Do Nudges Reduce Disparities? Choice Architecture Compensates for Low Consumer Knowledge

Kellen Mrkva, Nathaniel A. Posner, Crystal Reeck, Eric J. Johnson

Research output: Contribution to journalArticlepeer-review

41 Citations (SciVal)


Choice architecture tools, commonly known as nudges, powerfully impact decisions and can improve welfare. Yet it is unclear who is most impacted by nudges. If nudge effects are moderated by socioeconomic status (SES), these differential effects could increase or decrease disparities across consumers. Using field data and several preregistered studies, the authors demonstrate that consumers with lower SES, domain knowledge, and numerical ability are impacted more by a wide variety of nudges. As a result, “good nudges” designed to increase selection of superior options reduced choice disparities, improving choices more among consumers with lower SES, lower financial literacy, and lower numeracy than among those with higher levels of these variables. Compared with “good nudges,” “bad nudges” designed to facilitate selection of inferior options exacerbated choice disparities. These results generalized across real retirement decisions, different nudges, and different decision domains. Across studies, the authors tested different explanations of why SES, domain knowledge, and numeracy moderate nudges. The results suggest that nudges are a useful tool for those who wish to reduce disparities. The research concludes with a discussion of implications for marketing firms and segmentation.

Original languageEnglish
Pages (from-to)67-84
Number of pages18
JournalJournal of Marketing
Issue number4
Early online date21 Jan 2021
Publication statusPublished - 1 Jul 2021


  • defaults
  • financial literacy
  • nudges
  • numeracy
  • socioeconomic status

ASJC Scopus subject areas

  • Business and International Management
  • Marketing


Dive into the research topics of 'Do Nudges Reduce Disparities? Choice Architecture Compensates for Low Consumer Knowledge'. Together they form a unique fingerprint.

Cite this