October 11, 2012 at 12:38 pm #54200
Our current process produces a widget with a residual chemical left on the widget with a mean = 42 ug/kg and a standard deviation of 16 ug/kg (n = 40). I want to remove a catalyst to see if I can drive the residual to <8 ug/kg. The Method Detection Limit is 8 ug/kg. How many samples would I have to run to determine that I can consistently meet the <8 ug/kg? I was looking at a 1 sample t test or maybe a 2 sample t-test but thought that I would be comparing 2 different distributions. The known distribution is normal but the proposed distribution from the removal of the catalyst would have a boundary at 8 ug/kg and it would be a rare event to have a detectable concentration, so I was thinking of a Poisson distribution (and then using a 1-sample Poisson sample size calculation). How shall I proceed?0October 18, 2012 at 12:15 pm #194248
EustonParticipant@dabrewajeff Include @dabrewajeff in your post and this person will
be notified via email.
From your description, logic tells me you’re comparing 2 normal distributions. Your answer is right in your question. Therefore, you should be able to plug the values into Minitab and get your sample size.
Don’t forget you need to define one (or more) levels of confidence that you’ve hypothesized correctly!
Hope this helps. Jeff0
You must be logged in to reply to this topic.