I have multiple Likert-Scale variables representing a common dimension (eg, 5 statements value labeled from 1 to 7 for the Reliability dimension, 4 statements about Responsiveness, 4 about Tangibles etc.).
I asked the respondents of my questionnaire to sort the 5 dimensions with perceived importance order (eg, 1=least important ... 5=most important),
so now I have a total sum of importance under each dimension (eg, 179 for Tangibles, 452 for Reliability ...) for a grand total score of 1635 [(1+2+3+4+5)*109 responses].
It turns out that Reliability is the most important and Tangibles the least important, with the other 3 dimensions in between.
My question is what's the correct 'statistical' way to weight my variables with that data.
I don't mind creating new weighted variables but I don't know how to calculate the weight coefficient to multiply them.
I could possibly convert them to percentage and multiply this number with my Likert-Scale variable value but something tells me this is completely wrong.