Sample Size for Variable Data

Hello, I am trying to determine whether or not our sample size criteria is correct. Some time in the past, my company adopted a standard practice document that lines out sample sizes based on certain criteria, i.e. variable vs. attribute, critical to quality vs. non, software, etc. My question is in regards to our calculations for variable data sample size. We use power and sample size tool for a 1-sample t test in Minitab to calculate our sample size (alpha = 0.05, beta = 0.20, delta = sigma). We typically do not have a hypothesized, but we do have standards that we need to adhere to (example, pull force between a connector and tubing is 15 N). The data is normal and we do not use a control group. My question is, is the 1-sample t test appropriate for calculating sample sizes when only one type of sample is being tested? Knowing the source of the document, I would tend to trust the numbers, but I already re-calculated one and came up with a different answer. Please let me know if more information is needed or if this path makes sense. Thanks in advance!


Well-Known Member

You need to have a hypothesis to calculate power, for example, the average size is 45cm and you want to verify it wasn't change by more than 0.2mm for any direction (2 tailed), then you choose sample size that will have the power to identify this change. if for example, if the new average is 45.1cm the test may not have the power to "say" that the average was changed.


Less is more. Stay pure. Stay poor.
One-sample ttest is fine and typically used to compare a sample to a constant value, so i dont get your question. How about you provide a full example to convey your question!


@Miner heads up this seems like its going to be industrial or quality related.