To gain insight into the behavior of the value of information, this paper identifies specific rules for a canonical decision problem: the two-act linear loss decision with normal prior probability distributions. Conditions are derived for which the expected value of perfect information increases when mean and standard deviation are both linear functions of an exogenous variable. A variety of richer decision problems can be adapted to the problem, so that the general results obtained here can be immediately applied to understand drivers of information value.
Keisler, Jeffrey, "Technical note: Comparative static analysis of information value in a canonical decision problem" (2004). Management Science and Information Systems Faculty Publication Series. 44.