An easy way to reduce socioeconomic disparities
Researchers from Columbia University and Temple University published a new paper in the Journal of Marketing that examines how choice architecture can reduce socioeconomic disparities.
The study, forthcoming in the Journal of Marketing, is titled "Do Nudges Reduce Disparities? Choice Architecture Compensates for Low Consumer Knowledge" and is authored by Kellen Mrkva, Nathaniel Posner, Crystal Reeck, and Eric Johnson.
As Mrkva explains, "Our research demonstrates that people with low socioeconomic status (SES), low numerical ability, and low knowledge are most impacted by nudges. As a result, 'good nudges,' designed to encourage selection of options that are in people's best interests, reduce SES disparities, helping low-SES people more than high-SES people." On the other hand, nudges that encourage selection of inferior options exacerbate disparities relative to "good nudges" because low-SES consumers are more likely to retain inferior default options. In other words, nudges are a double-edged sword that can either reduce disparities or make matters worse because they impact low-SES people most. The research team generalized its findings across three different types of nudges, several different consumer decision contexts, and real retirement decisions.
This research has major implications, including for the COVID vaccination process. Across the country, millions of people are now eligible to get a COVID vaccine. However, the signup process is often unnecessarily complex. New York's nycHealthy sign-up portal, for example, includes as many as 51 questions and requests that you upload your insurance card. As a result, many people, especially the elderly, poor, and less digitally literate, have struggled or failed to make an appointment. As Johnson explains, "Our research suggests that making beneficial behaviors like vaccination simpler has a crucial and underappreciated advantage—it reduces socioeconomic disparities. On the other hand, when these behaviors are unnecessarily complex, it is typically low-SES consumers who are harmed the most."
In five experiments as well as data from real retirement decisions, the researchers show that people who are lower in SES, domain knowledge, and numeracy are impacted more by a variety of nudges. As a result, "good nudges" that facilitate selection of welfare-enhancing options reduce disparities by helping low-SES, low-knowledge, and low-numeracy consumers most.
In Study 1, participants made five consumer financial decisions. For each decision, they were randomly assigned to a "no default," "good default," or "bad default" condition (the latter two pre-selected correct or incorrect options, respectively). After they made these five decisions, participants completed common measures of the three hypothesized moderators—financial literacy, numeracy, and socioeconomic status. As predicted, there was a large default effect. There were also interactions between the default condition and the three moderators; participants lower in these moderators were more impacted by defaults. These effects remained significant when adding survey engagement, comprehension, need for cognition, agreeableness, decision time, and their interactions with condition to the model as covariates.
Study 2 examines whether these effects generalized across three different types of nudges and three decision contexts. It replicated the SES and financial literacy effects of Study 1 across all nudges and contexts. Unlike Study 1 and all subsequent studies, the nudge x numeracy interaction was not significant. The key effects remained significant when controlling for a measure of fluid intelligence.
Study 3 uses syndicated data from stratified random samples of American households about their retirement investment decisions to examine a sample of people who work for companies that use defaults to automatically enroll employees into retirement contributions. Respondents reported whether they retained or opted out of the default contribution amount and default investment allocation. Evidence supports that lower-SES and less financially literate people are more impacted by nudges and thus less likely to opt out of these retirement defaults: Lower-SES participants were less likely to opt out as were participants with lower financial literacy.
Study 4 replicated these effects in the context of COVID-19 health decisions (e.g., deciding whether to wear a mask). Additionally, domain-specific health knowledge moderated default effects whereas other-domain knowledge did not. Studies 5-6 replicated the predicted moderators from Study 1 with incentives. Mediation models suggest that people with lower SES, domain knowledge, and numeracy were more impacted by nudges partly because they experience higher uncertainty and decision anxiety when making decisions.
Across the six studies, nudges influenced choice disparities across people. Posner summarizes the study by saying "Our results suggest that nudges that make behaviors such as retail purchases, vaccine sign-up, and retirement contributions more automatic can reduce socioeconomic inequities."