How to correlate groups of cases unified by non-numeric variables?

Hi. I am struggling to correlate "nations" of respondents in a multi-country study, and I can't figure out what the best way is, in SPSS or in any software for that matter.

As the "nationality" is an alphanumeric field (and I want to correlate based on many other fields that are numeric/continuous), I thought I would create separate binary 0/1 variables for each nationality. I then correlate these "nationality" variables, but I get negative correlations between all nationality pairs - thus rendering this approach meaningless. I assume this is because the correlation takes into account (and is heavily influenced) by the 1-0 in the nationality variable, thus making each nationality strongly negatively correlated against each other nation, and thus negative correlation *suffocates* any other meaningful correlation on the basis of the other variables. Is this the case or am I far off? Is there a better way to correlate based on groups derived from a non-numeric variable?

I would appreciate any advice.