It appears that you are using SPSS.  If it makes sense to treat the ordinal variable as a dependent variable, you could try using ordinal logistic regression via the PLUM procedure.  As stated in the documentation linked below (see Measuring Strength of Association), you could use one of the pseudo-R2 measures.  But as noted, their interpretation is not nearly as straightforward as R2 in a linear model. 

Since your independent variable is nominal (i.e., the 3 newspapers). I would recommend using chi-squar for the nominal dependent variables.

Dear Rashid

In a case of ordinal variables you can use Kruskal-Wallis test, and for a nominal variable ( categories ) chi-square will be suitable.

Good Luck

I'm not sure I understand your question. Content analysis usually calls for categorization of text according to meanings assigned to words and phrases by an expert coder, presumably you. There are some content analysis software packages like Provalis, pcat, or Atlas.ti, but SPSS is a quantitative statistical analysis package. The only statistic that I can think of that would be valid or useful for content analysis would be frequencies (counts) of significant words or phrases, and I don't know if SPSS has that capability.

I think the question is simply wrong. A statistical test should be undertaken to determine whether a relationship predicted by theory can be identified in the obtained data. The question should be what statistic is appropriate given the kind of data I have and the relationship I am testing for.