You meet someone special. Things go well. Your thoughts turn to the future. You’re in love, but you know you ought to think this through. If you get married, will it last?
People usually think forecasting has to do with weather, economics and politics. But forecasts are the foundation of every decision we make. We bet that an MBA degree will be worth the student-loan debt because it will pay off in the long run. We switch careers because we think a new one will be more rewarding.
These sorts of forecasts matter a lot to us, so we want them to be as accurate as possible. Unfortunately, for most of us, they aren’t.
Imagine that young man who’s thinking about buying an engagement ring. How will he decide whether the relationship is likely to last for life?
He thinks about the woman he loves. He imagines her smile. His heart flutters. But he’s more analytical than most and he knows he can’t let his feelings decide. So he thinks carefully about his girlfriend’s personality and his own. He contemplates her hopes for the future and compares them to his. He imagines how life would be if they woke up to each other day after day.
Finally, he makes up his mind. He can’t say the marriage is guaranteed to succeed. But it is extremely likely. It’s time to buy that ring.
So how accurate is his forecast?
It’s probably off by a substantial margin. The real probability of the marriage succeeding is lower than he thinks.
The culprit here is “optimism bias.” As psychologists have repeatedly demonstrated, when we estimate the likelihood of an event, whether or not the question is personal makes a big difference. If people are asked how likely it is that a new business will fail, they’ll give one answer. But if they’re asked, “How likely is it that your new business will fail,” they’ll give a much more optimistic answer.
Even thinking carefully about all the challenges his relationship will face isn’t likely to correct the bias. We all tend to uncritically accept information that supports what we believe, while hyper-critically dismissing information that does not. In fact, thinking long and hard about your relationship may only serve to strengthen your confidence that your faulty judgment is reliable.
There is a better way to approach forecasting.
Earlier this year, the Intelligence Advanced Research Projects Activity, a US government intelligence research agency, wrapped up a four-year forecasting tournament. One of its goals was to find outstanding forecasters and determine why they are better able to see what’s coming.
The program succeeded. With the help of more than 20,000 volunteers, a small number of truly excellent forecasters were identified—people dubbed “superforecasters.”
Not surprisingly, superforecasters are cautious, analytical thinkers. Like our young man contemplating marriage, they know they can’t trust feelings and intuitions, and they think carefully before drawing conclusions. But unlike that young man, they do that hard thinking methodically.
For example, superforecasters know that people tend to immediately focus on the particulars of a specific case when making a judgment. That’s sometimes called the “inside view.” Here, the inside view is the young man’s girlfriend, her personality and hopes, and the other details mentioned above. Superforecasters do consider the inside view—but only after a careful analysis of the “outside view.”
The outside view is the larger class to which this particular case belongs. Here, the outside view would be, “What percentage of marriages end in divorce?”
Or rather, that’s one outside view. It’s not a particularly good one, because it is so broad. Superforecasters would refine it.
Researching the subject, they would discover that there are large and growing disparities in divorce rates. One key factor is education. The more educated a married couple is, the less likely they are to divorce. If our young man and his girlfriend both have graduate degrees, a good outside view would be, “What percentage of marriages where both spouses have graduate degrees end in divorce?” It might even be possible to refine that outside view further with other relevant factors, like age, income, and the length of the relationship prior to marriage.
This will produce a hard number—a cold, clinical percentage—uninfluenced by feelings and biases.
Superforecasters would then turn to the inside view, the particulars of this case, and adjust the number up or down accordingly. If, for example, the young man’s girlfriend was so averse to commitment that she hated to lock into a cell phone plan, that might move the number lower.
This approach won’t eliminate psychological biases, logical mistakes, and other sources of error. But it can reduce them, resulting in more accurate forecasts.
Admittedly, this may sound a little Vulcan—particularly when the subject matter in question is love and marriage. But remember that forecasts inform decisions. And decisions shape our lives.
A little Vulcan can do us all a lot of good.