Noah Castelo, Elizabeth Hardy, Julian House, Nina Mazar, Claire Tsai, & Min Zhao
Summary. To improve efficiency and reduce costs, government agencies provide more and more services online. Yet, sometimes people do not access these new services. For example, prior to our field experiment intervention, Ontario spent $35 million annually on infrastructure needed for in-person license plate sticker renewals. In Canada’s most populous province, only 10% of renewals occurred online. Our intervention tested variations in messaging mailed with sticker renewal forms that encouraged consumers to renew online. We changed text and color on the envelope to try to make the benefits of the online service more salient. In addition, changes to text and color on the renewal form itself emphasized either consumer gains from online renewals or losses associated with in-person renewals. Each intervention increased use of the online service when compared to the unaltered messaging. The combination of salience and gain framing achieved the highest number of online renewals: a 41.7% relative increase.
It would be impossible to inventory all of the useful transactions people conduct on the Internet every single day. In 2014 alone, e-retailers based in the United States sold more than $300 billion of merchandise, according to the U.S. Department of Commerce.1 But valuable online activity includes much more than shopping. Most U.S. colleges and universities offer online courses, for example. People commonly publish books, raise money for deserving charities, and even find life partners online.
Governments, however, are not always as successful at convincing citizens to access public services online. For example, in a 2012 nationally representative survey, the majority (52%) of Canadians reported that they never or only sometimes access government services online, despite the fact that 84% in the same survey expressed interest in using such e-services.2 This is a missed opportunity, because conducting government business online, when done well, can cut costs and increase efficiency as well as consumer satisfaction.
In Ontario, Canada’s most populous province, the provincial government has made more than 40 services available online, including address changes, driver’s license and health card renewals, and copy requests for various records such as birth and marriage certificates. Expanding citizens’ use of just one online service, the annual or biennial renewal of automobile license plates, could potentially save millions of dollars. Since 2010, Ontarians have been able to go online to renew the stickers that display when their license plate fees expire. But as of 2013, only 10% of these transactions were conducted digitally. Servicing the remaining 90% of renewals in-person cost over $35 million that year; online renewals are accomplished for a fraction of that cost.
Efforts to fix such discrepancies can benefit from insights produced by researchers in the field of behavioral science, who have expertise in developing and testing methods that help people accomplish behavioral change. Beginning in 2013, we applied our expertise in a first-of-its-kind collaboration between Ontario’s Behavioural Insights Unit (BIU) and the Behavioural Economics in Action research hub (BEAR) at the Rotman School of Management of the University of Toronto. The results were promising. Not only did we increase online license plate sticker renewals in Ontario, but there is also evidence that our interventions increased the number of on-time renewals, too. These findings suggest that no-cost interventions such as those we tested can save a provincial government money and help citizens in ways beyond providing convenience. By renewing their stickers on time, vehicle owners avoid incurring citations and fines. At the same time, police forces potentially gain time to focus on more pressing enforcement matters.
Using Choice Architecture to Change Behavior
People frequently face a major challenge when it comes to acting on their good intentions. Behavioral scientists call this the last-mile problem.3 People who plan to save often put away less money than they need later, for example.4 People who fully believe in the importance of a good cause end up not donating to that cause.5 To help people to act on their good intentions, governments are increasingly applying behavioral science insights.6,7 One of the most successful approaches to helping people act on their intentions is to carefully design the choice architecture (that is, the way in which options are presented or preferences are elicited) in ways that nudge people to make desirable decisions while preserving their freedom of choice.8 For example, people tend to view outcomes in terms of losses and gains relative to context-dependent reference points, such as when they compare an existing investment portfolio’s worth with its maximum and minimum potential value. What is more, people are about twice as sensitive to differences framed as losses compared with the same differences framed as gains, a phenomenon known as loss aversion.9–11
In another example, partially due to loss aversion, the default option among a group of options tends to be “sticky.” That is, people are reluctant to give up a default because the losses inherent in doing so often subjectively outweigh the relative gains of alternative options. Accordingly, defaults have been shown to exert a strong influence on choices, even those as consequential as whether to become an organ donor12 or whether to enroll in a 401(k) retirement savings plan.13
Choice architecture interventions make use of such insights to help people make better decisions. For instance, informing people that the majority of citizens pay their taxes on time (communicating a descriptive social norm) led to a 15% increase in the percentage of British citizens who paid their overdue taxes in 2011 and 2012. During a 23-day experiment involving just 100,000 taxpayers, this resulted in an additional £9 million in government revenue. It has been estimated that this intervention could bring in approximately £160 million at very low cost when implemented countrywide.14,15 In a different field experiment, 6,824 insurance policy holders in the United States were asked to endorse the statement “I promise that the information I am providing is true” with their signature at the top rather than at the bottom of an audit form before self-reporting their cars’ odometer mileage. This precommitment to truthfulness (acting as moral reminder) significantly reduced the extent of dishonest self-reports, resulting in an estimated additional $1 million in insurance premium revenue over a 24-month period.16 Even subtle cues in a physical choice environment, such as queue guides and area carpets, can create the feeling that it is almost one’s turn and therefore increase the likelihood that consumers will stay in line rather than leave the queue.17
Leveraging Research on What Motivates People to Act
In our field experiment, we attempted to nudge more citizens of Ontario to renew their license plate stickers online rather than doing so in-person at ServiceOntario centers. These stickers need to be attached to vehicle license plates to show that owners have paid the required fees. Vehicle owners can choose to renew for one or two years, with no discount for renewing for two. The fees for renewing online are the same as those people pay when renewing in-person at ServiceOntario centers.
We tested three interventions that were developed using insights from two areas of behavioral science research. One body of work demonstrates the importance of salience, the degree to which something grabs a person’s attention. The other focuses on effects from framing situations or messages in terms that acknowledge losses versus gains.
The goal of our first intervention was to increase the salience of information regarding the online renewal option that citizens saw in the subject line on the mailing envelope carrying the sticker renewal form (see Figure 1). We made two adjustments motivated by the knowledge that when some information is emphasized more than other information in a communication, the emphasized content will have greater impact on people’s judgments.18,19
First, we embedded the black text of the subject line in a blue background to make it stand out on the exterior of an otherwise standard black-and-white ServiceOntario pressure-sealed envelope (see reference 14).20 Second, to increase the salience of benefits from online renewal, in the subject line on the envelope, we used the wording “Instant and easy renewal online” rather than “Renew online and receive a 10-day extension.” Our reasoning was that not all consumers require a 10-day extension. Further, because renewing license plate stickers online might seem complex and daunting to people inexperienced with the procedure, focusing on its benefits could help ease people into trying an unfamiliar process (see Figure 2).
Research has shown that when people think about an immediate action, they often primarily focus on hassles or the mental effort required to perform the action instead of the benefits. For example, when deciding whether to adopt a new technology such as a smart watch, consumers are often predisposed to think first about the learning costs (time, for instance) associated with new tools.21 Because of a well-documented process known as output interference, this can inhibit the consideration of potential benefits. In other words, the consideration of unwelcome factors interferes with the effective consideration of any subsequent factors.22–24
In light of these insights, we also changed the original messaging (see Figure 3) about the online renewal option printed on the back of the renewal form found inside the pressure-sealed envelope. Specifically, we designed our second and third interventions to emphasize in more detail the consumer benefits that accompany an online renewal. Our goal was to prompt consideration of the benefits before a person had time to think about any hassles or costs associated with undertaking an unfamiliar process. We framed the benefits of renewing online in two ways. In our second intervention, we directed people’s attention to the gains or positives associated with the online renewal (see Figure 4). In our third intervention, we applied one of the most robust findings in the behavioral science literature: that people are more averse to potential losses than they are attracted to equivalent potential gains (see research on loss aversion in, for example, reference 11).25 Here we highlighted the negatives, particularly the time cost, of not choosing to renew online (see Figure 5). In addition, in both of these framing interventions on the back of the form we also used color to try to heighten the salience of parts of the gain and loss messages by printing some text in red instead of black. The colors used in our interventions are visible in our Supplemental Material published online.
Testing Our Interventions
In our study, 626,212 owners of registered vehicles in Ontario received one of four different versions of a sticker renewal letter over eight weeks from December 2013 to February 2014. In the control condition, owners received the provincial government’s standard renewal letter (see Figures 1 and 3). The rest received envelopes with the color and text modifications described above (see Figure 2), as well as one of three different messages inside. What we call the salience-only condition (with the envelope altered but nothing else) featured the same message inside the envelope as the existing, standard renewal letter (see Figure 3). The salience-gain condition featured the gain-frame messaging and color modifications inside the envelope emphasizing the gains of renewing online (see Figure 4). And the salience-loss condition featured the loss-frame messaging and color modifications inside the envelope stressing the cost of not renewing online (see Figure 5).
We manipulated which version of the renewal letter was mailed to vehicle owners each week, according to the schedule in Table 1. Ninety days before each vehicle owner’s date of birth, when license plate stickers expire, ServiceOntario mails out renewal forms. As a consequence, vehicle owners’ assignment to one of our four conditions depended on their license plate sticker expiration date. Although complete randomization of assignment to condition would have been ideal, system limitations involved with printing and tracking hundreds of thousands of renewal forms mandated this approach.
As can be seen in Table 1, each condition was run two times for one-week periods, with four weeks between the periods. Our primary measure of success (our dependent variable) was the percentage of vehicle owners who renewed online. We also measured the percentage of vehicle owners in our sample who renewed on time. As can be seen in Figure 6, each of our three treatment conditions increased online renewals in comparison to the standard letter. The salience-gain treatment condition achieved the largest relative increase of 42.7% (increasing from 10.3% to 14.7%).
A technical description of our statistical analysis can be found online in our Supplemental Material. To summarize, we used two statistical techniques to ascertain whether the differences in online and on-time renewal rates between groups of people who had received different renewal letters was greater than would be expected by chance. The more common and less stringent was binomial logistic regression (Models 1 and 2 in Supplemental Table 1 online). It showed that people who received any of the three experimental letters developed for this trial were statistically more likely to renew their license plate stickers online and on time. Moreover, using this technique, we also found that both the salience-gain and salience-loss letters performed significantly better than the salience-only letter in terms of spurring online and on-time renewals. These results suggest that increasing the salience of the online service and emphasizing its relative gains helped encourage more consumers to take advantage of the online service.
The more stringent technique, called multilevel binomial logistic regression (Models 3 and 4 in Supplemental Table 1 online), is similar but was used to take into account a potential flaw in our trial design. Randomized controlled trials (RCTs) are normally conducted using random number generators or some other means of ensuring that each participant has an equal chance of receiving one of the various treatments (in this case, letters). In our trial, however, it was only possible to alter the version of the letter that was sent on a weekly basis. This created a potential problem, in that participants receiving a particular letter might exhibit behavior similar to the behavior of others receiving that same letter not because of the content of that letter but because of other coincident events occurring that week (for example, a weather event keeping people indoors and out of government offices). Because such factors could result in statistical errors, we created multilevel models to account for them. The results of these models suggest that only the salience-gain and salience-loss letters had significant positive effects on online renewals and that only the salience-gain letter had any effect on on-time renewals. This more stringent statistical analysis thus gives us greater confidence that the salience-gain letter prompted the best response of the four letters and that its apparent effect on online renewals is not due to chance.
The increased use of the online renewal service during our eight-week experimental period saved the government approximately $28,000 in transaction fees by reducing the number of in-person transactions in ServiceOntario centers, which are paid according to the number of in-person transactions processed. This estimate also takes into account marginal costs for the online service associated with meeting an increase in demand of this magnitude. We project that if the best-performing salience-gain treatment condition is adopted permanently in Ontario, approximately $612,000 would be saved annually, and this gain would be achieved at virtually zero cost to the government.
We also observed suggestive evidence that one of our interventions designed to increase online renewal service use also increased the number of on-time license plate sticker renewals. Such an additional benefit from increasing usage of an online service could be expected on the basis that online services are more convenient, are available at all times, and therefore pose fewer barriers that may encourage procrastination. However, as none of our interventions specifically addressed timely renewals, any effect on timely renewals would presumably be mediated through an increase in online renewals. Thus our evidence of this effect is less robust.
Figure 7 displays the on-time renewal rates by type of letter (for weekly results, see Table 1). Compared with the control condition, the salience-gain condition led to a 7.5% relative increase (from 70.4% to 75.7%) in timely renewals. The bottom half of Supplemental Table 1 online presents both the simple binary logistic regression results (Models 1 and 2) as well as the multilevel binary logistic regression results (Models 3 and 4). Both types of analyses agree on one observation: The salience-gain condition was most effective. In addition, the 1.3 times larger odds of on-time renewal in the salience-gain condition compared with the control condition can be considered a significant trend (highly significant according to the simple binary logistic regression and marginally significant according to the multilevel logistic regression). At minimum, we can conclude that our gain-frame messaging intervention on the back of the renewal form also had a significant, positive impact on timely renewals.
Subtle, Low-Cost Changes Have Impact
In a large-scale RCT in which people made real, consequential decisions, we observed that even subtle changes in messaging that apply behavioral principles can substantially influence people’s actions. It is important to note here that our interventions did not introduce citizens to the online service for the first time. The standard letter already directed vehicle owners to the ServiceOntario online renewal service and displayed its Internet address. Instead, the intervention increased participation in an existing government program, a benefit to both citizens and public policy due to its increased convenience and lower cost.
Our results add to the growing number of demonstrations of how simple and often cost-free interventions rooted in behavioral science research can be applied to help people make better decisions, without resorting to financial incentives or restricting freedom of choice. The interventions we tested, consistent with research on choice architecture and nudging (see reference 8),26 could be easily translated to other public policy realms. Highlighting communications materials to increase the salience of key messages and testing various gain and loss frames may be useful in helping citizens take advantage of any number of public services and benefits, such as the various government borrowing and savings programs that have seen improved participation due to interventions designed by the White House’s Social and Behavioral Sciences Team (SBST).27
Our results also reinforces the finding that using RCTs in the field is a feasible and valuable tool for testing some potential policy interventions.28 Staging such research allows for potent, rigorous comparison of multiple candidate interventions to determine which is the most effective in real-world contexts. Also, it can be implemented at very low cost when the proposed interventions are simply variants of existing processes, as was the case with our research.
It is also noteworthy that our findings are not consistent with the well-established finding that avoiding losses is a particularly strong motivator (see reference 11 and reference 25). Our messaging that emphasized potential gains (the gain frame) on the back of the renewal form was most effective and, in particular, more effective than the messaging that emphasized potential losses (the loss frame). This could be because people are already anticipating losses in the form of expected hassles when they consider using the in-person service instead of the online technology (see reference 21). That possibility may undermine effects from any additional loss thoughts. This unexpected finding illustrates the importance of testing even very fundamental assumptions before implementing a policy.
Our ability to explain this divergence from what has been observed in the loss aversion literature is limited by several imposed methodological constraints. That includes limitations regarding the number of conditions that we could run in an eight-week experimental period and the randomization capabilities of the ServiceOntario’s printer. For example, our salience-gain and salience-loss treatment conditions varied in several ways in addition to the framing. In particular, the salience-loss condition focused primarily on time savings and losses and presented both the negatives of not renewing online and the positives of renewing online. It did this in a particular order: It mentioned the negatives first, then the positives (see, for example, query theory; see reference 24). The salience-gain condition, on the other hand, focused only on the positives of renewing online, with no particular focus on time savings and losses. In addition, the salience-loss condition was much more specific when naming costs and benefits in that it mentioned very specific times, such as “wait in line (15 minutes),” in comparison to the benefits mentioned in the salience-gain condition (“save waiting time”).
Such differences make it difficult to tease apart the respective influences of each condition. Perhaps our findings reinforce the observation that a frame focused on loss is simply not effective when the focal attribute is time. That may be because people typically don’t view time investments as costs.29 Follow-up research could manipulate these elements in a more controlled manner to better understand their contribution. Exploring this and other potential mechanisms of our findings in laboratory studies could help refine and improve our interventions.
It is also worth noting that despite the success of our interventions relative to the status quo, the absolute number of citizens who used online renewal services was low—only 14.7% in the most effective condition. At this point, we can only speculate as to why. It is possible that many citizens did not read the content we added to the envelope or the renewal form with sufficient attention to encounter the online renewal opportunity. As a result, citizens may have habitually continued with the in-person service.
Another possibility is that some people are uncomfortable with conducting online transactions requiring personal information because of security and privacy concerns.30 This may particularly be true for older adults who are less familiar with such processes.31 Given that 84% of Ontarians have expressed interest in accessing government services online (see reference 2) but only 14.7% did so after our most successful intervention, we conclude that enormous potential continues to exist for behavioral science–based interventions to further narrow this intention–action gap.
Our experimental design, necessitated by practical constraints, introduced the possibility that our results could alternatively be explained by some other phenomenon that produced a four-week trend in online renewals, which was subsequently repeated the following four weeks. Although we see this alternative explanation as unlikely, it is impossible to rule out. It further illustrates the trade-offs between experimental control and the real-world field studies required to achieve external validity. In light of our positive results, however, ServiceOntario has embraced our most successful approach to nudge the people it serves to take advantage of its online renewal service. At the end of February 2015, it implemented the features of our salience-gain test condition by changing its standard mailing to all customers. The agency added the blue color and the text changes to the envelope and replaced the standard message about the online renewal option on the back of the renewal form with the positive gain-frame messaging and red color modification.
Aggregate online renewal rates every month after this province-wide implementation enabled us to compare year-over-year increases in online renewals for the seven months immediately before and after the implementation of the salience-gain letter. As can be seen in the dashed black line of Figure 8, after the salience-gain letter was implemented province-wide, there is a more pronounced increase in the percentage of license plate stickers renewed online each month compared with the same month in the previous year (represented by the solid gray line). To assess whether this increase is statistically significant, we conducted a mixed-method analysis of variance to control for any seasonal effects during the months in which the salience-gain letter was first implemented. An in-depth description of the statistical analysis and results can be found online in our Supplemental Material. To summarize, this analysis revealed a significant positive interaction, indicating that in the months after the salience-gain letter was implemented in Ontario, the year-over-year increases in the use of online services accelerated significantly, by about 1%.
In addition to our findings, we are sharing some reflections on the process we used to obtain these results. Our research is part of an ongoing collaboration between the Government of Ontario’s BIU and the Rotman School of Management’s BEAR to improve savings, efficiency, and well-being in Ontario. Through this collaboration, we’ve learned several practical lessons that may be of use to others working to apply behavioral science insights in the public policy realm:
- Whenever possible, a fully randomized experimental design is preferable. If this is not an option, it is crucial to come up with a design that limits the influence of confounds to ensure that measured effects are caused by the interventions.
- Data requirements for a study must be communicated clearly to all relevant parties. Further, data collection should be pretested before an RCT is conducted to ensure that all data are captured.
- A thorough application of behavioral insights with RCTs requires a research team with strong policy and program knowledge, an understanding of government, and support from academics with both theoretical and experimental expertise.
- Collaborations between scholars and government employees benefit from a division of labor that takes advantage of their respective expertise. These groups need strong administrative support.
- Researchers need allies to explore the promise of behavioral insights to improve public policy. Organized outreach, education, and support from senior leadership, and early adopters are all needed.
- A behavioral insights–driven philosophy can help a government better understand the people it serves.
- Outcomes that benefit citizens can lead to better outcomes for government programs.
This project has strengthened our confidence that applying behavioral insights and principles can produce measurable and positive outcomes at low cost. It has also reinforced our belief that collaborations such as the one described above can improve the performance of existing government programs. ♦
The authors are listed in alphabetical order. Julian House was responsible for the data analysis and interpretation; Nina Mazar was the research project lead (The findings and views expressed in this paper do not necessarily represent those of the World Bank or its Executive Directors). The research was conducted while Elizabeth Hardy was a manager in the Behavioural Insights Unit of the Government of Ontario (from 2013 to 2015). The research was supported by a TD Bank Group Research Fund awarded to BEAR.
Castelo, Columbia Business School, Columbia University; Hardy, Central Innovation Hub, Government of Canada; House, Behavioural Insights Unit (BIU), Government of Ontario, and Behavioural Economics in Action research hub (BEAR) at Rotman School of Management, University of Toronto; Mazar, BEAR, Rotman School of Management, University of Toronto, and Global INsights Initiative (GINI), World Bank; and Tsai and Zhao, BEAR, Rotman School of Management, University of Toronto. Corresponding authors’ e-mail addresses: firstname.lastname@example.org and email@example.com.
1. Enright, A. (2015). U.S. annual e-retail sales surpass $300 billion for the first time. Retrieved from Internet Retailer website: https://www.internetretailer.com/2015/02/17/us-annual-e-retail-sales-surpass-300-billion-first-ti
2. Jordan, M., Belmonte, D., Fowell, L. H., Tanaka, D., & Wagner, J. (2012). Next generation of eservices: Enhancing service delivery in the Canadian public sector. Retrieved from PwC website: http://www.pwc.com/en_GX/gx/psrc/pdf/citizen-compass.pdf
3. Soman, D. (2015). The last mile: Creating social and economic value from behavioral insights. Toronto, Ontario, Canada: University of Toronto Press.
4. Thaler, R., & Benartzi, S. (2004). Save More TomorrowTM: Using behavioral economics to increase employee saving. Journal of Political Economy, 112(Suppl. S1), S164–S187.
5. Bendapudi, N., Singh, S. N., & Bendapudi, V. (1996). Enhancing helping behavior: An integrative framework for promotion planning. Journal of Marketing, 60(3), 33–49.
6. House of Lords. (2011). Behavior change report. Retrieved from http://www.parliament.uk/business/committees/committees-a-z/lords-select/science-and-technology-committee/news/behaviour-change-published
7. Ly, K., & Soman, D. (2013). Nudging around the world. Retrieved from University of Toronto, Rotman School of Management website: http://cd-www.rotman.utoronto.ca/-/media/Files/Programs-and-Areas/behavioural-economics/Nudging%20Around%20The%20World_Sep2013.pdf
8. Thaler, R., & Sunstein, C. (2008). Nudge: Improving decisions about health, wealth, and happiness. New Haven, CT: Yale University Press.
9. Dinner, I., Johnson, E. J., Goldstein, D. G., & Liu, K. (2011). Partitioning default effects: Why people choose not to choose. Journal of Experimental Psychology: Applied, 17, 332–341.
10. Kahneman, D., Knetsch, J. L., & Thaler, R. H. (1990). Experimental tests of the endowment effect and the Coase theorem. Journal of Political Economy, 98, 1325–1348. http://doi.org/10.1086/261737
11. Kahneman, D., Knetsch, J. L., & Thaler, R. H. (1991). Anomalies: The endowment effect, loss aversion, and status quo bias. Journal of Economic Perspectives, 5(1), 193–206. http://doi.org/10.1257/jep.5.1.193
12. Johnson, E. J., & Goldstein, D. (2003, November 21). Do defaults save lives? Science, 302, 1338–1339. http://doi.org/10.1126/science.1091721
13. Madrian, B. C., & Shea, D. F. (2001). The power of suggestion: Inertia in 401(k) participation and savings behavior. The Quarterly Journal of Economics, 116, 1149–1187.
14. UK Cabinet Office Behavioural Insights Team. (2012). Applying behavioural insights to reduce fraud, error and debt. Retrieved from https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/%20%2060539/BIT_FraudErrorDebt_accessible.pdf
15. Hallsworth, M., List, J., Metcalfe, R., & Vlaev, I. (2014). The behavioralist as tax collector: Using natural field experiments to enhance tax compliance (NBER Working Paper 20007). Retrieved from National Bureau of Economic Research website: http://www.nber.org/papers/w20007
16. Shu, L. L., Mazar, N., Gino, F., Ariely, D., & Bazerman, M. H. (2012). Signing at the beginning makes ethics salient and decreases dishonest self-reports in comparison to signing at the end. Proceedings of the National Academy of Sciences, USA, 109, 15197–15200. http://doi.org/10.1073/pnas.1209746109
17. Zhao, M., Lee, L., & Soman, D. (2012). Crossing the virtual boundary: The effect of task-irrelevant environmental cues on task implementation. Psychological Science, 23, 1200–1207. http://doi.org/10.1177/0956797612441608
18. Alba, J. W., & Chattopadhyay, A. (1986). Salience effects in brand recall. Journal of Marketing Research, 23, 363–369.
19. Taylor, S. E., & Thompson, S. C. (1982). Stalking the elusive “vividness” effect. Psychological Review, 89, 155–181.
20. Kahneman, D. (2011). Thinking, fast and slow. New York, NY: Farrar, Straus and Giroux.
21. Zhao, M., Hoeffler, S., & Zauberman, G. (2007). Mental simulation and preference consistency over time: The role of process- versus outcome-focused thoughts. Journal of Marketing Research, 44, 379–388.
22. Anderson, M. C., & Neely, J. H. (1996). Interference and inhibition in memory retrieval. In E. Bjork & R. Bjork (Eds.), Memory handbook of perception and cognition (pp. 237–313). San Diego, CA: Academic Press.
23. Hoch, S. J. (1984). Availability and interference in predictive judgment. Journal of Experimental Psychology: Learning, Memory, and Cognition, 10, 649–662.
24. Johnson, E. J., Häubl, G., & Keinan, A. (2007). Aspects of endowment: A query theory of value construction. Journal of Experimental Psychology: Learning, Memory, and Cognition, 33, 461–474. http://doi.org/10.1037/0278-7318.104.22.1681
25. Kahneman, D., & Tversky, A. (1979). Prospect theory: An analysis of decision under risk. Econometrica, 47, 263–292.
26. Johnson, E. J., Shu, S. B., Dellaert, B. G. C., Fox, C., Goldstein, D. G., Häubl, G., . . . Weber, E. U. (2012). Beyond nudges: Tools of a choice architecture. Marketing Letters, 23, 487–504. http://doi.org/10.1007/s11002-012-9186-1
27. Social and Behavioral Sciences Team. (2015). Annual report. Retrieved from White House website: https://www.whitehouse.gov/sites/default/files/microsites/ostp/sbst_2015_annual_report_final_9_14_15.pdf
28. Haynes, L., Service, O., Goldacre, B., & Torgerson, D. (2012). Test, learn, adapt: Developing public policy with randomised controlled trials. Retrieved from United Kingdom Cabinet Office website: https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/62529/TLA-1906126.pdf
29. Leclerc, F., Schmitt, B. H., & Dubé, L. (1995). Waiting time and decision making: Is time like money? Journal of Consumer Research, 22, 110–119.
30. Carter, L., & Bélanger, F. (2005). The utilization of e-government services: Citizen trust, innovation and acceptance factors. Information Systems Journal, 15, 5–25.
31. Forsythe, S. M., & Shi, B. (2003). Consumer patronage and risk perceptions in Internet shopping. Journal of Business Research, 56, 867–875.