Which statistical test to use? Compare 5 categories with data (ratio scale)

M

MaxLenk

New Member
#1
#1
Dear all,
I desperately in need!
I have two samples (reports of 30 good and 30 bad companies). For each sample I classified statements in five categories (one being the "best" category, five being the "worst" --> ordinal scale). Within the categories, I have absolute frequencies. One example looks like this:

Category 1 Category 2 Category 3 Category 4 Category 5
Company A (good) 47 32 13 7 12
Company B (bad) 17 22 21 13 15

For a more detailed example, look at the attached file. I now want to know, if the reporting for the good companies is significantly better, than the reporting of the bad companies. Can anyone tell me which test or what kind of test I need to use??
I really appreciate any kind of help! Thanks a lot for your time!!
Best regards, Max
 
Karabiner

Karabiner

TS Contributor
#2
#2
I now want to know, if the reporting for the good companies is significantly better, than the reporting of the bad companies
Click to expand...
How do you define your outcome? It could for example be the relative frequency of selected categories (e.g. for each company, "% of reports which are in category 1 or 2"), or the absolute frequency of negative reports (or of positive reports), or the absolute difference between negative and positive reports, or simply the median category for each company. Any of these suggestions would lead to a single dependent variable, and you could compare this between groups using t-test or U-test. But maybe you have a different idea with regard to the outcome.

With kind regards

K.
 
M

MaxLenk

New Member
#3
#3
Hey Karabiner,
thanks a lot for taking your time to help me.
The data set is part of my thesis. In the end I would like to be able to say (backed up with a significant test) that the reporting in the good reports is better than the reporting in the bad reports. I could use t-tests to show if that is the case for each of the categories (relative and absolut frequencies). What I'm looking for is, if there is a test that not only looks at a single dependent variable but looks at all the frequencies in the five categories at the same time and analyzes it.
Best regards,
Max
 
Karabiner

Karabiner

TS Contributor
#4
#4
What I'm looking for is, if there is a test that not only looks at a single dependent variable but looks at all the frequencies in the five categories at the same time and analyzes it.
Click to expand...
Possibly a multilvelel model. Individual respones are nested within companies, companies constitute groups. Unfortunately, I do not know how well ordinal responses can be handled by multilevel models.

With kind regards

K.
 
You must log in or register to reply here.