Methodology

Expected utility

Assumptions

Actors have fixed, complete and transitive preferences over a set of outcomes.

• Completeness: All pairs of consequences are compared
• Transitivity: This means that preferences do not contradict themselves If you like A more than B, and you like B more than C, then it would be inconsistent to say you like C more than A. Transitivity says instead, that you must like A more than C.

Calculation

The expected utility for action $a_1$ is calculated by multiplying the probability of each outcome with its utility. In other words, the utilities are weighted by their probability and summed up.

$$EU(a_1)=p(c_1|a_1)\ast u(c_1)+p(c_2|a_1)\ast u(c_2)+p(c_3|a_1)\ast u(c_3)$$

• Three outcomes ($C_{1,2,3}$)
• $p$ = probabilities for the respective outcomes
• $u$ = utility of an outcome

Example: Utility of voting

Causal Inference

Confounding Bias