Package testing using likert scale

We recently finished a survey using a likert scale to compare attributes of consumer packaging(sample question below).

The same question was asked of two different consumer packages. Given the likert scale is ordinal

a)is it valid to calculate a weighted mean to compare the responses?
b)is valid to calculate confidence intervals of the weighted means, using non-overlapping confidence intervals to indicate a significant difference in the means.

PS. The responses are non-normally distributed.

C)any other advice on how to approach the analysis?

sameple question: To what extent does this packaging convey each of the following attributes to you?
Strongly(5)---.....----not all all(0)