1

In an experiment, I wanted to check the effect of image redness on subjects' choices of an image. I modified the images at three redness levels (I: Increased redness, N: Normal, and D: Decreased redness). I presented all the pairs to different subjects (I vs N, D vs N, and D vs I) and instructed them to choose the one that seems more appealing to them.

From the box plots with notches, I can visually see the effects, but I am looking for a proper statistical index to compare these conditions.

enter image description here

I have already used a t-test (number of selections per condition and comparing each pair), and I used Freidman Tests too. But I am not sure if this is the right way to do it. Could please introduce a proper test or documentation guide that I can go through?

kjetil b halvorsen
  • 77,844
Majid khalili
  • 111
  • 2
    How many individuals? What is the Y scale on your plot? What is the response, some numerical, or just which image they preferred? Where the comparisons shown in randomized order? See https://stats.stackexchange.com/questions/83005/how-to-calculate-ratings-rankings-from-paired-comparison-pairwise-comparison-o, https://stats.stackexchange.com/questions/208376/pairwise-comparison-with-bradley-terry – kjetil b halvorsen Jan 10 '23 at 17:51
  • 1
    Also, your title is very generic, you could see more interest by editing the title to describe your specific experiment – kjetil b halvorsen Jan 10 '23 at 18:23
  • 1
    @kjetilbhalvorsen Thanks a lot, actually the Bradley-Terry model is exactly what I was looking for – Majid khalili Jan 12 '23 at 20:18

0 Answers0