Hai everyone

[hazariza]

Hello
My name Hazariza
Im from Malaysia
I found this side because of my problem with colleration
If someone here can help me with my problem, im very grateful
Im not familiar with statistic
So, please help me...
Thank you

 

[hlsmith]

Hai,

Please describe your problem in more detail.

 

[hazariza]

Thanks for the reply

We have an instrument to count and measure size of micron size sample.Recently we buy another one but update version, more complicated software,but field of view for the new instrument is less than the older one.

Usually, we just use r2 linearity to make correlation. But this instruments. Is very hard to find the exact spot even in same instruments. So we think the r2 correlation is not so good to correlate this 2 instrument

What the best method, to correlate this 2 equipment???

Thanks for any advice

 

[rogojel]

hi,
check out the Bland-Altman plot. https://en.wikipedia.org/wiki/Bland–Altman_plot

 

[hazariza]

Thanks for the suggestion...
I've look at it...
But someone suggested GR&R..
Does this suitable??
Thanks

 

[rogojel]

Nope,
not for comparing two instruments. It is great tool to analyse measurement errors.

regards

 

[hazariza]

So, let say if i use bland- altman plot,the relationship between my 2 instrument is not so good, can we have some kind of coefficient which i can use to make data from instrument 1 is also valid if i run in instrument 2??

 

[rogojel]

I think not. A bad instrument/measurement system is just a bad instrument/measurement system. Recalibrate, repair, do a gage R&R and fix the problems, if possible. do not use the bad system any more, if it is not possible to fix it.

regards

 

[hazariza]

About the repair and recalibrate, we cant do that because the new instrument using different system and software, but still have same function with the old one. Even the vendor of this instrument say they can't give the exact reading with the old one. So, that why we need correlation between 2 instrument

 

[rogojel]

If this were possible no one would buy expensive instruments, right? You should not kid yourselves, a broken instrument is just that, there is no statistical wizardry to help .