# Should I Get Excited About a Correlation of 0.7, n=8

Six Sigma – iSixSigma Forums General Forums Methodology Should I Get Excited About a Correlation of 0.7, n=8

Viewing 5 posts - 1 through 5 (of 5 total)
• Author
Posts
• #54287

roarty
Participant

I have 8 coatings. I have measured the surface energy of each coating then coated the inside wall of a pump with each and measured the increase in pump efficiency.

In other words I have two sets of paired data, n=8. I have calculated the Pearsons correlation to be 0.73

My question is; how confident can I be in my conclusion that there is a dependence between a coatings ability to improve pump efficiency and its surface energy?

I would like to say something like “For this sample size, there is only an X % chance that a correlation of 0.73 is due to random error.

How do I calculate X in this statement? Thank you.

0
#194630

MBBinWI
Participant

@cognition – Robert, as usual, is correct, but left out that you can add repetitions of the experiment to solidify the results. If there is a relationship, your power value will increase.

0
#194634

roarty
Participant

Thank you Robert,

I like your second suggested approach.

I am using the ‘Analysis ToolPak’ plug-in on Microsoft Excel.

My understanding is that my software uses the following t-value in a one tailed t-test…

t = (slope – 0)/Standard error of slope

Am I correct in saying that this is the approach you suggested?

Thank you for your suggestion MBBinWI,

The pump efficiency tests are carried out by an external company and are relatively expensive and time consuming. Normally the expense would not be a problem (we accept that R&D is expensive) but up until recently we have had a lot of naysayers (myself included) who have said it is pointless developing a coating for pump efficiency as surface roughness is the only factor that counts – any coating, as long as it cures to a smooth film, is as good as the next.

I am now convinced that surface roughness is not the whole story and I am now prepared to spend some more money on this project. So I agree with you, more data including repetitions will be necessary to increase the power value before I move on to the product development stage.

0
#194640

Robert Butler
Participant

Yes, that’s the standard expression and that is what I recommended.

0
#194642

roarty
Participant

Thank you Robert

0
Viewing 5 posts - 1 through 5 (of 5 total)

You must be logged in to reply to this topic.