Enhanced active choice – Keller et al (2011) #JournalClub

Some colleagues of mine were quite enthusiastic about this paper from 2011 Enhanced active choice: A new method to motivate behavior change by Punam Anand Keller, Bari Harlam, George Loewenstein and Kevin G. Volpp. Press release/pop-sci write-up: The Science of Better Choices.

The introduction gives a nice summary of possibilities for structuring choices. Choice architecture was a term promoted by the seminal book Nudge (2009) by Richard Thaler and Cass Sunstein. I highly recommend reading that book. Or -TL;DR- read Nudging: A Very Short Guide where Sunstein summarizes it in 7 pages.

Default in
Keller et al highlight the power of defaults. Opt-out gets you higher conversion or compliance than opt-in. Organ donation is a famous example; in Germany 12% is potentially an organ donor, whilst in Austria that number is 99.98%. All Austrians are donor unless they specifically state that they do not want to be an organ donor, Germany has an opt-in system: you have to actively register yourself (Do Defaults Save Lives?, Science paper by Johnson & Goldstein, 2003).

Opt-out also has limitations, and the paper lists five:

  1. Opt-out is “less likely to engender the kind of committed follow-up that is often useful when it comes to implementing the decision
  2. opt-out ‘choices’in many situations are less likely to reflect decision makers’ true preferences than will more active choices
  3. in some situations, passive choices are more likely to result in waste or inefficiency
  4. legally or ethically unacceptable
  5. opt-out policies can be counterproductive if those who implement them view them as a substitute for other, more substantive, interventions

(Enhanced) Active Choice
To circumvent some drawbacks of opt-in, whilst retaining the pros, an alternative is active choice: forcing people to choose. The authors elaborate on that:

In this paper, we examine a modified approach that we call ‘Enhanced Active Choice’ that advantages the option preferred by the communicator by highlighting losses incumbent in the non-preferred alternative.

In four studies they tests these hypotheses:

In sum, our main hypotheses are (H1) that Active Choice (‘unenhanced’ or basic and ‘enhanced’) will result in more compliance than opt-in non-enrollment defaults, and (H2) that Enhanced Active Choice will result in more compliance than basic Active Choice.

In studies 1 and 2 participants (total N=55 and N=110 respectively, which translates to N≈18 and N≈27 per experimental group) were asked their intention to get a flu-shot.

In study 1, agreement for Active Choice and Enhanced Active Choice (62% and 75% agreed to get flu-shot) was significantly higher than for Opt-in (42%). Active Choice was framed as (Underlined portion was only shown in Enhanced Active Choice-condition):

“I will get a Flu Shot this Fall to reduce my risk of getting the flu and I want to save $50 or, I will not get a Flu Shot this Fall even if it means I may increase my risk of getting the flu and I don’t want to save $50.

Similarly, in study 2, Enhanced Active Choice (93% did not choose alternative “I want to remind myself to get a flu shot”) was higher than Opt-in (45%). However, in contrast to study 1, Active Choice also resulted in lower agreement (52%).

Studies 3 (telephone, N=9,950) & 4 (website, N=11,182) looked at real medicine prescription refills and strived to enroll people in an automatic refill program. In both studies, Enhanced Active Choice resulted in significantly more enrollment than Active Choice: study 3: 32.0% vs 15.7%, and study 4: 21.9% vs 12.3%.

Some criticism on stats en framing
Oddly enough, in studies 1 & 2 the authors employ ANOVAs to compare %s, in studies 3 & 4 they (correctly I think) use χ²-tests. Perhaps because for study 1, and generously assuming group sizes of N=20, the χ²=5.07, which results in a p=0.079 with two degrees of freedom?

Furthermore, I fail to see how enhanced (i.e. “highlighting losses incumbent in the non-preferred alternative“) the following choices are:

  • Study 2: “I want to remind myself to get a flu shot”
  • Study 3: “press 1 if they preferred to refill their own prescription by themselves each time
  • Study 4: “I Prefer to Order my Own Refills”

For study 1, the enhancement (i.e. the detriment of the non-preferred option) is quite clear: you risk getting the flu and it would save you $50.


Enhanced Active Choice is relatively easy to implement in settings in which there are regular opportunities to interact with potential target groups. It is best implemented in conjunction with a mandatory task such as employee benefits enrollment


  • I think Enhanced Active Choice is an effective way to shape choices
  • I like the summary on pros and cons for different choice architecture choices
  • The statistics and sample sizes of studies 1 & 2 are dodgy
  • I fail to see the “enhanced” in studies 2, 3 and 4

Geef een reactie

Vul je gegevens in of klik op een icoon om in te loggen.

WordPress.com logo

Je reageert onder je WordPress.com account. Log uit /  Bijwerken )

Google+ photo

Je reageert onder je Google+ account. Log uit /  Bijwerken )


Je reageert onder je Twitter account. Log uit /  Bijwerken )

Facebook foto

Je reageert onder je Facebook account. Log uit /  Bijwerken )


Verbinden met %s