To analyze a new safety standard's social utility, Section 1.1 divides that utility according to people the standard affects and according to goals those people have.When the new standard's outcome is uncertain, it is also helpful to divide its utility for a person according to its possible outcomes, perhaps, either a reduction or no reduction of workplace injuries. That utility is a weighted average of personal utilities of its possible outcomes, where the outcomes' probabilities supply the weights. This type of analysis is called expected utility analysis, because the weighted average is called an expected utility. It adds a new dimension to utility analysis, a dimension of possible outcomes.

This chapter presents a precise formulation of expected utility analysis. The new formulation incorporates several improvements over standard formulations. In particular, it de.nes an option's possible outcomes so that an agent's knowledge of their utilities requires only introspection. It also carefully attends to necessary idealizations, adopting and elaborating those Section 1.3.2 sketched. The next chapter defends and generalizes my version of expected utility analysis. Later chapters combine expected utility analysis with intrinsic utility analysis and group utility analysis.

My investigation of expected utility analysis assumes that expected utilities govern decisions. First, expected utilities generate options' utilities.