When trying to fit a probability distribution to quantitative results, sometimes the normal probability doesn't fit. Minitab has a wealth of distributions to pick from. Do you just pick whichever one Minitab tells you fits the best? Maybe not. Just because the distribution fits your data doesn't mean it's a good one to use. We review my top 3 distributions for product testing and some other ones that come up but may not be appropriate to use.
We'll also share what you need to think about when picking a distribution:
- think about your purpose of test
- consider your failure mode and how you're expecting your product to perform - does the typical use case of a distribution fit?
- when you have options, the simpler the distribution the better (i.e. choose a 2-parameter over a 3-parameter)
If not normal, try lognormal and a 2-parameter Weibull distribution first. If your analysis is really complicated and the stakes are high, ask a reliability engineer for some help fitting an accurate model.
Minitab has a help guide on distribution fit for reliability analysis. It lists the available distributions in Minitab, and you can read more about them. Bookmark the page as a starting point to help you. https://support.minitab.com/en-us/minitab/19/help-and-how-to/statistical-modeling/reliability/supporting-topics/distribution-models/distribution-fit/
Reliability engineers also use Gamma, Beta, and Log-Logistics distributions. Here is a link that explains them if you really want to know about these distributions. https://www.weibull.com/hotwire/issue56/relbasics56.htm