19.3 Information for Decision

Information for decision refers to the specific subset of available data and knowledge that is relevant to, and capable of improving, a particular decision. Not all information is decision-relevant: an enormous proportion of the data that could theoretically be accessed by a decision maker has no bearing on the choice at hand, and treating all available information as equally pertinent would overwhelm any decision maker's processing capacity. Information for decision is defined by its functional relationship to the decision it supports — by the degree to which it reduces uncertainty about which option is best, improves the accuracy of the model used to evaluate options, or enables the detection of consequences and risks that would otherwise be invisible to the decision maker.

Decision-Relevant Information versus General Information

The distinction between decision-relevant information and general information is functionally derived: the same piece of data may be highly relevant to one decision and entirely irrelevant to another. Information about a competitor's pricing strategy is decision-relevant when choosing a product's price point and irrelevant when deciding the internal organizational structure. Information about a patient's drug allergies is decision-relevant for prescribing medication and irrelevant for choosing the patient's billing account category. The relevance relationship is determined by the structure of the decision problem, not by intrinsic properties of the data itself.

This functional relativity of information relevance has an important implication: information gathering for decision support must be guided by an analysis of the decision's structure. Before seeking information, the decision maker or their support system must identify what the key uncertainties are, which variables most affect the expected outcome of each option, and what data could reduce uncertainty in those specific dimensions. Without this prior analysis, information gathering becomes either indiscriminate (gathering everything and creating overload) or driven by convenience and availability (gathering whatever is at hand rather than whatever is needed).

The Value of Information

In decision theory, the value of information is defined formally as the increase in expected decision quality — measured as expected utility or payoff — that results from receiving the information before making the decision. Perfect information about the true state of the world would allow the decision maker to select the optimal option in every circumstance; the expected value of perfect information is the difference between what could be achieved with perfect knowledge and what the best available decision achieves under current uncertainty.

This formal concept translates into a practical principle: information has value for decision making only to the extent that it changes what decision will be made. If the best available decision is the same regardless of what the information reveals, the information has no decision value — even if it is highly accurate and intrinsically interesting. Decision makers who pursue information that will not change their conclusions are not making better decisions; they are consuming time and resources in information gathering that yields no improvement in choice quality.

Types of Information Needed for Decisions

Different aspects of a decision problem require different types of information:

State-of-the-world information describes current conditions in the environment relevant to the decision. It establishes the baseline from which the decision's effects will unfold and is essential for calibrating any projection of what different options will produce.

Causal model information describes how variables in the decision's domain are causally related — what causes what, which levers produce which effects, how large the effects are, and under what conditions they occur. This type of information enables the decision maker to predict what will happen if each option is chosen. Without adequate causal model information, decisions are based on guesses about causal relationships, and errors in those guesses propagate into systematic option-evaluation errors.

Option inventory information describes the range of choices available to the decision maker. Decisions constrained to an incomplete option inventory may select the best available option but miss a superior option that was never considered. Comprehensive identification of available options is therefore a form of decision-relevant information that is often neglected in practice.

Preference and criteria information specifies the values, goals, and tradeoff parameters that determine which outcomes are better and which are worse. Even accurate causal predictions about what each option will produce cannot guide selection without a clear specification of what the decision maker most values. Clarifying preferences and criteria before evaluating options is a form of information acquisition that significantly improves decision quality.

Constraint information specifies the boundaries within which options must fall — resource limits, legal requirements, ethical constraints, stakeholder expectations, and operational requirements that rule certain options out or impose costs on approaches that violate them.

Information Timing and Decision Readiness

The value of information for decision making depends not only on its content but on its timing. Information delivered after a decision is made has no decision value for that choice, even if it would have substantially improved it. Information delivered too early — before the decision maker has formed enough of a decision model to interpret it — may be received but not properly incorporated into the decision process.

The concept of decision readiness captures the match between information timing and decision timing: information for decision is most valuable when it arrives at the point in the decision process where it can be absorbed, integrated, and acted upon. Organizations that time information gathering to the decision cycle — commissioning analysis before decisions are needed rather than after — improve their capacity to use information effectively.

Information Sufficiency and the Stopping Problem

A pervasive practical challenge in information gathering for decisions is knowing when to stop. More information can always potentially improve a decision, but information gathering consumes time and resources, and beyond a certain point, additional information yields diminishing returns in decision quality improvement. The decision of when to stop gathering information and proceed to choosing is itself a decision — one that must be made under uncertainty.

Practical decision making requires working norms about when information is sufficient: when the major uncertainties have been reduced to manageable levels, when the most decision-relevant data has been obtained, and when the probability of reversing the current best option through additional information gathering is low relative to the cost of further delay. These stopping norms are necessarily imprecise, but they are indispensable for avoiding the paralysis that results from indefinitely deferring decision until perfect information is available — a condition that never arrives.

Information Overload and Decision Quality

When information for decision exceeds the decision maker's processing capacity, quality degrades rather than improves. Information overload produces several pathological decision patterns: selective attention to whatever information is most salient rather than most relevant; cognitive simplification that ignores nuanced data in favor of easily processed summary signals; decision fatigue that reduces the rigor with which successive information elements are evaluated; and premature closure that terminates information processing before it should to escape the overload.

Effective decision support systems manage information load by filtering and summarizing — presenting decision makers with the most relevant information in processed form, structured to match the decision maker's cognitive architecture and the structure of the decision at hand, rather than exposing them to raw data volumes that exceed their processing capacity.