Non-parametric Analysis Discussion > Mrs
Hi Fari,
Great question! It sounds like you should be using a non-parametric correlation (e.g. Spearman's Rho).
Here is the syntax for that:
nonpar corr
/variables = variablename1 variablename2
/print = spearman.
With that said, it is common for people to treat ordinal variables (such as your 0-3 scale) as continuous and use a simple Pearson's correlation coefficient, but strictly speaking that is not correct, since the ordinal variable is most likely not normally distributed.
I hope that helps!
July 10, 2013 |
Jeremy Taylor
Jeremy Taylor
I am using a scale which is consist of 4 components, the value of each component is discrete values 0 (normal),1(mild),2(moderate),3(severe). I have used this scale for 200 patients, I am going to find the correlation of this scale with some continues variables, theses continues variable also measured for 200 patients, I don't think so, I can use multivariate or multiple-regression because the dependent variable ( each component in scale) is discrete values and independent variables continues values, Can you please suggest a method to find this correlation?