Johnson Branden B
Bureau of Risk Analysis, Division of Science, Research and Technology, New Jersey Department of Environmental Protection, P.O. Box 409, Trenton, NJ 08625-0409, USA.
Risk Anal. 2003 Aug;23(4):767-80. doi: 10.1111/1539-6924.00354.
Despite many claims for and against the use of risk comparisons in risk communication, few empirical studies have explored their effect. Only one study, published by Roth et al. in this journal in 1990, has tested the 1988 predictions by Covello et al. as to the public's relative preferences for 14 kinds of risk comparisons as they might be used by a factory manager to explain risks of his ethylene oxide plant. That study found no correlations between the Covello predictions and seven different measures of "acceptability" of Covello's examples of each type of comparison. However, two critics of the Roth study, as well as its own authors, suggested that a scenario involving local risks, a conflict-ridden situation, and a plant manager unknown to the townspeople might better evoke Covello-like preferences than the distant, calm, friends-involving scenario used by Roth. The research reported here replicated the Roth study using the same scenario, risk comparison examples, and evaluation measures, and added a second scenario intended to replicate the conditions suggested by critics. Over 200 New Jersey residents answered the study questionnaire. The replication scenario reproduced Roth's results, and the conflict scenario also evoked no rankings correlated with Covello's predictions. Furthermore, neither agreement nor disagreement with five statements representing "conflict"--respondents' reports that the industrial-plant scenario made them angry, they lived near industry, they were concerned about industrial risks, people in their home town were angry about industrial pollution, and they worried "frequently" about long-term effects of pollution--correlated with Covello's predictions. Over half of all ratings ascribed to the comparisons in aggregate were positive, and most detailed comments offered by respondents also were positive, despite many criticisms and suggestions for their improvement. The wide variability in individuals' rankings also undermines the notion of any single ranking of preferred comparisons. These findings have implications for use of risk comparisons, but also reveal the inaccuracy of the field's assumptions about public reaction to industrial risk information, including risk comparison.
尽管对于在风险沟通中使用风险比较存在诸多支持和反对的观点，但很少有实证研究探讨其效果。只有一项由罗斯等人于1990年发表在本期刊上的研究，检验了科维洛等人在1988年做出的关于公众对14种风险比较的相对偏好的预测，这些风险比较可能会被工厂经理用来解释其环氧乙烷工厂的风险。该研究发现，科维洛的预测与对每种比较类型中科维洛示例的七种不同“可接受性”衡量指标之间没有相关性。然而，罗斯研究的两位批评者以及该研究的作者都认为，与罗斯所采用的遥远、平静且涉及朋友的情景相比，涉及当地风险、充满冲突的情景以及城镇居民不认识的工厂经理的情景，可能更能唤起类似科维洛所预测的偏好。此处报告的研究使用相同的情景、风险比较示例和评估方法重复了罗斯的研究，并增加了第二个情景以重现批评者所建议的条件。200多名新泽西居民回答了研究问卷。重复情景重现了罗斯的结果，冲突情景也未产生与科维洛预测相关的排名。此外，代表“冲突”的五条陈述（即受访者报告工厂情景使他们生气、他们住在工厂附近、他们担心工业风险、家乡的人对工业污染感到愤怒以及他们“经常”担心污染的长期影响）的同意或不同意程度，均与科维洛的预测无关。总体而言，超过一半的对这些比较的评级是积极的，受访者给出的大多数详细评论也是积极的，尽管有许多批评以及对改进的建议。个体排名的广泛差异也削弱了存在任何单一首选比较排名的观念。这些发现对风险比较的使用具有启示意义，但同时也揭示了该领域关于公众对工业风险信息（包括风险比较）反应的假设的不准确之处。
