You don't get such numbers by asking people 'what would you prefer' - that would give wrong answers even w/o lying, since many of those things are such where sociologists know that the believed preference (what you think you'd choose) differs from the true preference (what you do choose in reality).

But you get useful results if you (a) ask people how happy they are (even if they exaggerate and understate, the ranking is generally accurate), and (b) ask people if they're getting X in the company, and then measure the correlations (and do a bunch of tricky adjustments for factors that are interrelated).

I.e., you don't ask "is a fair boss important for you? are office conditions important for you?" - but, if for example, on average the people who think they have a fair boss are feeling happier than those who think they have an unfair boss; but those who think that the office conditions really suck are just as [un]happy that those who feel that the office is okay - now, that's useful signal.