Hope someone can give some suggestions. I want to perform a MSA study for a discrete data type . I am confused as to what the sample size for the MSA should be? Since the data type is discrete, is there any number as to what the sample size should be?
By 'discrete data type' I'm assuming you mean a categorical type response (Pass/Fail for example) for how distinct product evaluators, would evaluate some characteristic and your primary focus is on seeing how consistent the evaluators are wrt to their determinations.
If that's the case, I'll take a stab at providing some general guidance. Usually sample size is one of the LAST things you think about when structuring an MSA. At the heart of every well structured, efficient measurement system evaluation study is a special form of a designed experiment. My suggestion is to start out by considering the practical problem at hand from a DOE point of view in terms of identifying factors, issues like crossing and nesting, etc. This will dictate an ultimate design, which in turn dicates the # of observations required.
My practical experience tells me that almost always resource or experimental execution constraints limit the number of observations that can be collected within the context of the experiment...not a hypothesis testing type framework.
In terms of setting up the design you may want to watch the first 11 minutes (I present general MSA concepts and some of the important DOE aspects of these studies) of Part 1 of this JMP On Demand webinar that I host: