quantile regression with a bounded dependent variable

Quantile regression is cool. But I wonder about using it when the dependent variable is bounded. I often work with happiness variables, emerging from a questionnaire where respondents select from a 0 - 10 scale. Running a quantile regression, it seems that pretty much every independent variable has a smaller "impact" (coefficient) on people in the higher quantiles. But perhaps this is simply a consequence of having a bounded scale? People in the higher quantiles have smaller prospects for moving further up the scale...

Is there a more general view about the advisability of using quantile regression in this sort of context?