Independence of computed values


New Member
Hi fellows, I'd like to ask you for help. I's sure my problem is trivial for some of you but I can't figure it out by myself..

The problem is that I have 11 gradients, just 11 values, that are calculated from a fingerprint, and I wish to know whether these 11 numbers are statistically independent or not. I've tried to google some stuff, but I've found nothing useful, like chi-square test of independence that needs a contingency table, but how can I create a contingency table from 11 values?

A huge thanks for any idea, suggestion, advice..


TS Contributor

One way to approach this ( and a particularly easy one ) is to use regression.

Fit the model X_i = m * X_j + b

is the simple linear regression of the ith measure on the jth.

if m sig different from 0 then not independant. if m is 0 then probably they are (although not necessarily).