Hi there!

I am a medical student working on my PhD, and I am analyzing some data for a meeting. Since we don't learn much about statistics (at least not here in Germany ), I have some beginner questions and I would be extremely happy if you could help me. The first question is the most important one at the moment, hence the bold writing.

I am working on a therapy trial with a small patient collective (n = about 70). At first, I did some descriptive calculations.

1. I do not know how I can calculate the significance of frequencies (nominally scaled data). For example: In group A, the frequency of factor X is bigger than in group B. So how do I calculate the p-value for this difference in frequency?

2. We are defining our therapy response with a scale from 1 to 6 (1 = "very much improved", 2 = "much improved" and so on until 6 = very much worse"). Is this an ordinal scale or an interval scale?

3. I calculated some correlations as well, and I am absolutely unsure whether the things I calculated are making any sense. For comparing two interval scales, I used the Pearsson corelation, and for correlations between an interval scale and a nominal scaled variable, I used crosstables and the Chi2 test. Is this correct?

4. I have two different sets of questionnaires: Self-assessment and assessment through the physician. Both have a different number of items and thus different total scores. I would like to know whether those assessments correspond to each other, so I did t-tests for paired samples. What I got was also a table with correlations (which were rather good and had siginificant p-values). Is this usable? Can I even compare those two questionnaires, even if they have different total scores?

I use PSPP, which is a freeware alternative to SPSS.

Thank you so much in advance!

Queen in the North