Incorporating structural models into research on the social amplification of risk: implications for theory construction and decision making