JUDGMENTAL
FORECASTING
Biases, etc.
Judgmental Forecasting

The statistical forecasting methods presented in the
text allow us to extrapolate established patterns
and/or existing relationships in order to predict
their
continuation,
assuming
that
such
patterns/relationships will not change during the
forecasting phase.
Judgmental Forecasting

At the same time, because changes can and do
occur, these must be detected as early as possible
to avoid large, usually costly, forecasting errors.
Judgmental Forecasting

However, when changes are detected, or if we can
know when they are about to occur, human
judgment is the only viable alternative for
predicting both their extent and their implications on
forecasting.
Judgmental Forecasting

Human judgment is also needed to incorporate
inside information and knowledge, as well as
managers’ experience, about the future.
Judgmental Forecasting


Before using our judgment for improving forecasting
accuracy, however, we must understand its biases
and limitations along with its major advantages.
Doing so allows us to combine the information from
our statistical predictions with those of our judgment
by exploiting the advantages of both while
avoiding their drawbacks.
Judgmental Forecasting


We make innumerable forecasts every day, but
expend little effort in evaluating them to find ways
of improving their accuracy. The reason is simple:
we do not want to be held responsible if our
forecasts go wrong.
However, unless we get feedback about the
accuracy of our predictions, it is not likely that we
can improve our performance when making similar
forecasts in the future.
Judgmental Forecasting

Because judgmental forecasts are much more
common than statistical ones, not only can we not
ignore them, but we must also be willing to accept
that judgmental forecasting errors cannot be
entirely avoided; we will be better off if we can
accept such errors while learning as much as
possible from them so we can improve our ability to
forecast more accurately in the future.
Judgmental Forecasting

The accuracy of judgmental forecasts is, on
average, inferior to statistical ones. This is because
our judgment is often characterized by considerable
biases and limitations.
The nature of judgmental biases and
limitations


We rarely do anything to remedy the deficiencies
of our judgment, mainly because we are unwilling
to accept that our judgment can be faulty or
biased.
Because judgmental biases are almost never
presumed to exist, it is extremely important to
expose them: empirical evidence clearly
demonstrates their existence and their negative,
damaging consequences.
Bias

The entire subject of judgmental biases could take
many volumes to treat thoroughly. (See Kahneman
and Tversky, 1979). We focus here on those
aspects of judgmental biases that most critically and
directly affect forecasting.
Bias

Inconsistency: being unable to apply the same
decision criteria in similar situations
 Formalize
 Create
the decision-making process
decision making rules to be followed
Bias

Conservatism: failing to change (or changing slowly)
one’s own mind in light of new information/evidence
 Monitor
for changes in the environment and build
procedures to take actions when such changes are
identified
Bias

Recency: having the most recent events dominate
those in the less recent past, which are downgraded
or ignored
 Realize
that cycles exist and that not all ups or downs
are permanent
 Consider
the fundamental factors that affect the event
of interest
Bias

Availability: relying upon specific events easily
recalled from memory to the exclusion of other
pertinent information
 Present
 Present
complete information
information in a way that points out all sides of
the situation being considered
Bias

Anchoring: being unduly influenced by initial
information which is given more weight in the
forecasting process
 Start
 Ask
with objective information (e.g., forecasts)
people to discuss the types of changes possible;
ask the reasons when changes are proposed
Bias

Illusory correlations: believing that patterns are
evident and/or two variables are causally related
when they are not
 Verify
statistical significance of patterns
 Model
relationships, if possible, in terms of changes
Bias

Search for supportive evidence: gathering facts that
lead toward certain conclusions and disregarding
others that threaten them
 Induce
disconfirming evidence
 Introduce
role of devil’s advocate
Bias

Regression effects: persistent increases (or
decreases) might be due to chance rather than a
genuine trend
 One
needs to explain that if the errors are random, the
apparent trend is unlikely to continue
Bias

Attribution of success and failure: believing success
is attributable to one’s skills while failure to bad
luck, or someone else’s error. This inhibits learning
as it does not allow recognition of one’s mistakes
 Do
not punish mistakes, instead encourage people to
accept their mistakes and make them public so they and
others can learn to avoid similar mistakes in the future.
(This is how Japanese companies deal with mistakes, in
general.)
Bias

Optimism, wishful thinking: people’s preferences for
future outcomes affect their forecasts of such
outcomes
 Have
 Have
forecasts made by a disinterested third party
more than one person independently make the
forecasts
Bias

Underestimating uncertainty: excessive optimism,
illusory correlation, and the need to reduce anxiety
result in underestimating future uncertainty
 Estimate
uncertainty objectively.
Consider many
possible future events by asking different people to
come up with unpredictable situations/events
Bias

Selective perception: seeing problems in terms of
one’s own background and experience
 Ask
people with different backgrounds and experience
to independently suggest solutions.
Conventional wisdom versus
empirical findings

Another type of judgmental bias that can threated
decision-making effectiveness is unfounded beliefs
or conventional wisdom.
Conventional Wisdom

The more information we have, the more accurate
the decision.
 The
amount of information does not improve the
accuracy of decisions, instead it increases our
confidence that our decisions will be correct.
Conventional Wisdom

We can distinguish between useful and irrelevant
information.
 Irrelevant
information can be the cause of reducing the
accuracy of our decisions.
Conventional Wisdom

The more confident we are about the correctness of
our decision, the more accurate our decision will be.
 There
is no relationship between how confident one is
and how accurate his or her decision is.
Conventional Wisdom

We can decide rationally when it is time to quit.
 We
feel we have invested too much to quit, although
the investment is a sunk cost.
Conventional Wisdom

Monetary rewards and punishments contribute to
better performance.
 Human
behavior is too complex to be motivated by
monetary factors alone.
Conventional Wisdom

We can assess our chances of succeeding or failing
reasonably well.
 We
are overly optimistic and tend to downgrade or
ignore problems and difficulties.
Conventional Wisdom

Experience and/or expertise improves accuracy of
decisions.
 In
many repetitive, routine decisions, experience and/or
expertise do not contribute more value to futureoriented decisions.
Conventional Wisdom

We really know what we want, and our preferences
are stable.
 Slight
differences in a situation can change our
preferences (e.g., most people prefer a half-full to a
half-empty glass of water).