Sample Size for expensive tests

Feb 2013
4
0
Seattle
How do we calculate sample size?

Some homosapiens have poked and grunted an assertion of approximately thirty, without giving reason. That is insufficient justification to present to a board for approval of expensive trials.

How then do we measure and assess the merits of sample size?

For a working example, a set of destructive sample test experiments on a product to determine fitness for purpose.
 
Oct 2012
751
212
Ireland
A sample size of 30 is a rule about testing with sample from a normal distribution. The t test is an exact test for this case but it can be replaced with the more simple Z test if the sample size 30 or larger, that's because as the sample size gets bigger and bigger the two tests become exactly equal and 30 is normally considered to be good enough for them to be equal.

The method used to calculate sample size will depend on the type of test you are doing. Are you just comparing the mean from one population to the mean from another? What is the hypothesis you are testing and what is your alternative hypothesis?