Sample Size too small?

[Hatter]

I have a question for the statisticians here. If I have 100 units, how many must I randomly select and test in order to evaluate the entire group with a margin of error of less than 2%?

Sadly, I've never taken a statistics course. :(

[Morbidity]

There are two issues here

Firstly how confident do you wish to be of the result
Secondly how close do you need to be to the result

A common statistical outcome is expressed as I'm 95% confident that the outcome is "x" +/- y%.

How many must you randomly select? Depends what you are trying to test? If you are trying to test for example that in a hospital situation all equipment had been appropriately sterilised then a single negative is dire and so a smaller number need to be tested. If you are trying to test that the average height of American males is 6 foot +/- 2 inches then you'll need to test more.

Do you have any knowledge of the underlying distribution?

Basically you'll need to give me a lot more info to assist you.

[Dawnstar]

There are two issues here

Firstly how confident do you wish to be of the result
Secondly how close do you need to be to the result

A common statistical outcome is expressed as I'm 95% confident that the outcome is "x" +/- y%.

How many must you randomly select? Depends what you are trying to test? If you are trying to test for example that in a hospital situation all equipment had been appropriately sterilised then a single negative is dire and so a smaller number need to be tested. If you are trying to test that the average height of American males is 6 foot +/- 2 inches then you'll need to test more.

Do you have any knowledge of the underlying distribution?

Basically you'll need to give me a lot more info to assist you.

Listen to the wise Morbs. ;)