Type 1 Sample size (Minitab)

[Mr Micron]

Hi again,

Can anyone shed some light as to what is acceptable as far as a minimum sample size for a type 1 study?

I know Minitab suggests 50 samples but a quick Google suggests anywhere from 25 upwards. I know that it can also depend on the required confidence interval. But I would like to run two sets of data by you guys (can't upload images for some reason) to see if you think the data is OK, enough to not having to repeat the studies with 50+ samples? Or should they be repeated?

Thank you.

1st study run chart
30 samples
Basic statistics
Ref 2.136
Mean 2.13647
StDev 0.003928
6xStDev (SV) 0.023569
Tolerance 0.3

Bias
Bias 0.00047
T 0.650708
Pvalue 0.520
(Test bias = 0)

Capability
Cg 2.55
Cgk 2.51

%Var (repeatability) 7.86%
%Var (repeatability & bias) 7.98%


2nd study run chart

28 samples

Basic statistics
Ref 31.72
Mean 31.7275
StDev 0.73898
6xStDev (SV) 4.43386
Tolerance 200



Bias
Bias 0.0075
T 0.05370
Pvalue 0.958

(Test bias = 0)



Capability
Cg 9.02
Cgk 9.02

%Var (repeatability) 2.22%
%Var (repeatability & bias) 2.22%

 

[Miner]

Given the magnitude of the Cg/Cgk values, you probably do not need to repeat the studies with a larger sample size. The lower confidence intervals will still likely be acceptable. If this is the same measurement device for both studies, you may want to perform a linearity study to look at the change in bias over size.

 

[Mr Micron]

Given the magnitude of the Cg/Cgk values, you probably do not need to repeat the studies with a larger sample size. The lower confidence intervals will still likely be acceptable. If this is the same measurement device for both studies, you may want to perform a linearity study to look at the change in bias over size.

The data is from two separate systems and studies. My thoughts are similar to yours. The problem I have is Minitab and the procedure state 50 samples. So I'm having to justify a shift from both "standards" to an auditor. I'm looking to implement a minimum sample size to the procedure with a clause to increase when necessary. With equipment like above, if it's showing high capability and nothing out of the ordinary, why keep going with larger sample sizes for goings sake. Especially when some of the samples are very costly and physically large. Or is my logic misplaced? Thanks for the reply Miner.