Here is an image from the New York Times:
This will be the first US label that contains a gas consumption metric ("gallons per 100 miles").
As the labels were debated last fall, one of my favorite features was a new number: The savings or costs of gas compared to an average vehicle calculated over 5 years. I'm very glad they kept this feature. It is behaviorally insightful in two ways.
First, it harnesses the power of "loss aversion," which is the idea that coming up short of some comparison point is painful--and people are motivated to get rid of the feeling of loss. In this case, the comparison is an average car. When a low MPG car is compared to the average car, the difference in gas costs will be in the thousands of dollars. For example, if the comparison is between a 16 MPG vehicle and a 22 MPG vehicle the difference in cost would be close to $4,000. Avoiding that cost will be attractive to some.
Second, it harnesses the power of "scale expansion." By expressing the cost over 5 years (not 1 year), the magnitude of the savings and costs are larger and more salient.
Loss Aversion
The idea of loss aversion comes from a theory of decision making proposed by Amos Tversky and Daniel Kahneman called Prospect Theory. One of the core ideas in Prospect Theory is that people value outcomes by comparing them to salient "reference points." Reference points are often the status quo (e.g., one's current car), but can shift. In this case the label offers a comparison with the average car. This is similar to the idea OPower has used when it provides feedback to electricity customers on their neighbor's average energy use. (Also, see this paper in Psychological Science by Schultz et al. in 2007.) Being below the average is unpleasant, and people are motivated to close the gap.
For a brief description of the Prospect Theory value function, see this article called Raising the Bar on Goals published the University of Chicago's Graduate School of Business. It summarizes research published in this Heath, Larrick, & Wu (1999).
Scale Expansion
Numerical information, such as gas consumption or costs, can often be expressed in different units: Per day, per week, per month, per year. For example, a donation to NPR can be thought of as $100 per year or 27 cents per day. Changing the time period (or more generally, scale) does not change the substance of the information. It is the same cost regardless of whether it is framed as per year or per day. However, increasing the denominator increases the magnitude of the numerator, and makes the amount psychologically more salient. The same costs seem more important when expressed per year than per day.
In this 2009 paper (Burson, Larrick & Lynch, 2009; also, see this forthcoming paper by Pandalaere and colleagues in the Journal of Consumer Research), we showed that people are more sensitive to an attribute (such as cost) when it is expressed on an expanded scale. In our first study, we had people consider cell phone plans that had two attributes: Rate of dropped calls and price.
- In Condition 1, dropped calls were expressed per 100 calls and price was expressed per year
- In Condition 2, dropped calls were expressed per 1,000 calls, and price was expressed per month
People expressed their preferences for one of two plans (A or B) that differed in dropped calls and price. In Condition 1, Option A was described as having 4.2 dropped calls per 100 calls and at a price of $384 per year. In Condition 2, Option A was described as having 42 dropped calls per 1,000 calls at a price of $32 per month. Fundamentally, both descriptions are identical--only the scale changed.
People's preferences shifted systematically with the expanded attribute. The numbers on the right show that when price was expanded (and number of dropped calls contracted) in Condition 1, only 31% favored the option higher in price (Option A). However, when dropped calls were expanded and price was contracted in Condition 2, 69% now favored the option higher in price.
Why do people change their preferences? We argue that differences become clearer when they are large. And larger differences motivate people to choose the option favored by the large difference.
Is scale expansion potentially a trick? It can be if the denominator is manipulated in an unrealistic way (costs per 100 years). But if the denominator is scaled to something practical and meaningful, such as costs over five years, it is a helpful tool. It gives people accurate, relevant information that they might not calculate otherwise.
More Behavioral Ideas
For a summary of behavioral ideas applied to energy metrics, see this talk which I recently gave at the annual meeting for the Center for Research on Environmental Decisions (CRED).
