It may not seem very important, I know,
but it is, and that's why I'm bothering telling you so.
--Dr. Seuss
Introduction
A decision is an allocation of resources.
It can be likened to writing a check and delivering it to the payee. It
is irrevocable, except that a new decision may reverse it.
In the same way that a check is signed by the account owner, a decision
is made by the decision maker. The
decision maker is one who has authority over the resources being allocated.
Presumably, he (or she) makes the decision in order to further some
objective, which is what he hopes to achieve by allocating the resources.
Key distinction: decision vs. objective
Example: To accelerate an R&D program is an objective, not a
decision. To allocate the funds in an effort to accelerate the program
is a decision.
Why it's important: The decision might not succeed in achieving the
objective. One might spend the funds and yet, for any number of reasons,
achieve no acceleration at all.
The decision maker will make decisions consistent with his values,
which are those things that are important to him, especially those that
are relevant to this decision. A common value is economic, according to
which the decision maker will attempt to increase his wealth. Others might
be personal, such as happiness or security, or social, such as fairness.
The decision maker might set a goal for his
decision, which is a specific degree of satisfaction of a given objective.
For example, the objective of the decision might be to increase wealth,
and the goal might be to make a million dollars.
A decision maker might employ decision
analysis, which is a structured way of thinking about how the action
taken in the current decision would lead to a result. In doing this, one
distinguishes three features of the situation: the decision to be made,
the chance and unknown events which can affect the result, and the result
itself. Decision analysis then constructs models,
logical and perhaps even mathematical representations of the relationships
within and between these three features of the decision situation. The
models then allow the decision maker to estimate the possible implications
of each course of action that he might take, so that he can better understand
the relationship between his actions and his objectives.
The three features of a decision situation
At the time of the decision, the decision maker has available to him
at least two alternatives, which
are the courses of action that he might take. When he chooses
an alternative and commits to it (i.e., signs and delivers the check),
he has made the decision and then uncertainties
come into play. These are those uncontrollable elements that we sometimes
call luck. Different alternatives that the decision maker might choose
might subject him to different uncertainties, but in every case the alternatives
combine with the uncertainties to produce the outcome.
The outcome is the result of the decision situation and is measured on
the scale of the decision maker's values. Since the outcome is the result
not only of the chosen alternative but also of the uncertainties, it is
itself an uncertainty. For example, an objective might be to increase wealth,
but any alternative intended to lead to that outcome might lead instead
to poverty.
Key distinction: good
decision vs. good outcome
Example: Someone who buys a lottery ticket and wins the lottery obtains
a good outcome. Yet, the decision to buy the lottery ticket may or may
not have been a good decision.
Why it's important: A bad decision may lead to a good outcome and
conversely a good decision may lead to a bad outcome. The quality of a
decision must be evaluated on the basis of the decision maker's alternatives,
information, values, and logic at the time the decision was made.
Types of decisions
A simple decision is one in which
there is only one decision to be made, even though there might be many
alternatives. An example of this is the very limited consideration of purchasing
collision insurance for an automobile. The decision maker might be interested
only in comparing three alternatives, such as no insurance, insurance with
$100 deductible, or a policy with $500 deductible. If at the same time
we attempt to add another decision, we have created a problem of strategy,
which is a situation in which several decisions are to be made at the same
time. Each of the decisions in the strategy will have different alternatives,
and the decision maker will attempt to choose a coherent combination of
alternatives. For example, if the decision maker is considering whether
to buy a new car or keep his 10-year old one, and at the same time he is
considering the insurance decision, he might compare only two candidate
strategies: keep the old car and not buy collision insurance, or buy a
new car and buy some level of collision insurance.
Key distinction: strategy vs. goal
Example: Launching two new products a year is a goal. Investing in
additional personnel, while at the same time stopping the funding of some
stalled projects, is a strategy intended to lead to that goal.
Why it's important: Strategy describes a collection of actions that
the decision maker takes. The outcome of the actions is uncertain, but
one of the possible outcomes is attainment of that goal.
An important special case of a strategy problem is the portfolio
problem, in which the various decisions faced in the strategy are of
a similar nature, and the decision maker does not have sufficient resources
for funding all combinations of alternatives. An example is an investment
portfolio, in which the decision maker is aware of a good number of investments
he would like to make, but is unable to afford all of them. Especially
in situations like this example, people sometimes address the problem as
one of performing a prioritization
of the various opportunities. If one opportunity is prioritized higher
than another, then, in the case of limited resources, the decision maker
would prefer to invest in the former than in the latter.
Key distinction: decision vs. prioritization
Example: To assert that one would rather fund development project
A than development project B, and project B than project C, is a prioritization.
Actually funding project A is a decision.
Why it's important: A prioritization might be an intermediate step
en route to a decision, and one might even use a prioritization as a tool
to aid in a decision.
Some decisions offer the opportunity to adopt a particular type of alternative
called an option. An option is an alternative
that permits a future decision following revelation of information. All
options are alternatives, but not all alternatives are options.
Key distinction: alternative vs. option
Example: To allocate the resources needed to drill an oil well is
an alternative. To pay money now to reserve the right to drill after geological
testing is done is an alternative that is also an option.
Why it's important: Options, as an important type of alternatives,
have the potential of adding value to a decision situation. A wise decision
maker is alert to that possibility, and actively searches for valuable
options.
Uncertainties
Decision making would be easy if we could predict reliably what outcome
would follow from the selection of which alternative. To this end decision
makers use forecasts, or predictions
of the future, to guide their choice of alternatives. They attempt to predict
the outcome, on all values of interest to the decision maker, associated
with each alternative that might be chosen. For example, the decision maker
may use forecasts of market size, market share, prices and production costs
in order to predict the profits associated with a new product. When the
quantities forecasted are uncertain, forecasters can describe their uncertainty
about these uncertainties using a probability
distribution. A probability distribution is a mathematical form for
capturing what we know about uncertainties, and how confident we are of
what we know. A probability distribution could record, for example, that
the decision maker (or his designated expert) believes that there is a
30 percent chance of a product having less than 10 percent market share
2 years after its launch, and a 60 percent chance of the product having
less than a 30 percent market share. After assigning probability distributions
to each uncertainty, one can examine the uncertainty associated with the
outcomes of the decision situation. For example, given probability distributions
for price, market share, market size, cost, etc., one can determine a probability
distribution for profits.
Frequently in studying a probability distribution one considers its
expected value (or mean), which is the
average of the values one would expect to obtain upon sampling from the
distribution a large number of times. The expected value weights each possible
value by its likelihood of occurring. Another way to summarize a probability
distribution is to consider some representative percentiles, for example
the 10th-, 50th-, and, 90th percentiles, or "10-50-90s"
for short. The 10th and 90th percentiles are possible low and high values,
respectively, for a given uncertainty, set so that there is a 10 percent
chance that the uncertainty, when revealed, will fall below the 10th percentile,
and a 10 percent chance that it will fall above the 90th percentile. Similarly,
the 50th percentile (or median) is that
number such that the realized value, when revealed, is equally likely to
be above as to be below. Given these percentiles, we can test the sensitivity
of the decision to the uncertainties. Here we attempt to answer the question
whether the decision maker would choose a different alternative if he could
know that the uncertainty would have the low (10th percentile) value from
the one that he would select if he had knowledge that it would have the
high (90th percentile) value. This sort of analysis helps to achieve clarity
of action. With clarity of action, the decision maker knows what he
should do, even if he does not know how it will turn out. This is
the aim of decision analysis.
Key distinction: sensitivity
of the decision vs. sensitivity
of the outcome
Example: The profitability of the new product we are developing is
sensitive to the market share we achieve. However it may be the case that,
as long as we achieve a market share within the range of our 10-50-90s,
we would still choose to develop the product. In this case, our development
decision is not sensitive to market share.
Why it's important: Everyone wants the outcome to be as good as possible,
and in that sense might be interested in knowing to what uncertainties
the outcome is sensitive. However, if one is interested in achieving clarity
of action, as opposed to predicting the future, one need only be concerned
about those uncertainties which would change the decision if we could know
in advance how they will turn out.
Outcomes and values
We consider decisions carefully because we care about the outcomes,
whose goodness we measure against our values. The most commonly studied
and discussed value is economic value,
which we assume to be measured in dollars. Given a stream of cash flows
over time, people often use the NPV (net
present value) to describe the current value of future cash flows. The
NPV is a calculation performed on cash flows over time, allowing one to
condense that stream of cash flows into a single number. Decision makers
often use the NPV of profits or cash flows as a measure of the value of
a project. The NPV calculation makes use of the
discount rate, which has several interpretations, but can be thought
of as a factor applied to future income to reflect the fact that it is
less valuable than income received now. It also reduces the impact of future
costs, since costs that can be deferred into the future are preferable
to those that must be paid now.
In thinking about the value of a scenario, it is helpful to distinguish
between direct and indirect
values. Direct values are cash flows directly related to a project,
for example, the profits resulting from the manufacture and sales of a
new product. Indirect values are things that the decision maker values
that are not likely to show up in accounting statements. For example, a
decision maker may experience "pride" or "goodwill" in producing some products
and value such an outcome beyond its direct economic value. These indirect
values could include costs associated with, for example, laying off workers,
or negative impacts on reputation. While some of these indirect values
are intangible, others are tangible but difficult to put a number on. For
example, increases in "goodwill" associated with one product may result
in increased sales of other products though this effect may be hard to
estimate.
Key distinction: direct vs. indirect values
Example: The direct value of a new product might be the current value
of the future cash flow associated with the manufacture and sale of the
product. The indirect value might include effects like increased goodwill
or strategic advantage that come from having the product but are not directly
associated with the manufacture and sale of the product.
Why it's important: Typically, maximizing the NPV associated with
a product is one of the decision maker's objectives. The decision maker
might, however, assign value in excess of a cash flow based NPV, and that
increment might be for what is sometimes termed "strategic value." These
indirect sources of value must be included in the NPVs, if one is to think
appropriately about values. It is better to put a rough value on these
indirect sources (so it can be discussed and evaluated) than to assume
they are worth precisely zero.
Often the decision maker will have values other than economic, and in this
case he will have to make trade-offs
between
values, which are judgments about how much he is willing to sacrifice on
one value in order to receive more of another. For example, in a personal
context, a decision maker may need to make a trade off between hours spent
at work (something he may wish to minimize so as to maximize the time he
spends with his family) and the amount of income he receives.
Risk
As decision makers ponder the possible outcomes of their decisions they
often think about risk, which is the possibility
of an undesirable result. In discussing this, it is convenient to consider
the notion of a risk-neutral decision
maker. Someone who is risk neutral is willing to play the long-run odds
when making decisions, and will evaluate alternatives according to their
expected values. For example, such a decision maker would be indifferent
between receiving $1 for certain and an alternative with equal chances
of yielding $0 and $2, since this is the average amount that the alternative
would yield if repeated many times. While an insurance company may evaluate
individual policies as if it were risk-neutral, for alternatives with substantial
risks, decision makers are often risk
averse, which means that they value alternatives at less than their
expected values. To make this definition of value precise, we define the
certain equivalent, (or certainty equivalent) of an alternative
as the amount that the decision maker would be indifferent between (1)
having that monetary amount for certain or (2) having the alternative with
its uncertain outcome. For example, a risk-averse decision maker might
have a certain equivalent of $500,000 for an alternative with equal chances
of yielding $0 and $2,000,000, even though the expected value for this
alternative is $1,000,000. In thinking about risk aversion, it is important
to remember that different decision makers have different attitudes toward
risk. While this gamble with equal chances of yielding $0 and $2,000,000
may be very risky for me, a billionaire or a big company may not view these
stakes as large and may have a certain equivalent close to the expected
value.
Decisions where risk aversion holds can be analyzed using a utility
function, which encodes a decision maker's attitude toward risk taking
in mathematical form by relating the decision maker's satisfaction with
the outcome (or "utility" associated with the outcome) to the monetary
value of the outcome itself. These utility functions can be indexed by
their risk tolerance, which is
a technical term describing the decision maker's attitude toward risk.
The greater the decision maker's risk tolerance, the closer the certain
equivalent of a gamble will be to its expected value. The risk tolerance
is a mathematical quantity that describes the decision maker's attitude
towards risk; it is not the maximum amount that the decision maker
can afford to lose, though generally decision makers with greater wealth
will have larger risk tolerances. The decision maker needs to think about
his risk tolerance only in cases where the stakes are large and he is not
comfortable basing his decision on the expected monetary value.
Key distinction: certain equivalent vs. expected
NPV
Example: The certain equivalent for a risky new product is the smallest
sum of money for which the decision maker would be willing to sell rights
to that product. The expected NPV for the product is the hypothetical average
NPV from numerous independent launches of identical projects.
Why it's important: Most projects cannot be repeated and even if
they could, when the stakes are large, most decision makers value gambles
at less than their expected values. Precisely how much less than their
expected value depends on the decision maker's attitude towards risk. This
attitude towards risk varies from decision maker to decision maker and,
even for a specific decision maker, may vary over time.
Index
Alternatives | Certain
equivalent | Clarity of action | Decision
| Decision analysis | Decision
maker | Direct values | Discount
rate | Economic value | Expected
value | Forecasts | Goal
| Good decision | Good
outcome | Indirect values | Mean
| Median | Models | NPV
| Objective | Option | Outcome
|
Portfolio | Prioritization
| Probability distribution | Risk
| Risk tolerance | Risk-averse
| Risk-neutral | Sensitivity
| Sensitivity of the decision
| Sensitivity of the outcome
| Simple decision | Strategy
| 10-50-90s | Trade-offs
| Uncertainties | Utility
function | Values
Author's Note
This short narrative is intended to provide, in non-technical terms and in a readable way, definitions of the words most
commonly used in a decision making context. I have written this because I have found that these words are frequently used
incorrectly. In a sense, misunderstanding language is more insidious than not understanding it at all, in that it distorts meaning.
Distortion of meaning, with its attendant confusion of thought, deters us from the goal of good decision making, which is clarity
of action, the right action. I hope you can read this in 10 minutes or less and benefit from it.
Several colleagues have contributed appreciably to the quality of this document, and I would like specifically to mention Bob
Clemen, Don Keefer, Craig Kirkwood, Bob Nau, and Jim Smith. The reader has no way to gauge their contribution, but I do.
Thanks, Dr. Tom Spradlin,
Confident Choices .
About the Author
Tom Spradlin started his career in the pharmaceutical industry in the International Medical Department of C.H. Boehringer Sohn in Ingelheim, Germany in 1969. After completing his Ph.D. degree in Biostatistics at the Johns Hopkins University in 1976 he joined the statistics department of Eli Lilly and Company in Indianapolis, Indiana. He spent 13 years in that position, serving as principal statistician on many regulatory submissions, particularly in anti-infectives and insulin products.
After three years as a project manager in Lilly’s International Medical component, Tom became interested in decision consulting, and was one of the charter members of Lilly’s Decision Sciences department. In his ten years in that role he consulted on a wide variety of decision situations, especially in research strategy and operations, litigation, manufacturing, and business development.
Tom retired from Lilly in 2001 and formed Confident Choices (www.confidentchoices.com), a consultancy dedicated to assisting clients facing difficult decisions to make the right choice efficiently and with confidence. Tom is a former member of the Council of the Decision Analysis Society of the Institute for Operations Research and Management Science, and is a founder of the Decision Analysis Affinity Group, a collection of industrial decision consultants from the USA, Canada, Europe, and Japan.
Citation
Spradlin, T., "A Lexicon of Decision Making", DSSResources.COM, 03/05/2004.
Tom Spradlin provided permission to feature this article and archive it at DSSResources.COM on Tuesday, December 02, 2003. A version of this article is posted at the Decision Analysis Society website (http://faculty.fuqua.duke.edu/daweb/lexicon.htm). You can contact Tom by phone: 317-844-0636 and email: confidentchoices@indy.rr.com. This article was posted at DSSResources.COM on March 5, 2004.