Living conditions, interviewer effects and perceived well-being of the elderly. A multiple correspondence analysis approach