Dear Jawad,

Your data for each question is independent. You can test each one in turn using a Mann Whitney U test.

I suggest that you construct interval-level scales from your variables, rather than analyzing them one at a time using Mann Whitney. This process will involve calculating Cronbach's alpha for each scale, and probably using Exploratory factor analysis to assign variables to scales.

Because there have been some many questions here about this topic, I have compiled a set of resources:

https://www.researchgate.net/post/How_can_one_help_people_create_simple_scales_from_Likert-scored_items

Jawad, I have been in the same waters in the past.

Though Peter's suggestion is not wrong that you can compare each attribute at at time Mann Whitney Test, I would rather go for the solution suggested by David. If you are able to combine attributes to higher level/order variables (also known as scale of latent variables), it will be much easier to make the comparison. One possible way is to do Factor Analysis. You can find more information on this in the link shared by David and some practical considerations are given at the following:

https://www.researchgate.net/post/Which_of_the_following_is_best_solution_against_Multiple_Factor_Analysis

If you already know that some of the attributes can be grouped under one heading and you have justification for that (theoretical or other), you may combine them to create a higher order variables for both importance and actual. Thereafter, you can compare mean scores of both high order variables to see if they are similar or different significantly, along with effect size.

Good Luck!