:bonk: Has anyone performed a alpha / beta risk estimate to TUR's or GRR's.
I would be interested in information pertaining to calculations for this.:bonk:

:bonk: Has anyone performed a alpha / beta risk estimate to TUR's or GRR's.
I would be interested in information pertaining to calculations for this.:bonk:

It would help if you clarified your question a bit. I don't know what you mean by "TUR," but I hope it's not this (http://www.cancerhelp.org.uk/help/default.asp?page=2873). :cool:

Do you want to take something like GR&R data and somehow translate it into risk estimates for supplier and customer?

Do you want to take something like GR&R data and somehow translate it into risk estimates for supplier and customer?

If this is what you're after, a gage performance curve should provide what you are looking for.

If you mean test uncertainty ratios:

http://elsmar.com/Forums/showthread.php?t=1050

The only thing I remember seeing was in a textbook. I think it was in the ASQ metrology book, where obtaining higher TUR would change the associated risks.

Hi Drikwah,

I have not performed the analysis that you are suggesting. However, I have determined measurement uncertainty. Part of that analysis is the coverage factor, k. When k = 2, the confidence interval is 95 %, therefore the alpha risk or level of significance is 1 - 95 % = 5 % at that k. Similar statements can be made for k = 2.5, k = 3, or other k values.

Wes R.

No Jim, not that. Test uncertainty ratio. I am attempting to assemble a spreadsheet showing the relationship of TUR's to Alpha / Beta Risks. This would be very similar to taking GRR data and translating it into risk estimates.
I would like to do this with TUR and GRR.

Wes, I believe this would be for the uncertainty budget part of the study that the TUR would be derived from. I would then from this uncertainty budget perform a TUR to the part or device being measured. Now I want to correlate this TUR to the alpha / beta risks.
An example would be from a simple type B uncertainty where the gage accuracy is .0001". The part or device to be verified has a tolerance .001".
giving a TUR of 10:1. The uncertainty budget would be .001005" (Type B) so now I want to know the alpha / beta risk values. Miner suggested the gage performance curve. I will examine this also.

Brad,
That is correct. I would like to create a chart showing the associated risks or correlation of TUR's to the alpha / beta risks. I have seen references to the gage performance curve but I don't know how to create one.

Hi Dirkwah,

If you have determined the combined measurement uncertainty, the k value is basically one standard deviation. Although the distributions of the uncertainty factors from each contribution may not be normal, the combined uncertainty is approximately normal.

Using a z distribution calculation, you can then determine the alpha risk as a function of the distance in standard deviation units from the population mean, which is estimated from your sample mean. Note use a t distribution if your sample size is less than 30.

The null hypothesis and alternative hypothesis would be:
H_0: mu < mu_0
H_1: mu > mu_0

Where mu_0 is the comparison value.

Wes R.

Is this to be the uncertainty study for the entire test method? If so, then a full analysis should be done, in which the calibration uncertainties are a piece, but ONLY a piece, of the whole.....and very likely the least significant piece.....

Hershal

Hershal,
sort of, but no. Just an example of what I'm after. The question that I am trying to answer is basically what is the difference in tha alpha/beta risks at differnet TUR's. Now if I have a specific TUR (Type B) how do I calculate the alpha/beta risks? I am hoping to do this in excel.

I seem to have misinterpreted the alpha and beta risks as Types A and B uncertainty components.....it appears I have a bit to learn.....

Hershal

Hershal,

Sorry if I am not clear on this. I believe these are also sometimes referred to as type I and type II risks. The risk of passing a bad part or failing a good one. I have created a Gage Performance Curve in excel based on the instructions given in the AIAG reference manual 1st edition. I am not sure it is working correctly since the example given isn't exactly the same.
If and when I get the bugs out of it I will post it on this site.