Hi, anybody know statistics?

[fs5]

Lets say I have a sample of 125,000. If I test 1,000 samples I know I will have a confidence level of 95% with a -/+ 3% margin of error.

Is there a way to tell the confidence level testing only say 200 or 500 machines? Is there a formula?

edit: google is my friend 🙂 I found some formulas online, thanks for the input anyways guys.

 

[bandana163]

A nearly LEET man like you should no.
Otherwise, all statistics are irrelevant.

 

[Mday]

there are many formulas, as statistics use models.

however, your problem is that you have to find out what model you have to use.

someone with more experience statistics will be able to give you an answer right away. but for you, you have to test the given numbers with existing forumlas to see which one it fits, and then apply it to the other situations.

 

[rudder]

I would say on average not many people here on ATOT know statistics.

 

[Analog]

Originally posted by: fivespeed5
Lets say I have a sample of 125,000. If I test 1,000 samples I know I will have a confidence level of 95% with a -/+ 3% margin of error.

Is there a way to tell the confidence level testing only say 200 or 500 machines? Is there a formula?

how do you get the 3%? Is that a given based on history?

 

[Spoooon]

I could look it up for you, since I have all of my statistics textbooks with me. But I don't know off the top of my head and I don't feel like looking it up.

I think if you google it, you'll probably find some info that will help.

 

[ghostman]

Dang, it's only been one summer and I don't remember squat from stats. From my sheet of formulas, this is the only one that looks like it might help. It solves for minimum sample size, so I don't know if it's of any use. Don't ask what those characters mean, because I don't remember (p should be p-hat, that is "p" with a "^" on it.). I also don't know what assumptions are made.

n = (z^2)*(ó^2)/err^2
n = (z^2)*p*(1-p)/err^2