Respondent characteristics' moderating effect on mixed-mode effects: an example of subjective well-being measurements