In my experiment participants rate their confidence concerning the distance between two stimuli previously presented on a 5 step scale (1 = sure they are near, 3 = don't know, 5 = sure they are far). The stimuli belong to 4 conditions (2x2 repeated measure design; see cond_A, cond_B in table). Here is a summary of the long format dataframe that shows the two repeated conditions, the ID, the proportions (n trial/total for each ID) for that relevant confidence rating.
cond_A cond_B ID 1 2 3 4 5 tot
1 congr far 5b6dada03ce3ab0001fd077c 0.0000000 0.2857143 0.1428571 0.4285714 0.1428571 7
2 congr far 5c5d73cf0edbd90001e1ee27 0.0000000 0.1666667 0.1666667 0.6666667 0.0000000 6
3 congr far 5c5e04ca6539fe00016e1afa 0.0000000 0.0000000 0.5714286 0.4285714 0.0000000 7
4 congr far 5c66f284146af300015604c2 0.1428571 0.1428571 0.0000000 0.4285714 0.2857143 7
5 congr far 5c7d57bb3454d70016499f1d 0.0000000 0.2000000 0.4000000 0.4000000 0.0000000 5
6 congr far 5da0a0ce117b92001465fa35 0.0000000 0.2000000 0.2000000 0.0000000 0.6000000 5
tail of the df
cond_A cond_B ID 1 2 3 4 5 tot
254 noncongr near 60e5cd4e3d91b6025c62b9fe 0.2000000 0.3000000 0.1000000 0.4000000 0.0000000 10
255 noncongr near 60e813572b209b50665a209c 0.5714286 0.1428571 0.0000000 0.0000000 0.2857143 7
256 noncongr near 60e9503c04c84d717015c4b0 0.4000000 0.2000000 0.2000000 0.2000000 0.0000000 5
257 noncongr near 60eafb146471ede7fb3773df 0.7500000 0.0000000 0.0000000 0.0000000 0.2500000 4
258 noncongr near 60eb0d6a4080e2a6cbc9c7d3 0.2222222 0.4444444 0.1111111 0.2222222 0.0000000 9
259 noncongr near 60ef4642202c8520a37e19b5 0.0000000 0.2857143 0.2857143 0.4285714 0.0000000 7
Please find here a plot that summarises the data
From that, I need to extract ROC curves (something like Chapter 3: Empirical ROCs https://rstudio-pubs-static.s3.amazonaws.com/371840_5694e490af85424eb7e7f3ae721cd67d.html#chapter-3-empirical-rocs) but cannot make heads or tails on how to do that.
Do you have any hints on how to run ROC analyses on this data? Or any other analyses that you think might be appropriate for this type of response? Thanks for your help