p********i
发帖数: 116 | 1 问一个统计问题，reviewer 问有没有用什么statistical correction ，这个是什么意
思？谢谢。 |
e*****t
发帖数: 642 | 2 more details. otherwise hard to help you. |
p********i
发帖数: 116 | 3
谢谢您，他是这样问的，
Given all the comparisons that were made in the article was any form of
statistical correction applied? A significance level of P<0.05 should be
noted that
all values are uncorrected.
看不懂这个问题，
【在 e*****t 的大作中提到】
: more details. otherwise hard to help you.
|
s******s
发帖数: 13035 | 4 are you sure it is "correction, uncorrected"
not "correlation, uncorrelated"?
【在 p********i 的大作中提到】
:
: 谢谢您，他是这样问的，
: Given all the comparisons that were made in the article was any form of
: statistical correction applied? A significance level of P<0.05 should be
: noted that
: all values are uncorrected.
: 看不懂这个问题，
|
v**********h
发帖数: 230 | 5 feels like he is asking unadjusted or adjusted...like adjusted for any
effect modification, confounding or precision...
【在 p********i 的大作中提到】
:
: 谢谢您，他是这样问的，
: Given all the comparisons that were made in the article was any form of
: statistical correction applied? A significance level of P<0.05 should be
: noted that
: all values are uncorrected.
: 看不懂这个问题，
|
s*****0
发帖数: 357 | 6 When you conduct multiple comparisons, you need to adjust your p-values to
ensure the overall significance (sometimes referred to as family-wise error
rate) at 0.05 level.
Google multiple comparisons or grab an entry level applied statistics book
for more details.
【在 p********i 的大作中提到】
:
: 谢谢您，他是这样问的，
: Given all the comparisons that were made in the article was any form of
: statistical correction applied? A significance level of P<0.05 should be
: noted that
: all values are uncorrected.
: 看不懂这个问题，
|
c*********t
发帖数: 340 | 7 ls正解.但是FWER是讲如果你的tests的purpose和content相近的情况（比如anova那种
情况下的pairwise comparison）
As noted by Hochberg and Tamhane (1987), "If these inferences are unrelated
in terms of their content or intended use (although they may be
statistically dependent), then they should be treated separately and not
jointly" (p. 6). （wiki）
如果文章当中的hypotheses不是这种情况，就没有FWER一说，FDR更适用，没有那么
Stringent |
s**f
发帖数: 365 | 8 Multiple testing will increase the chance of observing false results (i.e.
observing a difference purely by chance).
some corrections used to the significance level:
Bonferroni
Holms
Tukey
Dunnet
【在 p********i 的大作中提到】
:
: 谢谢您，他是这样问的，
: Given all the comparisons that were made in the article was any form of
: statistical correction applied? A significance level of P<0.05 should be
: noted that
: all values are uncorrected.
: 看不懂这个问题，
|
