Temptation bundling works when the treat is locked away: the gym-locker study

In the one experiment that tested it, temptation bundling, pairing something you crave with something you keep avoiding, worked only when the craving was available nowhere else, and only for a while. In the autumn of 2011, researchers at the University of Pennsylvania and Harvard assigned 75 people to a group whose page-turner audiobooks lived on an iPod in a locker at the gym entrance. In the first week, those people went to the gym 51 percent more often than a control group (p. 291). The people simply asked to keep their audiobooks for the gym gained about half as much, and that could have been chance; and after one holiday break the effect was gone (pp. 291, 294).

Eighty-two books you cannot put down

The researchers asked 233 undergraduates for "the five novels you have read in the last several years that you found it the most difficult to put down", and turned the answers, topped up from bestseller lists, into 82 audiobooks averaging nearly twelve hours each; The Hunger Games trilogy was the most chosen (p. 289). Temptation bundling, in the authors' words, couples "instantly gratifying 'want' activities" with "a 'should' behavior that provides long-term benefits but requires the exertion of willpower" (pp. 283 to 284).

Then 226 members of a university community who already used the gym's cardio machines, mostly undergraduates, were assigned at random to three groups (p. 286). The locker group got four novels of their choosing on a loaned iPod that had to stay in a monitored locker at the gym entrance whenever they were not checked in. The own-iPod group got the same novels on their own device, with a rule to try: "only let yourself enjoy these novels while exercising". The gift-card group got a $25 bookshop voucher, pretested as worth the same (pp. 287 to 288).

The locker group went about one and a half times as often, at first

In the first week after the intervention began, the locker group made 0.48 more gym visits a week than the gift-card group, "a 51% increase over the regression-estimated 0.94 visits for the control group"¹1 The 0.94 is a model's estimate after adjusting for how often people went before the study, not a raw count. ↩ (p = .004)²2 p is roughly how likely a gap this big would turn up if the locker did nothing; below .05 is the usual bar for calling a result real. ↩ (p. 291; Table 3, p. 292). In plain terms, about one extra visit every two weeks. Across the seven weeks before Thanksgiving, by a coincidence of numbers, an average of 51 percent of them visited at all in a given week, against 42 percent of the gift-card group (p. 290).

The own-iPod group, asked to police themselves, gained 0.27 visits a week at first, a difference that could have been chance (p = .092), and 0.14 over the seven weeks, which could too (pp. 290 to 291; Table 3, p. 292). The authors' conclusion: "individuals are limited in their ability to self-impose temptation bundling tying rules" (p. 294). The locker, not the suggestion, carried the result.

And then it faded

The locker group's advantage fell by 0.07 visits a week with every week that passed (p = .005); at that rate the whole first-week gain would be gone in about seven weeks (p. 291; Table 3, p. 292). Averaged over all nine weeks, with no allowance for that decline, the group's lead of 0.21 visits a week is not statistically significant (Table 3, p. 292). The gym closed for the Thanksgiving break, and in the two weeks after it the locker group visited less than the gift-card group did: 0.75 visits against 0.99, then 0.47 against 0.67 (Table 2, p. 290).

The authors' reading is that the break "eliminated our intervention's effectiveness", because the pull came from wanting the next chapter, and a forced spell without it "led the temptation bundled with exercise to lose its allure" (p. 294). Their proposed fix is rewards for re-engaging with the tempting content after holidays (p. 294). For scale, the two earlier pay-to-exercise studies the authors compare against raised later attendance by between 0.26 and 0.59 visits a week, and this cost a $15 reconditioned iPod and audiobooks from a dollar, or free from a library (p. 295).

Would you pay to have your own iPod locked away?

At the outtake visit, participants were asked what they would pay for a real month of the locker programme the next semester. Sixty-four percent called it appealing, the average offer was $6.91, and 61 percent offered at least a dollar (pp. 293 to 294; Figure 4, p. 294). They were buying the right to have their own iPod taken away from them.

The twist is who bid most. The gift-card group, who had never used the locker, offered $9.87 on average; the locker group offered $6.01 and the own-iPod group $4.77 (p. 293; Table A.2, p. 297). One gift-card participant bid $100; drop that person and their group's average is $8.59, still the highest, and the locker group's gap holds while the own-iPod group's gap shrinks to one that could be chance (Table A.2, p. 297). The authors could not tell whether the untested overvalued it or the tested undervalued what it had done for them (p. 293).

What it suggests trying

1. Pick a want and a should that are really yours. The authors concede that exercise is not a chore for everyone, nor a page-turner a treat for everyone (footnote 1, p. 284).
2. Take the rule out of your own hands. The audiobook lives on a device that stays where the task is, or someone else holds the login; the result in this study came from a staff-checked locker, not from a promise.
3. After a lapse, get hooked again before you expect the pull to return. That is close to the authors' own suggestion, rewards for re-engaging with the tempting content after holidays; the "before you expect it" part is ours (p. 294).

A worked example: the podcast you keep saving lives only on the old tablet that stays on the desk with the tax folder, and after the holidays you start it again with the episode you most want to hear.

Where Yappie fits

We make Yappie, a focus app with a pet beside the timer, so read this as an interested party. A soundscape you can play any time is not a bundle, and the own-iPod group in this study, the closest thing to a suggestion in software, did not reliably move. The parts of Yappie that exist only inside a session, the pet's reactions that only happen while the timer runs and the progress that only moves then, are the shape this study tested. Whether that shape holds over a screen and without a locker is our bet, not their finding.

References

1. Milkman, K. L., Minson, J. A., & Volpp, K. G. M. (2014). Holding the Hunger Games hostage at the gym: An evaluation of temptation bundling. Management Science, 60(2), 283 to 299. https://doi.org/10.1287/mnsc.2013.1784↩