Do the sunk cost effect and cognitive dissonance increase risk perception? An empirical study in the context of city smog