E

I am doing my M.Sc. thesis and I am a statistics dummy which couldn't figure out a solution from the statistics for dummies book

I would be very grateful if someone could help me with the following problem:

I conducted a content analysis of 44 policy documents from 44 different countries and inserted numbers of references under categories in Excel (e.g. number of references from each country case concerning marine ecosystems - where marine ecosystems is one category).

For each category - which is essentially a variable - I assigned 4 classes of quartiles depending on the number of references:

Class 1: 0 references

Class 2: x number of references that are >0 up to 0.25

Class 3: x references >0.25 up to 0.75

Class 4: x references >0.75

Each country can belong to only one class for each variable.

What I want to do now is to compare 2 variables by using the classes to see if they are associated (related). I tried with contingency tables by using only the values from the low (class 1) and high (class 4) groups, and then test the table with a chi square but I am not sure if this is the correct method.

My table looked something like this:

(Then I tried to put the remaining 2 categories and have a complete set 1, 2, 3, 4).

The cell counts refer to the number of countries corresponding to each class.

But I dont think I am really doing the correct thing.

After doing the chi test I would get a message in the statistics sofware saying that I need to have values of more than 5 in each cell count for the chi test to be reliable. Should I verify variable dependency with another method?

Should I use correlation better or some other method?