Читать книгу Figure It Out - Stephen P. Anderson - Страница 54

Whereas priming suggests a concept that somehow correlates with present or future concepts, anchoring need not exhibit such a correlation. With anchoring, introducing an irrelevant number biases your judgment. Adding a completely arbitrary factor—say the last two digits of your Social Security number—affects an unrelated piece of information, say the estimated value of a bottle of wine or a trackball mouse. In fact, this is precisely what Professor Dan Ariely and others did in a classroom experiment.¹⁴

After being shown an array of objects—wines of different values, computer equipment, books, boxes of chocolate—students were asked to do several things.

• First, translate the last two digits of their Social Security numbers into a dollar amount. For example, if the last two digits of your SSN were 37, then you would write this as a dollar amount: $37.

• Next, students wrote this dollar amount next to each object, and were asked if they would pay this amount for each object. Yes or no. Students were reminded that this was a completely arbitrary number and should have no correlation with the actual price value of the object.

• Finally, students were asked to write the maximum amount they would be willing to pay for each object.

Here is where things got interesting. The assessed value of these objects showed a strong correlation with the last two digits of the students’ SSNs. Students with Social Security numbers in the lowest 20%, priced (on average) a bottle of ’98 Côtes du Rhône wine at U.S. $8.64; students on the other end of the bell curve, the 20% with the highest Social Security numbers priced (on average) the same bottle at U.S. $27.91.

While completely arbitrary, the number written down appears to influence, or “anchor,” the value attribution. This is the basic idea of price anchoring, that when asked to make a numerical estimate where we have little information to go on, we’re often biased by some initial starting number. This happens in everything from auctions to retail price settings. If a retailer prices an object at a high dollar amount, then any discount off that amount will seem to be a good deal. Conversely, this same fixed amount might seem like too much, if the initial anchor were set much lower.

In his book Predictably Irrational, Ariely comments on this price anchoring research, stating:

Social Security numbers were the anchor in this experiment only because we requested them. We could have just as well asked for the current temperature or the manufacturer’s suggested retail price. Any question, in fact, would have created the anchor. Does that seem rational? Of course not.

While we’re focused here on one popular study, anchoring is a well-studied phenomenon, with research going back to at least the 1950s by researchers such as Muzafer Sherif.

On their own, these studies were remarkable. In the context of this chapter, though, anchoring should be less remarkable: it’s simply an association. By suggesting a detail—even a detail that may have little or nothing to do with the subject—we introduce extra information into our conscious (or unconscious) mind. This detail becomes part of the web of related associations that our brain has to sort out, in a never-ending job as a perceptual organ. The brain doesn’t necessarily care that information is unrelated—it’s all information to reconcile into a sensible pattern, if there is one. In economic terms, we might describe this as arbitrary coherence, where the brain attempts to maintain a decision consistent with previous decisions (coherence), even if the initial decision was arbitrary.

While the majority of studies investigate anchoring with numeric values, there are exceptions, studies that explore other—non-numeric—forms of anchoring. In a study of “cross-modality” anchoring, researchers wanted to push the boundary conditions for anchoring, to assess at what point a piece of irrelevant information would no longer influence an individual’s judgments.

With this study, students were given a survey with a series of unrelated questions, one per page. Among the decoy questions was an activity that required students to draw a series of either short or long lines. Later, one of the questions would ask students an estimation question, such as the length of the Mississippi River or the average temperature in Honolulu in July. The results? Researchers found a strong correlation between being primed with drawing long or short lines and the river length or temperature estimates, concluding that the “boundaries of anchoring effects might be much wider than previously thought.” Anchoring might work not just with irrelevant numbers, but also irrelevant shapes, sounds, temperatures, weights, and so on.

In their conclusion, researchers speculated on other ways that anchoring might be manifest:

Students filling out class evaluation forms might have lower evaluations of the class if the forms are completed with golf pencils rather than regular pencils. The fact that the taller candidate won 80% of the presidential election contests between 1904 and 1980 (Gillis, 1982) could be due partly to the fact that their heights biased people’s estimates of their positive qualities. Waiting in a long line to get tickets at the theater may bias people to then think that the cost of theater popcorn is not exorbitant. This is hardly an exhaustive list; clearly, there are many far-reaching applications and implications of these findings.¹⁵

Of course, our interest in all this is to show how we reason and think via associations—even with irrelevant associations. While there are worthwhile distinctions between anchoring bias, priming effects, decision framing, and other named effects, these distinctions seem to blur into each other, coalescing around this more fundamental idea of associations.

Let’s turn them to another form of associative thinking, also triggered through words and language, but one that is much more robust, developed, and explicit—that of narratives.