I have been stuck in a problem recently. I designed a survey, in which the participants observed several images. These images are gathered from different cultures, e.g. Culture A, Culture B and Culture C. For each image, I asked the participants a question: To what extent do you believe this image comes from each culture and three Likert-scales from 1-5 (1 for completely disbelieve, 5 for completely believe) are offered. After gathering the answers, I'd like to do some analysis on the them. Since culture B and C have more similarities with each other than they with culture A, what I'd like to know is, after observing an image, if the extent on believing it from culture B and C are more 'similar' or their distributions show no significant difference. What I am considering is one-way ANOVA (since there are 3 groups) or pairwise t-test (since the variables are somewhat correlated), which is more proper for this issue?
Hope someone could offer me some suggestions. Thanks in advance :-)