Return to tldreport Forum - Message Thread - FAQ

Username: fabrcop
Date/Time: Sun, November 12, 2000 at 9:29 AM GMT
Browser: Microsoft Internet Explorer V5.01 using Windows 95
Score: 5
Subject: Rough evaluation

Message:
 

 
---> You cannot take 500 people from an Internet forum and say that the statistical trends seen amongst those 500 people represent the statistical trends of millions of users. <---

I can.

You said 500.
Suppose that 90% support IOD. This means 450 out of 500.
Standard deviation: this is (roughly) square root of 450,
around 21.21 .
I consider 3 standard deviations,
so the gaussian area guarantees that my result is 99.73% accurate.
3 standard deviations is 3 times 21.21 , say 64 .

So: 450-64 = 386
    450+64 = 514

386 means 77.2 % of 500 .
514 overpasses 100% of 500 .

So I can say this:
if all Internet users were informed about
the matter, at least 77.2% of them would support IOD.
This result is rough but accurate at 99.73% (confidence level).

>This is further complicated by the fact that nearly everybody here >has some knowledge of the situtation, whereas only a miniscule >percentage OF a percentage of the Internet population even knows >what's going on here

We were forced to take a small sample.
However, statistics can predict that this can propagate to
the whole population mantaining a similar strong "bias".
Of course we can't be sure about 90% - this is may become
as low as 77.2% .

About "biasing": read my previous post
(people *became* IOD supporters after they knew IOD's histor
and situation).

Of course, this fails in the case of the President of the USA,
since we have percentages around 50%.

This is a rough but useful evaluation for our purpose,
even though it is off-topic
(now you can really tell I am a funny guy).

Regards

Fabrizio Coppola
Italy


 


Message Thread:


Privacy Policy | Terms of Service | Cookies Policy