R help

#1
I want to analyse the change over time for 2 groups (over 35 years), each group in each year has been recorded for 6 categories.
I want to see if there is a difference in the frequency recorded in each of these categories over time for each newspaper, and to compare between newspapers. I know I can use the Chi squared test for homogeneity, but I want to see in what direction the change takes place if there is a difference; I am interested to know if there is a shift from A to F over time (the categories A to F are not on a scale that is equal, but they are on a scale which goes to a certain direction).

The A, B, C etc. refer to certain key words used to refer to animals in the newspaper articles, which I have tallied up and changed into (100%) percent frequencies for each year (row) per newspaper type.

So I am basically interested to know if newspapers have changed their attitude (the terms act as a proxy) to animals.

For example this is how the data looks like

Liberal newspaper Conservative newspaper
% A | % B | %C | %D | % E | %F % A | % B | %C | %D | % E | %F
1994 1 15 25 28 20 11 3 12 26 27 18 13
1995 10 15 26 37 13 8 8 11 24 20 33 4
1996 4 11 24 20 33 8 8 15 26 37 13 10


Can anyone help me to understand what statistical test I should run in r?
 
Last edited:

gianmarco

TS Contributor
#2
Hello,

from an exploratory point of view, you could use simple Correspondence Analysis in this case. I happened to come across something like that in relation to the analysis of newspaper headlines.

You would need to reshape (or recode) your data a bit.

For instance, you may want to build a cross-tabulation with each row corresponding to a given year and an indication of what type of newspaper; also, you need the counts instead of %. Columns would correspond to your categories A, B, C, etc.

See attached images to make up an example.


Best
Gm
 

Attachments