What's luck got to do with it?
A conversation with Sam Bodily
A popular TV game show may not seem like a serious tool for teaching future business leaders how to make better decisions, but that is what "Deal or No Deal" has become for the University of Virginia's Darden School of Business.
With the help of an anonymous donor, first-year master's of business administration students in the decision analysis class participated in a game to pick "Darden's Luckiest Student." The winner, Hideki Inoue, won the opportunity to pick one of two briefcases. One briefcase held $17,500, the other had none. Unfortunately for Inoue, his choice came up empty.
But don't say he chose poorly, said professor Sam Bodily, who helped design Virginia's game. Even if Inoue had chosen the case with the money, it wouldn't have meant his decision was good or bad.
The point of the exercise, Bodily said, is to teach students that there is always an element of randomness in life. A good decision-making process includes understanding that good decisions can still have bad outcomes.
Bodily recently spoke with Washington Technology Editor Nick Wakeman about risk and decision-making.Q: What do you hope the students learned from the "Darden's Luckiest Student" contest?Bodily:
It is about risk preference. They were facing a 50/50 lottery, zero or $17,500, or they could take an amount the banker offered. How do you make that trade-off?
They had to complete a questionnaire and state what offer they would accept. Inoue wrote $8,000, which was reasonable because the certain equivalent, or average value of the two briefcases, was $8,750. Others would have taken less than $8,000. A few students chose more than $8,750, which is risk seeking. That's not behavior you see in many business managers, so we talk about that.
We also discuss how the banker figures out what to offer, assuming the banker is trying to conserve funds to use for another year. That's a competitive bidding problem.
You do this kind of thing when bidding on various jobs, particularly when it is not particularly clear what it is going to cost to do the job, and it isn't clear what other bidders might bid.Q: What are you learning from games such as "Deal or No Deal?"Bodily:
Researchers are interested in shows such as "Deal or No Deal" because there is no skill involved. It is purely risk preference. My hypothesis is that people who are consistent in the way they decide between certain and risky propositions make more money on that show.Q: What makes a consistent decision-maker?Bodily:
You choose a level of risk tolerance and put that into your decision model and use it every time.
What I've observed in "Deal or No Deal" is that there are people who turn down an offer that is above the average value of the briefcases left unopened. Then later, they will accept one that is well below the average value. Those people tend to make less money.Q: What are the components of a good decision?Bodily:
There are four steps: What is the sequence of decisions and uncertainties? What are the payouts? What are the probabilities of the payouts? And then you put all that together.
In class, we use the example of making a movie. You have to decide to make the movie, and then you have to decide about making a sequel.
For the analysis, you have to look ahead to where you may end up and do the analysis backward. You have to think about the kind of downstream decisions you'll have to make and work backward to where you first decide whether to make the movie.
You can make a decision diagram that reflects choices and uncertainties and then the follow-on choices and uncertainties. You look at the payouts for all the possible paths.Q: How does the magnitude of a decision affect the process?Bodily:
If I'm standing in the buffet line and deciding between soup and salad, I don't draw a decision diagram. But if you get into the mind-set of thinking through the consequences of choices, you don't have to.
Depending on the nature of the problem, you can structure a decision diagram and do a first-cut analysis.
Sometimes, you have choice A, and you look at the range of outcomes and look at choice B and look at the range of outcomes, and B doesn't overlap A. You don't need to go any further. You can make the decision.
The ones you spend more time on are the ones that overlap. You want to ask why. What are the various components? Sometimes, you can find yourself with a spreadsheet of everything that relates to the decision, and you try to sort out the items that matter and the ones that don't.