Civil Engineering Reference
In-Depth Information
OVERCONFIDENCE IN DIAGNOSIS (KLEINMUNTZ, 1990)
This applies to most judgments that we make in daily life. Ask a football fan of
Manchester United what he thinks is the probability that Manchester will beat Arsenal.
Although statistics of recent football matches show a probability of 0.6, he may say 0.9.
OVERESTIMATION OF SMALL NUMBERS AND
UNDERESTIMATION OF LARGE NUMBERS
It is difficult to judge small probabilities. A driver may expect that the probability of a
police speed control is 0.02 (2%). In reality, speed controls are much rarer, but since we
think of them as being significant events, the number is exaggerated.
SUMMARY
To summarize, we use heuristics all the time. In most cases, we don't even reflect on how
a decision was made, and we are not aware of the particular heuristics we adopt to deal
with events in real life. The use of heuristics increases our ability to deal with decisions
in real time and to arrive at reasonable solutions to everyday problems. Sometimes there
may be unwanted effects and very poor decisions, as illustrated in the case study below of
the Three Mile Island accident.
EXAMPLE: CONFIRMATION BIAS AT THREE MILE ISLAND NUCLEAR
POWER PLANT
At the time of the nuclear power plant accident at Three Mile Island in the U.S., operators
focused on a display that indicated that a relief valve had closed. This turned out to be
wrong information: the relief valve was in fact still open. They then searched for
confirming evidence that the water level was too high, although it was actually too low.
In so doing, their attention was diverted from many contradictory factors, and the
hazardous situation kept building up until it became a disaster (Wickens and Hollands,
2001). There are three reasons for the confirmation bias:
1. People have cognitive difficulties in dealing with negative information.
2. To change a hypothesis requires effort.
3. The final formulation of the decision becomes a self-fulfilling prophecy. We start off
with a cautious attitude and keep building up the confirming evidence.
The main challenge is to make decision makers consider disconfirming evidence. This
is a very difficult problem to solve in a nuclear power plant. Perhaps it would be possible
to emphasize disconfirming information by using different displays. Such displays could
express the semantics in the situation, summarize various decision making alternatives,
and keep the alternatives alive until the final decision is made.
