Chance of Showers and Decisions
Today we have a 40% chance of Bob making the wrong decision on his big case today, and a partly cloudy prediction of what he’s going to have for dinner.
Wouldn’t it be nice if we could work out those statistics in our head before the big case, or the test, or our dinner selection? Statistical approaches to uncertain decision making help us better understand when and where to make risky decisions and when to just walk away. Wouldn’t it be nice to just put in some variables and come out with solid numbers, everyone likes numbers, right?
In Hillel J. Einhorn’s article titled, Accepting Error to Make Less Error, he concludes that by using a statistical approach such as our weather related percentage prediction above, numbers do prove to be better than experience. He states that “using a simple rule that is consistently applied” (p. 181) is the most efficient way to make good decisions. Einhorn explores this idea by separating decision making strategies into two main analysis categories: clinical and statistical.
Clinical decision making can be defined as a causal understanding of factors before a decision is made. This “prediction depends on backward inference” (p. 184). To use this strategy you must know all of your options and consider them based on previous experiences. However, these memories may be influenced by, mood, fatigue, anxiety, or hundreds of other factors just waiting to throw off your decision.
On the other hand, statistical analysis accepts error, but “trusts no one and takes little faith.” (p. 186). This approach is based on the uncertainty of the world and simplifying it into models such as a weighted linear regression model, among other options. Comparing both of these decision making strategies, can be difficult, but Einhorn has constructed a simple “decision matrix for comparing the clinical and statistical approaches” (p. 186):
Einhorn believes that the systematic choice and random state of nature is a more clinical than statistical approach, while the random choice and systematic state of nature is a more statistical than clinical approach. He concludes that clinical reasoning has a higher risk value than statistical reasoning, because statistical decision making has shown better outcomes on average.
Text Credit: Dowie, Jack & Elstein, Arthur, Professional Judgment: A Reader in Decision Making, Accepting Error to Make Less Error, pp 181-195. Cambridge, UK, 1988.
Photo Credit: http://www.flickr.com/photos/syrisstudios/436710585/