SUBSIM Radio Room Forums



SUBSIM: The Web's #1 resource for all submarine & naval simulations since 1997

Go Back   SUBSIM Radio Room Forums > General > General Topics
Forget password? Reset here

 
 
Thread Tools Display Modes
Prev Previous Post   Next Post Next
Old 03-01-06, 02:20 PM   #16
Type XXIII
Machinist's Mate
 
Join Date: Oct 2005
Location: Norway
Posts: 129
Downloads: 0
Uploads: 0
Default

Quote:
Originally Posted by August
Quote:
Originally Posted by Type XXIII
When it comes to statistical surveys, the total population is not relevant. As long as you interview a representative sample of the total population (choosing totally at random is the best,) the results from that sample will with a high degree of probability be very close to the actual situation. This can be proven mathematically.
Yeah i understand the theory, but it'd be hard to prove (to me at least) that such a small number could be representative of the total population. They claim an error rate of +/- 3% but it could easily be far greater, especially since the poll was run only once.
Also directed to the other people in this thread:

If those subjects are, as stated in the article, chosen completely at random from the total population, they are a representative sample, again because of the law of high numbers.

Of course, there can be questions raised as to the credibility of the people making the survey. Be sceptical of statistics, they can easily be made deceptive.

But don't call 1000 people a small selection. I did the math on that certain statistic, and from the numbers in the article it can be concluded that, with 99% probability, between 19% and 25% of the population for which this sample is representative (from what the article indicates, american adults that own and answered the phone), knows the names of the five simpsons.

For those math interested, the formula for the margin of error is equal to z*SQRT(p(1-p)/n).

z is a variable that depends on the desired confidence, for 99%, it is approx. 2.58. If you want the definition, go study some statistics.
p is the estimate for (population with ability)/(total population).
n is the number of subjects.
Type XXIII is offline   Reply With Quote
 


Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT -5. The time now is 12:34 AM.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2025, Jelsoft Enterprises Ltd.
Copyright © 1995- 2025 Subsim®
"Subsim" is a registered trademark, all rights reserved.