meof 31 taken from a normally stributepopulation with a varianof 529, a 99% confinintervfor the population mewill have a lower limit closest to: A.23.64. B.25.41. C.30.09. A is correct.To solve, use the structure of Confininterv= Point estimate ± Reliability factor × Stanrerror, which, for a normally stributepopulation with known variance, is representethe following formula:X‾±zα/2σn\overline X\pm z_{\alpha/2}\frac\sigma{\sqrt n}X±zα/2nσ For a 99% confininterval, use z0.005 = 2.58. Also, σ = 529\sqrt{529}529 = 23. Therefore, the lower limit =31−2.58×2365=23.639831-2.58\times\frac{23}{\sqrt{65}}=23.639831−2.58×6523=23.6398 我们需要用到【置信区间结构】的计算公式解决本题the structure of Confininterv= Point estimate ± Reliability factor × Stanrerror即X‾±zα/2σn\overline X\pm z_{\alpha/2}\frac\sigma{\sqrt n}X±zα/2nσ当置信区间=99%的时候,Z0.005=2.58,且 σ = 529\sqrt{529}529 = 23. 所以,下限为 =31−2.58∗2365=23.6398=31-2.58*\frac{23}{\sqrt{65}}=23.6398=31−2.58∗6523=23.6398 请问老师,我有两个问题,第一个是如果题干没有告诉我应该用到什么分布的话,我怎么判断是应该用z分布还是t分布呢?第二个问题是,我现在很混乱z分布和t分布的计算有什么区别呢》唯一的区别就是t分布的计算是自由度-1是吗,z分布就直接用sample zise的数量?
NO.PZ2017092702000114问题如下 For a sample size of 65 with a meof 31 taken from a normally stributepopulation with a varianof 529, a 99% confinintervfor the population mewill have a lower limit closest to:A.23.64. B.25.41. C.30.09. A is correct.To solve, use the structure of Confininterv= Point estimate ± Reliability factor × Stanrerror, which, for a normally stributepopulation with known variance, is representethe following formula:X‾±zα/2σn\overline X\pm z_{\alpha/2}\frac\sigma{\sqrt n}X±zα/2nσ For a 99% confininterval, use z0.005 = 2.58. Also, σ = 529\sqrt{529}529 = 23. Therefore, the lower limit =31−2.58×2365=23.639831-2.58\times\frac{23}{\sqrt{65}}=23.639831−2.58×6523=23.6398 我们需要用到【置信区间结构】的计算公式解决本题the structure of Confininterv= Point estimate ± Reliability factor × Stanrerror即X‾±zα/2σn\overline X\pm z_{\alpha/2}\frac\sigma{\sqrt n}X±zα/2nσ当置信区间=99%的时候,Z0.005=2.58,且 σ = 529\sqrt{529}529 = 23. 所以,下限为 =31−2.58∗2365=23.6398=31-2.58*\frac{23}{\sqrt{65}}=23.6398=31−2.58∗6523=23.6398 为什么样本标准差不是除以根号64呢?
NO.PZ2017092702000114问题如下 For a sample size of 65 with a meof 31 taken from a normally stributepopulation with a varianof 529, a 99% confinintervfor the population mewill have a lower limit closest to:A.23.64. B.25.41. C.30.09. A is correct.To solve, use the structure of Confininterv= Point estimate ± Reliability factor × Stanrerror, which, for a normally stributepopulation with known variance, is representethe following formula:X‾±zα/2σn\overline X\pm z_{\alpha/2}\frac\sigma{\sqrt n}X±zα/2nσ For a 99% confininterval, use z0.005 = 2.58. Also, σ = 529\sqrt{529}529 = 23. Therefore, the lower limit =31−2.58×2365=23.639831-2.58\times\frac{23}{\sqrt{65}}=23.639831−2.58×6523=23.6398 我们需要用到【置信区间结构】的计算公式解决本题the structure of Confininterv= Point estimate ± Reliability factor × Stanrerror即X‾±zα/2σn\overline X\pm z_{\alpha/2}\frac\sigma{\sqrt n}X±zα/2nσ当置信区间=99%的时候,Z0.005=2.58,且 σ = 529\sqrt{529}529 = 23. 所以,下限为 =31−2.58∗2365=23.6398=31-2.58*\frac{23}{\sqrt{65}}=23.6398=31−2.58∗6523=23.6398 为什么这个n用的65,但是上一题t分布里面n用的n37,查表用的 36呢?