Estimation: Sample size?

Question

What role does the sample size play in estimation?

Thanks for the answers. But I was looking for more of a list of what role sample size plays in estimation. I already know that it determines accuracy, but what else?

Answer 1

The bigger the sample size, the more accurate the estimate.

Say I was estimating the percentage of red m&m's in a pack.

If I sampled 10 m&m's, I might I might find that the sample contains 1 red one. This could lead me to think that a pack contained 10% red m&m's.

However, if I sampled 100 m&m's, I might find that the sample contains 15 red m&m's, leading me to think that a pack contains 15% red m&m's. This answer is impossible using the sample size of 10, as you can't get 1.5 m&m's of a particular colour.

If I sampled 1000 m&m's I might find that I have 143 red ones, leading me to think that the pack contains 14.3% red m&m's.

So, as you can see, the bigger the sample size, the closer the estimation comes to the actual answer.

Answer 2

Sample size is the determinant of accuracy in estimation.Without sample there is no estimate i.e if sample size = 0 then estimate = 0 if sample size is greater then the accuracy is better depending on your estimator.

That is if you have a reasonable sample size that can give you better result but you use wrong estimator you may not be accurate in your estimation. You can not use "Mean" when you suppose to use "Mode" or "Median"

Answer 3

Well if you are speaking about sampling peoples responses to questions posed by pollsters, there are strict mathematical rules of probability, Obviously if you poll everyone involved your chances of success and the transference into estimation are pretty sound - but still not infallible! The simple answer is ... The larger the sample the better the estimate! But these devious pollsters word the questions in a way to "shape" the estimate to the way their clients want the response! After all, he who pays the piper, calls the tune - did I get that right?