I have three variables, x,y, z. My sample n is about 17,000.

My pearson coeficients & r^2:

for x -> z it is 0.187, 0.034

for y -> z it's 0.311, 0.096

testing for significance, I receive a z-score of 12.3 & p < 0.00

So, I know the difference is significant, but I want to say in (somewhat layman's terms) how strong it is. So, I have looked at coefficient of determination, r^2, to get the ratio: 0.096:0.034, or 2.8.

Is it then correct/accurate to say the correlation between y & z is 2.8 times as strong as the correlation between x & z?


It is accurate to say the explained variance between y & z is 2.8 times as strong as the explained variance between x & z.

Or, that z explains almost 10% of the variance of y, but only an estimated 3% of y.

More generally, you could highlight that the correlations you found are both positive and small to moderate. And that given the size of the sample the difference is significant (for a smaller n of, say 400, the difference would not have been significant).

| cite | improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.