In this paper we use data from an SP study on flood safety in the Netherlands, and elicit individual preferences for reduction of risk to life and limb. We perform context analysis to test the robustness of fatality risk valuation throughout choice experiments. The main interest of this paper is VOSL sensitivity to the valuation of correlated risks (scope effect). Besides, we explore the role of cognition on the stability of valuation across choice experiments using age and education. We pool data from multiple choice experiments and apply nested and mixed logit models in our analysis. We confirm statistically significant sensitivity to scope, comparing VOSL estimates for the test group in a choice experiment where correlated risks were present (risks of fatality, injury and evacuation) to an experiment where only fatality risk is valued. We find that the origin of differences in VOSL valuations across the choice experiments lies in differences in age and educational attainment, and may therefore be related to cognitive abilities of respondents. In particular, we conclude that higher VOSL sensitivity to scope is most prominently present among respondents of senior age (65 and older) and respondents without college education. This finding has important implications for discrete choice modeling and the use of obtained values in cost-benefit analyses.