I have the same group of participants (n=31,000) that answered the following question "What is your favorite subject?" by selecting one of three answer options (Science, Math, Art) at two different timepoints (Year 1 and Year 2). In the data table, the answers are coded as 1, 2, and 3, respectively, in two columns (one for each year), and every row is a participant. What method do I use to analyze if the subject favoritism differed between Year 1 and Year 2? Chi-squared? McNemar? Repeated measures ANOVA? Cochran Q? Other?
Summary table: Year 1: 30% Science, 42% Math, 28% Art; Year 2: 20% Science, 58% Math, 22% Art
