The use of multi‐level modelling in risk research. A secondary analysis of a study of public perceptions of genetically modified food

Many studies have examined the relationship between various individual variables and people's perceptions of genetically modified (GM) food. A problem with this type of research is that contextual factors are completely ignored. This article explores the use of multi‐level modelling in the field of risk research, by re‐analysing a recent British study of public perceptions of GM food. As the study employed a multi‐stage sampling strategy, it could be used to examine simultaneously the individual and spatial variation in trust and the acceptability of GM food. While the geographical variation in acceptability was largely due to compositional differences between the sampling points, a geographical variation in trust remained after controlling for individual differences. The analysis demonstrated that city‐dwellers commonly have more trust in the regulation of GM food than other respondents. Next to being associated with a number of socio‐demographic variables, both acceptability and trust were related strongly to voting intention. Moreover, the results suggest that there is a link between vulnerable groups, feelings of exclusion, and (dis)trust. The article is concluded by arguing that multi‐level modelling provides new opportunities for simultaneously examining the individual and contextual basis of public perceptions of controversial risk issues.