Cognition and Chance - How good are you at making decisions in uncertain conditions?

In preparation for writing a book review for "The Mensa Research Journal," I am reading Cognition and Chance: The Psychology of Probabilistic Reasoning (Raymond S. Nickerson, Lawrence Erlbaum Associates, 2004, 520 pages).  Here are a few of the most interesting conclusions Dr. Nickerson draws from his overview of several decades of research.

1. Once we have a tentative answer in mind, it is easier for us to see data which confirm that tentative answer than those which contradict it.
2. We are lousy at estimating risk.  For example, many people worry more about dying in an airplane crash than in a car wreck, even though the latter is far more likely to occur.
3. Our estimates of time and cost are almost always too low.  We assume the probability that all steps of a project will go according to plan is high, when in fact it is usually very low.
4. We see correlations where they do not exist, particularly if we expect to see the correlation.  This is a bit like artist James C. Christensen's saying that "believing is seeing."
5. We tend to believe we have more control over chance events than we really do.  For example, we believe that lottery numbers we pick ourselves are more likely to be winners than randomly assigned numbers.
6. Information acquired early in an investigation tends to carry more weight than information acquired later on.  Nickerson cites studies, for example, that jurors in mock trials are most heavily influenced by information presented very early in the trial.
7. Sometimes we fail to modify existing beliefs when presented with new evidence.  On the other hand, sometimes we are too quick to change our beliefs based on single, dramatic incidents or data.

What does all of this mean to those of us who are trying to run businesses?  You can probably draw many conclusions, but here are a few of mine:

• Always complete a Potential Problem Analysis on any important plan of action (see #3 above), and never assume all steps will go according to plan.
• Let the data, and statistical analysis, lead you to conclusions about correlation.  Don't trust your intuition to tell you two conditions are related, because your intuition is often wrong.
• Treat data carefully.  Try to ensure that later data carries as much weight as early data.  Try to carefully consider how new data might change your earlier assumptions.  Actively seek data that contradict your original, tentative answer.  Listen to people who have not been involved in your planning or decision making - ask them to help you see where you are misusing new data.

What other conclusions do you draw?