Analyzing & weighing satisfaction survey data results.

[brnjnsvld]

I have a question for the group, say I'm a restaurant owner that distributes a satisfaction survey to my customers, asking them to rate multiple aspects of their visit on a scale from 1-5 (1 being terrible, 5 being excellent), then say I also ask them how important/significant is each aspect towards their overall experience (let's say also 1-5). What's a good way to combine these two related metrics in order to get some sort of weighted evaluation on each of these aspects?

I was thinking something like: (their satisfaction rating for a particular aspect/5) * an importance multiplayer like 0 if they said 1, 1if said 2, 10 if said 3, 50 if said 4 and 250 if they said 5. I sort of trying to borrow off of the algorithm described below:

https://www.youtube.com/watch?v=m9PiPlRuy6E

Like them, in the end I would like to aggregate all of the individual's responses to a single value per customer.

Thanks!

 

[mmercker]

Hi, I would use multinomiale ordinal regression, where the rating score is the outcome variable, the different aspects (food quality, cleanliness,...) are the levels of a categorical predictor, and the "significance score" are a-priori weights. However, like this you would not get a single value per customer but per aspect... Not sure if this helps