由买买提看人间百态

topics

全部话题 - 话题: deviations
1 2 3 4 5 6 7 8 9 10 下页 末页 (共10页)
t*d
发帖数: 1290
1
多谢大家的回复。
没想到 standard error 和 standard deviation 的定义在这个版上也会有争议啊。
比如一个服从标准正态分布变量的 standard deviation 就是 1。一个来自标准正态分
布的样本,如果把它的均数作为一个随机变量的话,它的 standard deviation 就是 1
/n^0.5,其中 n 为样本大小,对吧?那么这个均数的 standard deviation 也就是样
本的一个 standard error,对吧?由于样本均数是总体均数的一个 estimator,所以
我的理解是 standard error 就是 estimator 的deviation。
有些场合,比如meta-analysis 的文献中,很多提到用 inverse-variance-weighting
的方法。由于没有明确指明,我就不明白这个variance 用的是样本本身的variance,
还是estimator 的variance?inverse-variance-weighting 方法的中心思想就是根据
各个 study 的可靠性来做个加权... 阅读全帖
t*d
发帖数: 1290
2
有些时候文章中讲到 standard deviation 实际上指得是 mean 的 standard
deviation (也就是样本的 standard error),而不是样本的 standard deviation。
怎么样才能区分清楚他们讲的到底是mena 的 standard deviation 还是样本的
standard deviation?
顺便请教一下,meta-analysis 中,inverse-variant weighted 方法用的是 1/(
standard variation of samples) 还是 1/(standard deviation of mean)?我觉得
应该是后者,因为这样即考虑了样本的方差,也考虑的样本的大小。不知是否理解正确?
多谢!
k*********g
发帖数: 791
3
原始的deviation,即population standard deviation;
就是众所周知的那个;
假如有n个数据的话,
在计算population standard deviation的时候;
根号里面除“n”;
但是population standard deviation比较适合用于data complete的情况;
对于只能获得一部分data的实际情况;
藕们应该用sample standard deviation;
in which 根号里面除“n-1”;
假如同样是n个数据的话;
藕试了2、3个小时;
发觉其中有一步过不去;
推不出来;
故此请教各位大圣;
C*******O
发帖数: 147
4
【 以下文字转载自 Statistics 讨论区 】
发信人: CAEFIMOTO (CXO), 信区: Statistics
标 题: How to calculate overall standard deviation for these two cases?
发信站: BBS 未名空间站 (Wed Jan 5 09:15:49 2011, 美东)
Case 1:
There is only 1 sample with 3 tests.
Test 1, test points=n1, mean=mu1, standard deviation=s1;
Test 2: n2, mu2, s2;
Test 3: n3, mu3, s3.
How to calculate the overall standard deviation for this case?
Case 2:
There are 2 samples, each has 3 tests like Case 1. How to calculate overall
standard deviation for this ... 阅读全帖
a**m
发帖数: 184
5
laugh
lz was talking about standard error not standard error of the mean
standard error is an estimate of standard deviation
say, you cannot get the standard deviation of the height of chinese women,
cuz you simply cannot sample all of them
however, you can sample (btw, here "sample" refers to measure their height)
1000 of them and get the standard deviation, which, to be more technical,
should be called the standard error.
and this standard error is an unbiased estimate of the standard deviatio... 阅读全帖
a**m
发帖数: 184
6
laugh
lz was talking about standard error not standard error of the mean
standard error is an estimate of standard deviation
say, you cannot get the standard deviation of the height of chinese women,
cuz you simply cannot sample all of them
however, you can sample (btw, here "sample" refers to measure their height)
1000 of them and get the standard deviation, which, to be more technical,
should be called the standard error.
and this standard error is an unbiased estimate of the standard deviatio... 阅读全帖
K****D
发帖数: 30533
7
来自主题: Mathematics版 - 弱智问题,有关standard deviation
假设一个变量x属于normal distribution, standard deviation是A, mean是X.
另一个变量y也是normal distribution, standard deviation是B, mean是Y.
X和Y不相关.
那么X*Y的standard deviation是多少? @_@
X+Y的standard deviation又是多少? @_@
t****r
发帖数: 702
8
大哥,你得定义standard deviation of what?
Standard error: estimated standard deviation of a Statistic!
Sample standard deviation: estimated standard deviation of a population!
虽然同样是估计,但是他们针对的对象是不一样的,前者的估计目标是sampling
distribution的参数,后者估计的事一个variable的distribution 的参数。
C*******O
发帖数: 147
9
Case 1:
There is only 1 sample with 3 tests.
Test 1, test points=n1, mean=mu1, standard deviation=s1;
Test 2: n2, mu2, s2;
Test 3: n3, mu3, s3.
How to calculate the overall standard deviation for this case?
Case 2:
There are 2 samples, each has 3 tests like Case 1. How to calculate overall
standard deviation for this case? (Sample 1 and 2 supposed to be the same
with production variations).
Thanks much in advance!
d******e
发帖数: 7844
10
http://mathworld.wolfram.com/StandardError.html
There appear to be two different definitions of the standard error.
The standard error of a sample of sample size n is the sample's standard
deviation divided by sqrt(n). It therefore estimates the standard deviation
of the sample mean based on the population mean
这种有专门的称呼,是SEM。
The standard error of an estimate may also be defined as the square root of
the estimated error variance sigma^2 of the quantity,
自己看。这个是大家常说的定义。
你所说的“特殊东西”不外乎是一个empirical ... 阅读全帖
t**t
发帖数: 3
11
I have a STAT problem. Ask for help!! For sample size N=30,60,120,240;generate
m=4 sets of random standard normal deviates (z scores). Does that mean for
each sample size, generate 4 times. but what is deviates z scores? Thanks
alot!!
h*********o
发帖数: 151
12
来自主题: Statistics版 - standard deviation v.s. standard error
问一个简单的统计问题哦,standard deviation 和 standard error究竟有什么差别阿
?只是说standard error是estimated standard deviation 么?
再来,如果我现在做CI,一般有公式 95% Limit = estimate +/- s.e.*critical
value.
如果是假设normal distribution, critical value = 1.96.
但是如果我假设estimate是服从一个复杂的分布,而critical value只能用bootstrap
估计的话,那么 95% Limit = estimate +/- bootstrapped s.e.,这里的应该是s.e.
还是s.d.呢?
l*****0
发帖数: 179
13
【 以下文字转载自 Quant 讨论区 】
发信人: liujx80 (xuxu), 信区: Quant
标 题: compare the deviation of data points in two different group
发信站: BBS 未名空间站 (Tue Jun 22 18:39:17 2010, 美东)
suppose I have two groups(A,B) measuring the stock price with the same unit.
and I have a data point A1 in group A and a data point B1 in group B. The
distributions of group A and B are not known and are different.
Is there any way I can compare which point deviates larger from its group
center(mean or median)?
Assume there is no out
k*******a
发帖数: 772
14
还是没看懂啊
比如说第一天 测量n1个值(对同一个物理量吗?)
如果那样,岂不是相当于测量了n1+n2+n3次同一个量?
你最后想要的是什么?定义一下什么是overal standard deviation

times, may at different time or days).
deviation=s1;
calculate or estimate (whichever is doable) the overall
q**j
发帖数: 10612
15
反了吧?standard error是estimtor的standard deviation.
standard deviation一般是给普通或者original random variable的。
看看这个。
http://en.wikipedia.org/wiki/Standard_error_%28statistics%29
d******e
发帖数: 7844
16
是你弄混了。standard deviation是要知道population density的,而density是
unknown的,所以只能用standard error来估计。
你给的这个wiki的链接无非就是换个说法而已。The standard error is the standard
deviation of the sampling distribution of a statistic。 所谓的sampling
distribution,就是population density的一个empirical estimator罢了。
M****7
发帖数: 13407
17
来自主题: Statistics版 - 问个STANDARD DEVIATION的事
一篇PAPER里说“black students score about 0。7 standard deviations below
whites in language arts”
请问这个0。7 STANDARD DEVIATIONS BELOW怎么理解?
谢谢!
x**g
发帖数: 807
18
Hi, I need some help here.
I used PROC SURVEYMEANS and I got the standard error of mean. How can I get
the standard deviation of mean?
Here is part of the output from PROC SURVEYMENAS.
Number of Clusters 91
Number of Observations 34572
Sum of Weights 34312.5015
The standard error of mean is 0.007918.
(BTW, the standard deviation from PROC MEANS is 0.725688)
Thanks a lot!
g********d
发帖数: 2022
19
来自主题: Statistics版 - Deviations in SAS and STATA
Anyone knows why the deviations from STATA and SAS are different from simple
statements like this? And how can I get the deviation number in SAS to
match that in STATA?
data test;
input x y;
cards;
2 1
3 4
5 3
7 6
4 3
6 6
;
run;
proc means;
weight y;
var x;
run;
/*STATA*/
sum x [fweight=y]
c********r
发帖数: 73
20
because there are four sample size. Maybe you just need to generate
one for each sample size, which make the total as four.
z score=[x-mean(x)]/standard deviation of x.
So, I think you just need to generate one standard normal distribution
, which has zero mean and unit variance, for each sample size.

N=30,60,120,240;generate
d******e
发帖数: 7844
21
sort的复杂度+N/2*算standard deviation的复杂度
y******y
发帖数: 107
22
来自主题: Biology版 - 如何计算ratio的 standard deviation
有两组数据,A组每组三个值,比如1,2,3. B组也是三个值,比如4,5,6。 那么该
如何求 A/B 的值以及standard deviation。谢谢。
g**********y
发帖数: 423
23
比如测量中国女人的身高,你得到150, 155, 160, 165, 170
你可以得到stdev。
sem = stdev/sqrt(n==5)
想像你从中国女人中抽1000次,
第一次的mean 为 160, 第二次159, 第三次161.5 。。。
sem 就是你计算160, 159, 161.5,。。。的stdev。
所以standard error = standard deviation of sampling of means。
s******y
发帖数: 28562
24
两者的意义完全不一样。
如果你是为了比较同一组数据里的分布情况,应该用standard deviation
如果是为了比较两个不同组的平均值的差别,那就应该用standard error of mean
l*******e
发帖数: 170
25
“比较两个不同组的平均值的差别”
这个一定要用SD(standard deviation of the difference between the means)吧?
l**********1
发帖数: 5204
26
你这个 有机会 和 錢肯 Kenneth R. Chien 说下
http://www.hms.harvard.edu/dms/BBS/fac/chien.php
看他这位 黄药师 啥反应
>
发信人: lostashoe (跑丢一只鞋), 信区: Biology
标 题: Re: 大家发文章用standard deviation,还是standard error?
发信站: BBS 未名空间站 (Mon Oct 15 13:13:57 2012, 美东)
这个问题我也想了下,我认为统计本身是客观的,“相对”准确的。
但是,搞生物的人做了10次试验,最后非常“精心地”找出了自己喜欢的3个结果,统
计的结果(P,R,Q,X^2啥的)都是基于这3个自己喜欢的结果的。所以统计结果非常
好看。
第二个人随机做一次试验,就重复不出来啦,原因是第一个人的做法就是违背统计学的
初衷的,第一个人应该统计所有10次试验的结果的。
>>
D*a
发帖数: 6830
27
为什么就是 unbiased estimate了?那我要是取了五个样本,同样可以得出一个SD啊。
当然我们大家都会说五个不够。但是从统计学上怎么理解最后一句话“ this standard
error is an unbiased estimate of the standard deviation of the height of
chinese females.”?

)
of
g**********y
发帖数: 423
28
比如测量中国女人的身高,你得到150, 155, 160, 165, 170
你可以得到stdev。
sem = stdev/sqrt(n==5)
想像你从中国女人中抽1000次,
第一次的mean 为 160, 第二次159, 第三次161.5 。。。
sem 就是你计算160, 159, 161.5,。。。的stdev。
所以standard error = standard deviation of sampling of means。
s******y
发帖数: 28562
29
两者的意义完全不一样。
如果你是为了比较同一组数据里的分布情况,应该用standard deviation
如果是为了比较两个不同组的平均值的差别,那就应该用standard error of mean
l*******e
发帖数: 170
30
“比较两个不同组的平均值的差别”
这个一定要用SD(standard deviation of the difference between the means)吧?
l**********1
发帖数: 5204
31
你这个 有机会 和 錢肯 Kenneth R. Chien 说下
http://www.hms.harvard.edu/dms/BBS/fac/chien.php
看他这位 黄药师 啥反应
>
发信人: lostashoe (跑丢一只鞋), 信区: Biology
标 题: Re: 大家发文章用standard deviation,还是standard error?
发信站: BBS 未名空间站 (Mon Oct 15 13:13:57 2012, 美东)
这个问题我也想了下,我认为统计本身是客观的,“相对”准确的。
但是,搞生物的人做了10次试验,最后非常“精心地”找出了自己喜欢的3个结果,统
计的结果(P,R,Q,X^2啥的)都是基于这3个自己喜欢的结果的。所以统计结果非常
好看。
第二个人随机做一次试验,就重复不出来啦,原因是第一个人的做法就是违背统计学的
初衷的,第一个人应该统计所有10次试验的结果的。
>>
D*a
发帖数: 6830
32
为什么就是 unbiased estimate了?那我要是取了五个样本,同样可以得出一个SD啊。
当然我们大家都会说五个不够。但是从统计学上怎么理解最后一句话“ this standard
error is an unbiased estimate of the standard deviation of the height of
chinese females.”?

)
of
i********c
发帖数: 54
33
来自主题: Mathematics版 - help!! about normal deviate
meet a coefficient called "the normal deviate corresponding to (1-x) upper
percentile".
What is it? How to calculate?
Bow! Thx a lot!
j*******d
发帖数: 55
34
ratio of an expected value and an std. deviation, 就是: mu/sigma, 数学上有
什么术语表达吗?我只知道skewness 和kurtosis,不知道这个一次项的叫什么,或者
根本就没有这么个术语?谢谢。
l*****0
发帖数: 179
35
suppose I have two groups(A,B) measuring the stock price with the same unit.
and I have a data point A1 in group A and a data point B1 in group B. The
distributions of group A and B are not known and are different.
Is there any way I can compare which point deviates larger from its group
center(mean or median)?
Assume there is no outlier, we can use (A1-mean_A)/(standard_deviation_A) .
But, I don't think we can compare the # for A1 and B1 since they are from
different distribution, can we?
Assum
l*****0
发帖数: 179
36
it's actually trying to compare the probability of the point deviates from
its mean.
l******1
发帖数: 292
37
请问standard deviation和standard error有什么共同点?很不明白他们有什么共同点
?谢谢
B****n
发帖数: 11290
38
Standard error usually means the standard deviation of the sample mean.
f*****g
发帖数: 260
39
手头上有一篇biological方面的文章。作者所作的图表,其中的standard deviation,
coefficients of variation, error bars等数据全是居于2个数据点。我的印象是,虽
然2个数据点也能用于计算上述的三项与偏差相关的criteria,但其degree of freedom
只有1。
理论上,一般至少应需有3个samples,其数据才比较有说服力吧?我不是学统计的,这
方面的知识较薄弱。想请大家指教一下,好吗?谢谢!
a********a
发帖数: 3176
40
来自主题: Statistics版 - Standard Deviation in Sudaan
Sudaan 能出Standard Deviation吗?LIKE PROC CROSSTAB? 谢谢啦!
p********0
发帖数: 186
41
来自主题: Statistics版 - What's deviation mean in logistic regression
HI,
in matlab when I called function
[B DEV] = glmfit(Xs, Ys, 'binomial', 'link', 'logit');
Xs is a 3 column 159 variables, Ys is 159 (0, 1).
my DEV = 204,
what does deviation mean? It is not the std of the new Variables, where
newVariable = B*Xs.
my newVariable has std=0.4245
l*****0
发帖数: 179
42
it's actually trying to compare the probability of the point deviates from
its mean.
a********6
发帖数: 14468
43
啥分布产生standard deviations用于simulation 较靠谱?因为SD都要求大于0,
lognormal的更要求大于1,正琢磨呢?谢谢。
q**j
发帖数: 10612
44
好像 correlation (叫r) 的分布是t distribution,standard deviation是 sqrt((1-r
^2)/(n-2)),当然假设x,y都是normal的。请问哪个地方有这个结论的推导?多谢了。
e*****t
发帖数: 642
45
来自主题: Statistics版 - 问个standard deviation test的问题
一组100个病人。在24小时内,每个小时测12次血压。想看看血压的variation(因为
variation更能体现hypertension)。文献上都是把这100个人,每小时的血压standard
deviation 求个平均值,也就是说,求 mean of every one's SD。那如果想比较第一
个小时和第二个小时mean of SD是否有统计上的差异,是用F test 还是t test?我知
道F test用来检验SD,但这里是mean of SD,不知道该用那种test。多谢
d**********r
发帖数: 24123
46
如果方差是相对一个估计值来言,就叫error
如果方差是相对一堆数据来言,就叫deviation
d******e
发帖数: 7844
47
standaard error就是standard deviation的estimator
d******e
发帖数: 7844
48
好吧,我也不跟你抠字眼了。
standard error是standard deviation的估计,这个你承认不?
q**j
发帖数: 10612
49
估计个啥。standard error是一种特殊东西的standard deviation.
p***r
发帖数: 920
50
来自主题: Statistics版 - 问个STANDARD DEVIATION的事
2 standard deviation around the mean covers 68% of the population
1 2 3 4 5 6 7 8 9 10 下页 末页 (共10页)