Minitab Gage R&R using Defect Count per Unit Area

  • Thread starter Brad Gover - 2010
  • Start date
B

Brad Gover - 2010

Hi all! Just got minitab two weeks ago and am still scratching my head on how to use it. I was given a study with results for analysis. The gage counts colony forming units (CFU) on a patch or petri dish. The gage reports counts like 65 or 374 colonies of germs for each petri disk. We did two operators with three trials per operator. I tried the 2 sample Poisson rate in basic stats but that does not tell me the repeatability or reproduceability. I would (if I am not crazy like to stick to defect count per unit if possible). If that does not work, how about a two factor crossed random ANOVA model with interaction in the gage study tab off of the quality tools tab? Can I resolve differences between operators, and interactions with samples with attribute data in counts? Should I treat the data as variables and forget the unit area as it is all the same unit size? Appriciate any thoughts.
 

Miner

Forum Moderator
Leader
Admin
You do have several options available. First, the Crossed Gage R&R routine in Minitab IS an ANOVA, so the only difference between running it and the EQUIVALENT ANOVA is the graphical and session window output which is more specialized for the R&R. However, the ANOVA option provides more flexibility in your study. But, you have to know what you are doing.

You are in luck. With large count numbers and equal areas of opportunity, you could use either the count data or the count per unit area data treated as if they were variable data.

To be more conservative (since Mintab makes it easy) you can use the Freeman-Tukey transform on the data (Calc > Calculator > Transform Count (Transform Proportion)). However, if you go this route, you will also need to transform your tolerances. With count number as large as you describe, it will probably not be necessary to use the F-T transform.
 
Top Bottom