Department of Health Management and Policy, University of Michigan, Ann Arbor, MI, USA.
Health Econ. 2013 Aug;22(8):965-86. doi: 10.1002/hec.2873. Epub 2012 Oct 11.
From a policy standpoint, the spread of health conditions in social networks is important to quantify, because it implies externalities and possible market failures in the consumption of health interventions. Recent studies conclude that happiness and depression may be highly contagious across social ties. The results may be biased, however, because of selection and common shocks. We provide unbiased estimates by using exogenous variation from college roommate assignments. Our findings are consistent with no significant overall contagion of mental health and no more than small contagion effects for specific mental health measures, with no evidence for happiness contagion and modest evidence for anxiety and depression contagion. The weakness of the contagion effects cannot be explained by avoidance of roommates with poor mental health or by generally low social contact among roommates. We also find that similarity of baseline mental health predicts the closeness of roommate relationships, which highlights the potential for selection biases in studies of peer effects that do not have a clearly exogenous source of variation. Overall, our results suggest that mental health contagion is lower, or at least more context specific, than implied by the recent studies in the medical literature.
从政策角度来看，量化社交网络中健康状况的传播非常重要，因为这意味着健康干预措施的消费存在外部性和可能的市场失灵。最近的研究得出结论，幸福感和抑郁可能在社会关系中具有很强的传染性。然而，由于选择和共同冲击，结果可能存在偏差。我们通过使用大学室友分配的外生变化来提供无偏估计。我们的研究结果表明，心理健康没有明显的整体传染，特定心理健康措施的传染效应也不大，没有幸福感传染的证据，只有适度的焦虑和抑郁传染的证据。传染效应的强度不能用避免与心理健康状况不佳的室友接触或室友之间普遍较低的社交接触来解释。我们还发现，基线心理健康的相似性可以预测室友关系的亲密程度，这凸显了那些没有明确外生变异源的同伴效应研究中存在选择偏差的可能性。总体而言，我们的研究结果表明，心理健康的传染程度低于（或至少更具体）最近医学文献中研究的结果。
