Interview: Dan Goldstein on Decision Science
Decision Science News is a website about decision research in Psychology, Economics, Business, Medicine, Law & Computer Science. Dan Goldstein is the editor of Decision Science News.
What are the biggest misconceptions business and technology people have
about the science of decision–making?
Dan Goldstein: Business people interpret decision making with two extreme views. One view is that optimal solutions exist and that expensive software can get a company closer to them. The other extreme is that the best intuitions come from the gut, without decision aids. Both views are misinformed.
The optimizers fail to realize that the space of problems that have optimal solutions is rather small, and that optimal solutions are easy for competitors to imitate. But tough problems have no optima.
What are the optimal acoustics for a concert hall? What's the optimal speed to drive to work? No one can say. The intuiters, in contrast, think that nothing beats instinct. This is long been proven wrong. In many domains, simple linear models (a basic form of decision aid) can outperform even the intuitions of experts.
Can you list some key decision science books for business managers to read, and explain why they are important?
Dan Goldstein:
- Rational Choice in an Uncertain World – Reid Hastie & Robyn Dawes – Yes, it’s a textbook, but it is the best introduction to the field
- The Black Swan – Nassim Taleb – To drive home the lesson that the world is less predictable than our models suggest
- How To Lie With Statistics – Darrell Huff –To help realize that “the data” are never univocal
- Gut Feelings: The Intelligence of the Unconscious – Gerd Gigerenzer – To learn when intuition works and doesn’t
- Meditations – Marcus Aurelius – To keep everything in perspective
Can technology design improve through the study of decision science, and if
so how?
Dan Goldstein: I don’t know if decision science can improve technology design, but technology design could certainly improve decision making. Good decisions need knowledge, and knowledge has been, until recently, hard to come by.
Remember back in the 80s when we had to slug to the library, make photocopies, or spend the day on the phone to gather information? Now the internet connects us to information very quickly, but the problem is now the reverse, there's simply too much. What I see as a primary challenge for technologists is getting the right information to decision makers in the form that they need it. Google is the world champion at this, but there is a long way to go.
Google gives you rank–ordered search results, but it takes specialist sites to break information into fields that are useful. Right now you can’t ask Google to show you two bedroom apartments within 3 miles of the West Village with dishwashers (well you can, but don't expect much more than agency advertisements). However, if the world were tagged up properly, all manner of devices could return you the same tables of information you now have to pay for. But we can’t (and shouldn’t) expect people to tag everything up. An automated way to add metadata to the world would be darn useful to people making decisions.
Is there any software out there today you would consider outstanding
in its implementation of decision science?
Dan Goldstein: Bill Sharpe, Eric Johnson and I have created a cognitively–friendly tool to help people express their preferences in a complex domain: that of investments. We grounded the tool in psychological principles. For example,
instead of bombarding people with probabilities, we allow people to deal with simple counts, or frequencies. Instead of expressing these quantities with numbers, we express them graphically.
This allows people who don’t know the first thing about probability theory to communicate what sorts of risks they are willing to take by expressing them in a purely visual way. We have a paper on it, and think it could really change how investing is done.
What is one thing businesses can do today to improve decision making?
Dan Goldstein: Companies make lots of mistakes when it comes to hiring people. They tend to put too much weight on interviews and not enough on past performance. There are some in my field who recommend dispensing with interviews altogether. I have found over a number of years that simply testing people on the job they are to perform is a much better predictor than betting on their “promise,” yet amazingly few companies do this.
Resources
- In Weird Math of Choices, 6 Choices Can Beat 600 “Research subjects who were asked to select from an extensive array of alternatives … were less satisfied with their choices, found the choices themselves less attractive, and felt more frustrated and regretful than other subjects who were given only a limited number of options to choose from.”
- The article How Do Managers Think? relates management to how doctors reach a diagnosis.
- Understanding the ‘Want’ vs. ‘Should’ Decision examines the two decision states and shows how you design for one or the other to create sales.
- High–Stakes Decision Making: The Lessons of Mount Everest is really about the system surrounding and influencing high–stakes decisions. Uses the Everest tragedy as the contextual framework for understanding how to diagnose and prevent serious decision failures in any organization.
- An article about Festinger’s theory of cognitive dissonance explains, “Harmon-Jones believes that dissonance is essentially about action, specifically about everyday decision making and choices and commitments. In his view, it's not just abstract “cognitions” in collision.”
- People are persuaded by that which is quantifiable, making it a primary front of information warfare. Playing With Numbers explains the phenomenon when you mix human nature, high–stakes decisions and the U.S. News and World Report college rankings.
- PLAYING BY THE NUMBERS helps you understand the decision science of behavioral economics using baseball as a backdrop. The article covers Hindsight Bias: How good decisions can get you fired, and why people make defensible decisions in favor of good decisions.
- Mental accounting theory explains Why It’s Easy to Blow the Tax Refund and Hard to Catch a Cab in the Rain.
- 26 Reasons What You Think is Right is Wrong covers over two dozen common cognitive biases, from the bandwagon effect to zero–risk bias.
- Days that shook the world explains the black swan theory at the core of The Black Swan: The Impact of the Highly Improbable by Nassim Taleb.
- What You Don’t Know About Making Decisions compares and contrasts advocacy and inquiry. Problems occur when advocacy crowds out inquiry, or when inquiry becomes an end unto itself (paralysis by analysis).
- Offering airfare predictions based on historic data not unlike stock forecasts is Farecast. With such an insight into how humans reach decisions, Farecast qualifies as an information technology.
- Mathematics, marriage and finding somewhere to eat explains the 37% rule. After having seen 37% of houses, restaurant reviews, prospective mates or job applications, “a coherent picture of the ideal employee is built up and the next person to fulfil these criteria gets the job.”
- Programming and the Sunk Cost Fallacy is one argument for spending more time studying decisions and humans.
- MapsKreig is a Web 2.0 style mashup of CraigsList and Google Maps. It provides an interface to view Craigslist apartment listings visually using the Google Maps API.
- Business Week design guru Bruce Nussbaum explains how pointless it is to hire, measure or fire individuals. Instead Nussbaum recounts the advice of Bob Sutton: HR should concentrate on hiring, evaluating and even firing networks of people. And while Nussbaum takes the why we hate HR view, the basic premise is to redesign the HR value proposition. Looking at the HR ‘product’ as users and stakeholders do, business managers can start thinking like interaction designers.
- “Following the crowd may have a pay off even if the chosen course of action fails. …If a bad outcome occurs, but the action was consistent with approved conventional wisdom, the hit to the manager’s reputation is reduced. …When a manager follows a management consultant’s advice in selecting a human relations system, he is buying insurance against a bad outcome.” Management fads: A behavioral explanation. Professor Stephen Bainbridge looks beneath herd behavior to the payoff: decision insurance.
- Miller and Hartwick give us the anatomy of a fad: Simple, Prescriptive, Falsely Encouraging, One–Size–Fits–All, Easy to Cut–and–Paste, In Tune with the Zeitgeist, Novel, not Radical, and Legitimized by Gurus
- Pfeffer and Sutton make a critical contextual observation, “Organizations can have amazingly good evidence, but it has no effect on the decisions they make if it conflicts with their ideology.” Consequently they suggest management adopt the techniques of evidence–based medicine.
- Game theory and the Traveler’s dillemma produces some unexpected but interesting insights into decisionmaking because people consistently reject the ‘rational’ choice, which produces a better result.
- 200% of Nothing: An Eye Opening Tour Through the Twists and Turns of Math Abuse and Innumeracy and Lying with Statistics explains there are several varieties of information: Information, Disinformation, Misinformation and Noninformation. To qualify as information technology a computer system design must acknowledge and address this — and almost none do.