If you’re a reader of the Nudge blog, you know all about Walter Mischel’s marshmallow test. (References to it are popping up in media everywhere these days.) Maybe you are curious how you might have done on the test if it had been you and one of his assistants at the Stanford lab decades ago. Would you have had the self-control to hold out for the second marshmallow?
If you watch many TV shows on Hulu, you might have already gotten a clue.
Hulu streams ads during its programs. It uses three advertising models, Standard, Premium, and something called Hulu Exclusive Format. When you watch shows under the third format you are presented with a choice before you start viewing. You are offered the option to watch a “long-form commercial” up front in exchange for a showing of your program without additional ads. Or, you can watch your program with “normal commercial breaks.” In other words, you can start watching sooner, or you can wait a bit and experience the pleasure of uninterrupted viewing.
If you watch enough shows under this format, you quickly get a sense of how long a “long-form commercial” runs. The Nudge blog watched one this week that lasted 90 seconds. But if it’s your first time encountering this advertising format, you are a bit uncertain about the commercial’s length. 1 minute? 2 minutes? More? You also aren’t necessarily sure whether the long-form commercial is going to be shorter or longer than the total ad time in the program with “normal commercial breaks.” Do you have to experience more commercial pain up front to get the enjoyment of subsequent commercial-free viewing?
It seems, at least, that is not the case. When the Nudge blog watched the same 22-minute program under both formats, the normal commercial breaks ended up running 15 seconds longer than the single long-form commercial at the beginning. In fact, even under the normal format, Hulu forces you to watch a 30-second commercial before your program starts. The test would be more interesting if you actually got to start watching right away, or if Hulu allowed sponsors to push the boundaries of “long-form.” How pleasurable is uninterrupted viewing? What’s tempting enough to get you to forsake it?
Hulu also employs a bit of choice architecture to nudge you toward the long-form commercial. Of the two boxes shown at the start of the program, the long-form commercial box is already checked. And there is a 15-second countdown clock, after which, if no choice has been made, the long-form commercial kicks in. So the long-form commercial sponsor clearly wants you to exhibit a bit of self-control for their message’s sake.
Does the patience of Hulu viewers mirror the results of Mischel’s test? Only Hulu knows. For now, it hasn’t said anything.
Enviromedia produces a new social norms ad about teen smoking in Texas.
The ad follows a similar structure to the “15 and falling” campaign about teen smoking in Canada by combining a descriptive norm message.
So do social norms ad targeted at teens need a dash of teen humor to make them stick?
The Nudge blog asked Enviromedia creative director Doug Irving for a peek behind the curtain about the ad’s origins. Doug said his team wasn’t aware of the “15 and falling ad” campaign, but definitely drew from the lessons of social norms work.
We knew from research (our own (qualitative research) and (the) Texas Dept. of State Health Services’ (quantitative research)) that peers are the biggest influencers among teen smokers. We’re also familiar with the power of social norming as a behavior-change and myth-busting technique. So when we came across a stat that so vividly illustrates how smoking has become a fringe habit, bingo…
That’s the science. As far as the art, we know you have to present your compelling fact in an entertaining way that cuts through the eight zillion other advertising messages teens have learned to tune out in a typical day. Humor isn’t the only way, but when it’s on the money, it’s hard to beat.
When our creative teams share their initial raw ideas, I’m searching for something that’s got the right balance between memorable, repeat-viewing-worthy weird and still delivering that message. Anything that drifts toward finger-wagging or adult versions of “cool” never makes it to the client meeting. If it makes everyone laugh the fifth time you present it, that’s a good sign.
In this case, spots involving a falconer, a coughed-up lung and the shushing foot seemed to hit the sweet spot. Some additional testing with teens in the target confirmed the foot as the winner. And here we are today.
A new PNAS paper “Using Implementation Intentions Prompts to Enhance Influenza Vaccination Rates” provides evidence of the power of plan-making in getting a flu vaccine shot (Full copy here. Hat tip: Prashant Srivastava). In an experiment, random sets of employees received a mailing that asked them to write down the date they planned to be vaccinated or the date and time they planned to be vaccinated. The first plan-making prompt didn’t work. The second did.
Vaccination rates increased when these implementation intentions prompts were included in the mailing. The vaccination rate among control condition employees was 33.1%. Employees who received the prompt to write down just a date had a vaccination rate 1.5 percentage points higher than the control group, a difference that is not statistically significant. Employees who received the more specific prompt to write down both a date and a time had a 4.2 percentage point higher vaccination rate, a difference that is both statistically significant and of meaningful magnitude.
Public health officials may be the ones most likely to read this study, but marketers at pharmacies should see a possible business opportunity here. When a customer arrives to pick up a prescription, a pharmacist could easily offer a flu shot, give the customer a piece of paper, ask them to make a plan, and tell them to take that plan with them and post it at home on their calendar.
Shape Up! 30 reps before it stops buzzing. Good news, the clock only weighs 1.5 pounds. Bad news, even at a pec-tearing pace, that’s 15-20 seconds of buzzing annoyance (that excludes the time needed to prepare for your reps).
Note: The following post is a revised and expanded version of an earlier one on the paycheck cycle.
On the eve of each new month, a consumer ritual unfolds at Walmarts around the country.
At around 11 p.m. “customers start to come in and shop,” Walmart’s CEO of U.S. Business Bill Simon told a conference of investors last year (pdf of transcript here). Shoppers fill their carts with staples. Baby formula, milk, bread, and eggs. They browse until midnight when their government electronic benefits cards activate. Walmart’s dead-of-night sales zoom well above its daily average over the month.
Retailers have long known about this phenomenon, commonly called the “paycheck cycle,” in which cash-strapped consumers make big purchases when they get paid and are forced to cut back to the bone later in the cycle until the next paycheck arrives. In this tough economy, said Simon, the paycheck cycle is “extreme.” It can affect Americans at all income levels, but at the end of the month, that extremity is most crushing to the poor and the working class.
Among economists, Pay-as-you-Drive (PAYD) automobile insurance has been an intriguing idea for a number of years. The idea is straightforward: Drive more, pay more. Proponents say PAYD can lower the social costs of driving—think carbon emissions and traffic congestion—by leading people to shoulder the true burden.
On the industry side, Progressive has been one of the most aggressive innovators with PAYD insurance. It first began offering a voluntary PAYD program called MyRate in six states in 2008. Three years later, thanks to inexpensive, if not quite cheap wireless technology, the idea is now available nationwide through an initiative called Snapshot. Progressive gives a discount to policyholders based on information about their driving habits collected over a month-long period. Drivers put a sophisticated little tracking device in their cars for six months. At the end of the first month, certain “good drivers” may be eligible for a discount of up to 30 percent based on that “snapshot” of driving behavior. (At the end of the six months, Progressive uses all the data to calculate a renewal rate.)
As Steven Levitt notes, the interesting part of PAYD insurance programs, which are voluntary, has always been how to structure the incentives for participation. The concern is that only low-mileage drivers will sign up.
The clearest winners are (low-mileage users), who can drive the same distance they used to drive and pay less. What’s less obvious is whether Progressive will be a winner; there are, in fact, a couple of situations in which Progressive could lose out. If all (PAYD) accomplishes is to give Progressive’s low-mileage customers the rate cut they deserve, then Progressive is doing little more than lowering its own revenues. It could, of course, try to compensate by raising rates on all its high-mileage (drivers), but then there’s nothing to stop (them) from buying…insurance elsewhere. (Of course, losing its riskiest customers to other companies might also prove profitable for Progressive.)
Where Snapshot is most likely to give Progressive an edge is if it can help the company better forecast the future liabilities it will be on the hook for. That means better predicting accidents. To do that, however, Progressive needs to get all kinds of drivers, not just low-mileage users, to try it and feed the data back to home base. To appeal to all drivers, Progressive has had to move beyond traditional economics and consider behavioral economics as well.
To encourage customers to test out Snapshot, Progressive draws from the lessons of overconfidence and loss aversion. Humans are overconfident creatures, especially when it comes to driving. Ninety three percent of people think they are above average drivers. Even among bad drivers, accidents are not the modal driving experience, and there’s always someone else to blame for them.
Snapshot is primarily a program about usage, but it’s sold as a program about “good driving.”
On its web site, Progressive doesn’t shy away from calling Snapshot “usage-based insurance.” But its media advertisements headlined by cheery customer service rep “Flo,” refer to driving behavior more generally. “Just plug it in and it keeps track of your good driving habits,” Flo explains. “So the better you drive, the more you save.”
What makes a good driver? The speed you drive at and the locations you drive to and from are not part of the equation. Taking speed off the table eases some customer fears since speeding is the most salient mark of good (or bad) driving. Taking driving location off the table eases fears of those who worry about privacy and whether Progressive is tracking their every move. The device doesn’t have a GPS.
Instead, “good driving” depends on 1) Time of day you drive, 2) Number of miles you drive, and 3) How hard you brake (“bad drivers” brake hard). What’s common about all three items? Drivers have very little control over them. When and how far your drive depends largely on where you live and work, which are both quite sticky in the short term. And how hard you brake is primarily an automatic reaction built up over years of commuting. For any single braking experience you can push the pedal more smoothly, but maintaining that consistency over an entire month if you are a hard braker is a bit like trying not to blink for a minute.
According to the Wall Street Journal, a Progressive executive admitted that one of these items predicts accident potential as well as all the usual demographic markers like age, gender and marital status. Although the executive didn’t say which one, the peak time for accidents is between midnight and 4 a.m. If there is one habit worth trying to change over your month with Snapshot, it could be the itch to party. At least think about taking a taxi or hitching a ride with your friend.
Progressive doesn’t think overconfidence, by itself, is enough to get people to try out Snapshot. So it has to remove any risk of losses by promising drivers they won’t pay any more for insurance than they are paying now. Freed of potential losses, overconfidence can kick in, prompting “good” and “bad” drivers to give it a shot and see what kind of a discount they get. After all, they can’t be penalized for any “bad” behaviors. But since Snapshot is about “good” and “bad” driving risk, not “good” and “bad” driving skill, Progressive doesn’t even care if you are involved in an accident while using the device since your risk profile is determined by calculations based on driving patterns. (Theoretically, you could be involved in a minor fender bender with Snapshot and Progressive would never know.)
In essence, Progressive doesn’t want a snapshot of your driving habits. It wants a snapshot of its overall customer base, assuming that base looks a lot like the overall car insurance market. It wants Snapshot to work less like a camera and more like a random sample. From that perspective, the success of economists’ preferred auto insurance depends on whether some non-economists can get the behavioral economics of offering the product just right.
Congestion isn’t just a highway problem. It happens on subway systems, too. Zhan Guo, a professor of urban planning and transportation policy at NYU, says that part of this congestion stems from how transportation authorities draw maps of their systems. In his paper, “Mind the Map” (working paper pdf here), Guo looks at the London Underground system and argues that the Tube map “has a tremendous impact on a passenger’s perceptions and his or her usage of the transit system” since “passengers often trust the Tube map more than their own travel experience on deciding the ‘best’ travel path.” Maps are more than two times as influential as actual travel times, even for experienced transit users. The lesson, Guo says, is that maps themselves can become cost-effective ways to ease planning and operation problems.
Transit maps are not scaled models of physical reality. In the London Underground map, for instance, the correlation between actual distance and transit map distance is just .22. Distance in the Underground map represents just 4 percent of the variation of the actual spatial distance between stations.
This distortion affects travelers’ perceived options of final locations, route choices, and the attractiveness of different routes. The biggest misperception is conflating transit map distance with actual travel time. Transfer stations that look “convenient” can be mobbed by crowds, leading to long wait times. Train trips that look “long” on a map can actually be reached more quickly on foot. Ultimately, travelers “(mis)trust a transit map more than their actual experience; they often take a path that looks shorter on the system map but is longer in reality compared with alternative paths.”
Here is one particular example on the Tube:
There are two alternative paths traveling from Paddington to Bond Street station, path 1 transferring at Baker Street and path 2 transferring at Notting Hill Gates (Map a). Path 2 is about 15% slower by in-vehicle time than path 1, and the Notting Hill Gate station is to the opposite direction of the destination Street (Map b). We would expect that few passengers would choose path 2. However, more than 30% of passengers chose path 2, probably because, on the schematic tube map, path 2 is about 10% shorter than path 1, and Notting Hill Gate station is shown to the south not west of Paddington (Map b).
How should planners adjust maps recognizing the influence they have on travelers’ choices?
There are lots of options. They could redraw maps to better reflect actual distance or actual travel times. They could simply post travel times or draw attention to popular, meaning crowded, transfer stations. Planners could take the focus off individual stations and onto areas. They might note “hot” zones where crowds often form. Travelers who don’t mind a bit of inefficiency in their routes might be willing to take “quieter,” if longer, trips – especially if they have a better chance at scoring a seat. Alternatively, planners might actually remove maps and drive customers to other travel planning options, such as a mobile phone, that can direct them to the most efficient route when travelers type in information about their trip. If the map is always going to influence travelers’ decisions, maybe the best strategy is to lessen travelers’ reliance on any visual version of it?
Addendum: Tim Waters observes: “Its amazing how close everything is. When you are in the tube, the distances seem so much further. I remember a few years ago getting the tube between Leicester Square and Covent Garden! The tube map although great for planning journeys really does distort distances.” Hat tip: Martin Delaney
Imagine a set of 65-year-old identical twins who plan to retire this summer after long careers. We’ll call them Dave and Ron. They have worked for different employers and have accumulated retirement benefits worth the same amount in dollars, but the benefits won’t be paid out the same way.
Dave can count on a traditional pension, paying $4,000 a month for the rest of his life. Ron, on the other hand, will receive his benefits in a lump sum that he must manage himself. Ron has a lot of choices, but all have consequences. For example, he could put the money into a conservative bond portfolio and by spending the interest and drawing down the principle he could also spend $4,000 a month. If Ron does that, though, he can expect to run out of money sometime around the age of 85, which the actuarial tables tell him he has a 30 percent chance of reaching. Or he could draw down only $3,000 a month. He wouldn’t have as much to live on each month, but his money should last until he reached 100.
Who is likely to be happier right now? Dave or Ron?
The Nudge blog is the online companion to Richard Thaler and Cass Sunstein’s “Nudge: Improving Decisions About Health, Wealth, and Happiness.” Here you’ll find much more about nudging, choice architecture, libertarian paternalism, and many other terms you won’t read about in standard economics books.
Cass Sunstein is currently the Administrator of the White House Office of Information and Regulatory Affairs and has no affiliation with the Nudge blog.
The Nudge blog is edited by John Balz.
Tell us about a nudge
The possibilities for great nudges are everywhere. For a list of favorites from the book, check out our dozen nudges. We invite readers to send their own nudge suggestions to firstname.lastname@example.org.
What is Choice Architecture?
Decision makers do not make choices in a vacuum. They make them in an environment where many features, noticed and unnoticed, can influence their decisions. The person who creates that environment is, in our terminology, a choice architect. The goal of Nudge is to show how choice architecture can be used to help nudge people to make better choices (as judged by themselves) without forcing certain outcomes upon anyone, a philosophy we call libertarian paternalism. The tools highlighted are: defaults, expecting error, understanding mappings, giving feedback, structuring complex choices, and creating incentives.
For a user-friendly introduction to choice architecture, check out this paper.