Actually this is not possible to answer since you don't have quantitative data. You can only test which activity has the highest probability to have a higher Likert scale than any of the other activities. This is a slightly different question than the one you asked, but this complication is introduced by nto having the data that would actually have been interesting (i.e., the actual time spent, rather than a rating about the length of the time; even a relatively coarse statement about the actual time would have been much more instructive that an ordinal rating!).

Since the data about different activities is paired within each subject, you can use the sign test to compare two groups of activities. To find which one is the activity with the highest probability of spending longer times you would need to compare all activities (I don't know how many you checked). Since this is a kind of a screening, you would then need to adjust the p-values for multiple testing (you can use Bonferroni-Holm).

However, with just 2 subjects I see a problem. The lowest p-value you can get from a sign test with n=6 is 0.03125. If you have to adjust this for multiple testing, the result cannot reach any conventional level of significance. So I can already conclude that your data is not sufficient to identify the "most attractive" or "most time consuming" activity among 3 or more different possible activities with conventionally acceptable confidence.

Jochen Wilhelm Thank you for your response. I don't understand any of the tests you recommended, but I will find out. I just want to compare the ratings and see which one is biggest.

Sign test: https://en.wikipedia.org/wiki/Sign_test

Adjustment: https://en.wikipedia.org/wiki/Holm%E2%80%93Bonferroni_method

Following