Mechanical Engineering Sample Test Size

[Mt.mandar]

You need to conduct an engineering 3-Point Bending. How many samples should you test?

I was recently asked this in an interview but was not sure how to answer. I tried saying that it would depend on time, the readiness of the samples, and the cost of the tests but the interviewer said to ignore those factors. I said there should be enough samples to make have a statistically significant outcome. So I guessed 100 samples. The interviewer seemed happy with the answer but wanted to know the minimum sample size. So I took another guess at 10 samples and the interviewer did not like that answer.

What is the correct way to answer this question??

 

[Miner]

What was the industry? If automotive, I think I know where they are coming from.

 

[Mt.mandar]

The company was Apple. You thinking it could have to do with 6sigma?

 

[Miner]

Possibly. However, Six Sigma teaches you to determine the same size based on the practical difference you want to detect as well as the desired alpha risk and power. The interviewer apparently never provided these. In automotive, the OEMs established n=100 as the minimum sample size for a capability study based on the confidence intervals of the mean and of the standard deviation. Prior to that the minimum was n=50. The automotive industry increased it because suppliers would claim capability, but the OEMs still saw product out of spec, so n=100 was somewhat empirically derived. It is possible that your interviewer came from the automotive industry.