Hi all,
I came across this article:
http://blogs.bu.edu/bumcpostdocs/2013/11/13/effects-of-remote-retroactiv...
Which claims that a research that was done on the effect of prayers on patients healing (retroactively!) wasn't done properly. It claims that:
"Of course there are many problems with this paper... It appears that most of the significance of this study can be ascribed to one outlier in the control group, whose stay in the hospital was extended. However, without access to the raw data it is hard to prove this. The fact that the median does not differ between the two treatment groups is another hint, i.e. that the results might look very different when the outlier is removed"
I didn't understand what he means, because as far as I know Median, Upper quartile and Lower quartile are not affected by extreme values in the list. So even if there was a patient who stayed for a much longer time in the hospital, it shouldn't have change the results in the final table.
So, do you agree with the article's author claim, and if so why?
Thanks.
Subscription Note:
Choosing to subscribe to this topic will automatically register you for email notifications for comments and updates on this thread.
Email notifications will be sent out daily by default unless specified otherwise on your account which you can edit by going to your userpage here and clicking on the subscriptions tab.
All values effect (or should I say affect?) the median. How much depends greatly on the data.
-----------------------------
/e
I read the page you linked. It described two different studies. One that tested deaths from a certain condition. And another that seemed to imply it was a test (using the same medical condition) of how long the person stayed in the hospital.
In each case testing retroactive prayer for a control group and a test group.
The first question that came to my mind was: Were these two studies done on the same population? I mean did they take a population of (past) patients with the medical condition in question; sort them into "survived" or "died". Then divided each of those into their respective test and control groups?
Yes, I fixed that thanks.
Can you show me an example of a list of lot of numbers where one single extreme value changes these statistic parameters in a significant way?
n values with:
I tested it, it's staying the same.
1. First I tried a list of a 200 random values between 1 to 10:
10, 5, 4, 10, 3, 8, 9, 9, 4, 3, 7, 10, 4, 5, 5, 2, 3, 9, 10, 4, 9, 1, 2, 6, 8, 1, 8, 1, 9, 8, 5, 10, 9, 9, 3, 9, 4, 1, 6, 4, 8, 4, 4, 10, 6, 10, 2, 2, 5, 5, 3, 4, 4, 1, 4, 9, 6, 6, 9, 10, 7, 7, 6, 2, 9, 10, 2, 9, 7, 8, 4, 9, 1, 4, 4, 7, 6, 7, 4, 3, 1, 1, 8, 4, 1, 3, 4, 1, 9, 9, 7, 4, 2, 7, 5, 7, 7, 10, 6, 1, 3, 3, 4, 1, 2, 8, 5, 5, 1, 8, 3, 4, 4, 6, 4, 6, 9, 7, 3, 9, 6, 1, 8, 5, 1, 2, 6, 3, 9, 5, 3, 5, 2, 5, 10, 1, 9, 1, 1, 10, 3, 1, 3, 5, 10, 5, 1, 10, 8, 10, 5, 4, 8, 10, 10, 9, 6, 6, 4, 2, 2, 1, 2, 7, 6, 5, 10, 4, 6, 2, 10, 6, 9, 9, 9, 2, 9, 10, 1, 1, 1, 1, 6, 5, 4, 9, 2, 7, 5, 1, 1, 8, 3, 10, 3, 9, 5, 5, 1, 10
and I got:
Quartile Q1: 3
Quartile Q2: 5
Quartile Q3: 8
2. Then I replaced the last value with 3628800 (= 10!) like so:
10, 5, 4, 10, 3, 8, 9, 9, 4, 3, 7, 10, 4, 5, 5, 2, 3, 9, 10, 4, 9, 1, 2, 6, 8, 1, 8, 1, 9, 8, 5, 10, 9, 9, 3, 9, 4, 1, 6, 4, 8, 4, 4, 10, 6, 10, 2, 2, 5, 5, 3, 4, 4, 1, 4, 9, 6, 6, 9, 10, 7, 7, 6, 2, 9, 10, 2, 9, 7, 8, 4, 9, 1, 4, 4, 7, 6, 7, 4, 3, 1, 1, 8, 4, 1, 3, 4, 1, 9, 9, 7, 4, 2, 7, 5, 7, 7, 10, 6, 1, 3, 3, 4, 1, 2, 8, 5, 5, 1, 8, 3, 4, 4, 6, 4, 6, 9, 7, 3, 9, 6, 1, 8, 5, 1, 2, 6, 3, 9, 5, 3, 5, 2, 5, 10, 1, 9, 1, 1, 10, 3, 1, 3, 5, 10, 5, 1, 10, 8, 10, 5, 4, 8, 10, 10, 9, 6, 6, 4, 2, 2, 1, 2, 7, 6, 5, 10, 4, 6, 2, 10, 6, 9, 9, 9, 2, 9, 10, 1, 1, 1, 1, 6, 5, 4, 9, 2, 7, 5, 1, 1, 8, 3, 10, 3, 9, 5, 5, 1, 3628800
and again I got the same result:
Quartile Q1: 3
Quartile Q2: 5
Quartile Q3: 8
You can try it yourself here, just copy paste the values list:
https://www.thecalculator.co/math/Quartile-Calculator-471.html
n! Should have been 200!
for a quick example with n=4:
Great example thanks!
1. I tried:
1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1
and I got:
Quartile Q1: 1
Quartile Q2: 1
Quartile Q3: 1.5
2. Then I tried:
1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 1, 1, 1, 2, 17
and I got:
Quartile Q1: 1
Quartile Q2: 1
Quartile Q3: 2
Strange, why wasn't it working on my first list of numbers? :-/
Sorry my mistake, but it's a too large value and the calculator can't handle it.
But anyhow, such a large value is not represents what we see in this research, look at the table he shows, the largest value is 320.
Well if you have the data and methods used; just do it both ways and compare.
Yes I will compare the lists, thanks!
edited .I reread the post. I don't understand enough to comment.
RE: OP: So the Christians failed at doing their study correctly? And that is surprising.... why? You understand it was a Christian organization reporting honestly on their own findings. If they thought there was a problem with the study they would have trashed it and conducted another to show prayer effective.
Maybe you are right.
Well an unbiased research group was contracted to present results, to the US national prayer group that funded the study...if you are talking about the 2009 heart surgery patients group.
All of which is a moot point, as prayer is a physical impossibility for humans. It takes 13 tev energy to run one particle smashing extravaganza at the LHC in cern...and I don't see any people instantaneously incinerated to dust upon initiating prayer.
@doG
"All of which is a moot point, as prayer is a physical impossibility for humans. It takes 13 tev energy to run one particle smashing extravaganza at the LHC in cern...and I don't see any people instantaneously incinerated to dust upon initiating prayer."
I understood that perfectly. I'm fine as long as you keep it simple and avoid jargon.