Forums

General Betting

There is currently 1 person viewing this thread.
ballabriggs
02 Nov 12 12:09
Joined:
Date Joined: 11 Jul 11
| Topic/replies: 534 | Blogger: ballabriggs's blog
If you have a population of 600, how big a sample do you need to get a 95% and a 99% confidence level that you have an accurate answer? 

Assume its a heads/tails probability of 0.5 . 

Also, how would this change if the probability was 0.75 v 0.25?

Post your reply

Text Format: Table: Smilies:
Forum does not support HTML
Insert Photo
Cancel
sort by:
Show
per page
Replies: 10
By:
JLivermore
When: 02 Nov 12 14:21
that question does not make much (if any) sense.

1 - The population size is irrelevant if your question has a variable sample size
2 - accurate answer for what?
3 - CI can always be calculated - I have a sample of 1 observation I can estimate a confidence interval for the true probability of heads, it's going to be 0%-100%.
By:
ballabriggs
When: 02 Nov 12 14:29
If the population is 600, I can choose what size sample I want.  Its not a "variable sample size".  I am looking to find how many people I would need to sample to get a statistically representative/significant guide to the overall population.
By:
JLivermore
When: 02 Nov 12 14:41
Well if you can chose your sample size then I would say it is variable, in that it can be chosen.  Although I was wrong about irrelevance of population.

Try this:
http://www.surveysystem.com/sscalc.htm

If you want to be 95% confident that the population average lies within +- 5% points of your sample average then you need to sample 234 people.
By:
ballabriggs
When: 02 Nov 12 14:51
Thanks for the link.

Ok, so 95% of the time, if I sample 234 people, I will get an answer of 112 to 122 , if p=0.5 ? 

Is that right?
By:
ballabriggs
When: 02 Nov 12 14:59
Actually, will rephrase this question, just to make absolutely clear. 

There is a population of 600, of which I do not know the results for any of them.  I wish to take a sample of this population, and get a reliable guide to the overall population with 95% confidence, that I have a result which I can say with 95% confidence is accurately representative of the overall population.  What size sample do I need to take to get that?
By:
JLivermore
When: 02 Nov 12 15:12
This is exactly what that calculator is answering, so IF you define 'accurate' as + or - 5% then 234 people.

I just pulled +/- 5 out of thin air.
By:
ballabriggs
When: 02 Nov 12 15:42
Thanks JLivermore.
By:
JLivermore
When: 02 Nov 12 15:59
np
By:
Mr.Angry
When: 02 Nov 12 19:54
Anyone who cannot use apostrophes won't be able to understand the answer.
By:
dave1357
When: 02 Nov 12 22:11
m'lud I would like to come to the defence of my learned friend OP.  He has contracted the word "whizzes" to "wiz's" - which isn't really an english word, but does follow the phonetic rules of text-speak.  So his crime (if it is a crime)is merely to write in the fashion of today's youth and not the greatest of grammar sins, the misused apostrophe.
sort by:
Show
per page

Post your reply

Text Format: Table: Smilies:
Forum does not support HTML
Insert Photo
Cancel
‹ back to topics
www.betfair.com