Dear Ahmad,

If you are talking about two independent variables, the Spearmen's Rank-order correlation is used for non-parametric data.

Now, if in your case one of these two variables is independent (I would guess that the 'numerical noncontinuous ranks' in your case) and the other is dependent on that, you are associating one variable to the other. In this case you can use many types of tests:

1) You can transform your dependent variable to be normal and use a traditional linear model.

2) You may use the raw data in a non-parametric anova (i.e. Kruskal-Wallis).

3) Another option is using multinomial logistic regression.

All of these 3 suggestions depend on how you data look like.

I hope that was helpful. Thanks, Nick

Dear Nick

Thanks a lot for your answer.

I'll give you more details about my data. My research point is to find correlation between correction in Pirani scoring in one hand and correction in radiological angles (talo-calcaeneal AP and lateral view, talo-first metatarsal angle AP view), both measured before and after correction using Ponsitti method . Pirani score data is considered numerical ranks (ordinal data) and radiological data is considered scale non-parametric data. We may consider all variables as independent or consider the Pirani score ranks as dependent and radiological angles as independent. In the light of this data, what do you suggest?

Hi,

To find the correlation between numerical nonparametric data and ordinal data by Spearmen's Rank-order correlation (between -1 and +1)

Kindly see the link:

http://www.mash.dept.shef.ac.uk/Resources/MASH-WhatStatisticalTestHandout.pdf

Spearman's rank correlation coefficient is appropriate for both continuous and discrete variables including ordinal variables 

Spearman's rank correlation coefficient is appropriate for all