Correlation-Matrix for Mixed Values (nominal, numeric)

Hi,

I am just wondering about my results of the correlation matrix. My use case is based on nominal and numeric attributes, stored in a Postgres DB. One table is referenced by a foreign key. If I use the ID of the foreign as my second value to be correlated, my results are not similar than in the case of a join and the usage of a text/string value instead of the ID.

Can someone of you explain these big differences?

Best

WaggaWagga