What is the best correlation method?

[Carlos Gonzalez]

Hello everyone, normally in my workplace exchanges of programs between different cmm or measurement equipment are experienced and being honest I have not found the best method to document these practices as well as to establish the best limit of correlation between the measurement equipment, in last days I found that the statistical control tool has a so-called "Dispersion Analysis" and offers something called "Pearson Coefficient" which provides a data indicating the strength of the correlation between two variables, does anyone have a method to document the correlation between different measuring equipment?

[mckenzie]

im not quite sure exactly what you're asking but, here goes, at first guess could you not run a gage r&r using the same part on the various cmm's/measurement equipment?

[louisd]

We used paired T tests, where we measure 36 sample parts on equipment A, assess average of the resultant data to the same 36 parts on equipment B. If the delta between means is less than 10% (of the product's dimensional tolerance), the equipment is considered equivalent.

[louisd]

Here's Paired T explanation from Minitab help:

Use Paired t to do the following:

• Determine whether the mean of the differences between two paired samples differs from 0 (or a target value).
• Calculate a range of values that is likely to include the population mean of the differences.

The paired t-test is useful for analyzing the same set of items that were measured under two different conditions, differences in measurements made on the same subject before and after a treatment, or differences between two treatments given to the same subject.

For example, the manager of a fitness facility uses a paired t-test to determine whether a group of participants improved their fitness after an 8-week program. If there is a difference in fitness measurements before and after the program, the manager can use the confidence interval to determine whether the difference is practically significant.

[Carlos Gonzalez]

im not quite sure exactly what you're asking but, here goes, at first guess could you not run a gage r&r using the same part on the various cmm's/measurement equipment?

Definitely using the MSA offers a good amount of tools to evaluate the measurement system even in the Minitab is able to deliver this dispersion analysis study, my topic is only to observe the level of correlation between different cmm through a percentage of correlation of determination or Pearsons coefficient with the level of correlation.

[Carlos Gonzalez]

Here's Paired T explanation from Minitab help:

I only have doubt about whether the best way to consider the averages, I think that it is not enough to observe the total dispersion and that the values that are close to the specification limits are left out since what is considered are the averages.

[louisd]

the theory is if you have already confirmed the measurement method is sufficient to pass a gage study (Crossed or Anova Gage R&R), and both pieces of equipment have comparable accuracy, there is no reason to conduct a Gage R&R study again on the similar piece of equipment. A statistical assessment of the measured values using Paired T, also looks at the standard deviation. if stddev is significantly higher for the new equipment, it would be a concern. otherwise, paried t comparing the means of each sample set will suffice.