Math Forum Discussions

Richard Slater

Posts: 3
Registered: 12/12/04

Help please.- really simple
Posted: Jul 10, 1996 1:44 PM

Plain Text

I am a researcher, stuck with a problem.

Other research papers in my field use coefficients of correlation.
This is fairly standard analysis.

My sample is skewed. Skew of 1.234, and kurtosis of 1.68.

I understand that because of this non-normally distributed sample, I
cannot use parametric tests on the data - unless I use
transformations, which I don't want to use.
I can use non - parametric tests.

My question is this - does regression / correlation constitute a
parametric or non-parametric test, or neither ?, ie does it matter
for the purposes of coefficients of correlation that the data is so skewed ?

any help gratefully etc.

Richard.