Journal of Educational Psychology © 2016 American Psychological Association 2016, Vol. 108, No. 3, 374–391 0022-0663/16/$12.00 http://dx.doi.org/10.1037/edu0000098 Using Design Thinking to Improve Psychological Interventions: The Case of the Growth Mindset During the Transition to High School David S. Yeager Carissa Romero and Dave Paunesku University of Texas at Austin Stanford University Christopher S. Hulleman Barbara Schneider University of Virginia Michigan State University Cintia Hinojosa, Hae Yeon Lee, and Joseph O’Brien Kate Flint, Alice Roberts, and Jill Trott University of Texas at Austin ICF International, Fairfax, Virginia Daniel Greene, Gregory M. Walton, and Carol S. Dweck Stanford University There are many promising psychological interventions on the horizon, but there is no clear methodology for preparing them to be scaled up. Drawing on design thinking, the present research formalizes a methodology for redesigning and tailoring initial interventions. We test the methodology using the case of fixed versus growth mindsets during the transition to high school. Qualitative inquiry and rapid, iterative, randomized “A/B” experiments were conducted with ϳ3,000 participants to inform interven- tion revisions for this population. Next, 2 experimental evaluations showed that the revised growth mindset intervention was an improvement over previous versions in terms of short-term proxy outcomes (Study 1, N ϭ 7,501), and it improved 9th grade core-course GPA and reduced D/F GPAs for lower achieving students when delivered via the Internet under routine conditions with ϳ95% of students at 10 schools (Study 2, N ϭ 3,676). Although the intervention could still be improved even further, the current research provides a model for how to improve and scale interventions that begin to address pressing educational problems. It also provides insight into how to teach a growth mindset more effectively. Keywords: adolescence, growth mindset, incremental theory of intelligence, motivation, psychological intervention One of the most promising developments in educational psy- how students view their abilities, their experiences in school, their chology in recent years has been the finding that self-administered relationships with peers and teachers, and their learning tasks (see psychological interventions can initiate lasting improvements in Ross & Nisbett, 1991). student achievement (Cohen & Sherman, 2014; Garcia & Cohen, For instance, students can show greater motivation to learn 2012; Walton, 2014; Wilson, 2002; Yeager & Walton, 2011). when they are led to construe their learning situation as one in These interventions do not provide new instructional materials or which they have the potential to develop their abilities (Dweck, pedagogies. Instead, they capitalize on the insights of expert teach- 1999; Dweck, 2006), in which they feel psychologically safe and ers (see Lepper & Woolverton, 2001; Treisman, 1992) by address- connected to others (Cohen, Garcia, Apfel, & Master, 2006; Ste- This document is copyrighted by the American Psychological Association or one of its allieding publishers. students’ subjective construals of themselves and school— phens, Hamedani, & Destin, 2014; Walton & Cohen, 2007), and in This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. David S. Yeager, Department of Psychology, University of Texas at Austin; Dean of Humanities and Social Sciences at Stanford University, Angela Carissa Romero and Dave Paunesku, PERTS and Department of Psychology, Duckworth, a William T. Grant scholars award, and a fellowship from the Stanford University; Christopher S. Hulleman, School of Education, University of Center for Advanced Study in the Behavioral Sciences (CASBS) to the first Virginia; Barbara Schneider, School of Education Michigan State University; and fourth authors. The authors are grateful to Angela Duckworth, Elizabeth Cintia Hinojosa, Hae Yeon Lee, and Joseph O’Brien, Department of Psychology, Tipton, Michael Weiss, Tim Wilson, Robert Crosnoe, Chandra Muller, Ronald University of Texas at Austin; Kate Flint, Alice Roberts, and Jill Trott, ICF Ferguson, Ellen Konar, Elliott Whitney, Paul Hanselman, Jeff Kosovich, International, Fairfax, Virginia; Daniel Greene, Gregory M. Walton, and Carol S. Andrew Sokatch, Katharine Meyer, Patricia Chen, Chris Macrander, Jacquie Dweck, Department of Psychology, Stanford University. Beaubien, and Rachel Herter for their assistance. The look and feel of the We thank the teachers, principals, administrators, and students who revised intervention was developed by Matthew Kandler. participated in this research. This research was supported by generous funding Correspondence concerning this article should be addressed to David S. from the Spencer Foundation, the William T. Grant Foundation, the Bezos Yeager, Department of Psychology, University of Texas at Austin, Austin, Foundation, the Houston Endowment, the Character Lab, the President and TX 78712. E-mail: [email protected] or [email protected] 374 DESIGN THINKING AND PSYCHOLOGICAL INTERVENTIONS 375 which putting forth effort has meaning and value (Hulleman & Yeager & Dweck, 2012). These psychological processes can result Harackiewicz, 2009; Yeager et al., 2014; see Eccles & Wigfield, in academic resilience. 2002; also see Elliot & Dweck, 2005; Lepper, Woolverton, Mumme, & Gurtner, 1993; Stipek, 2002). Such subjective con- struals—and interventions or teacher practices that affect them— Design Thinking and Psychological Interventions can affect behavior over time because they can become self- To develop psychological interventions, expertise in theory is confirming. When students doubt their capacities in school—for crucial. But theory alone does not help a designer discover how to example, when they see a failed math test as evidence that they are connect with students facing a particular set of motivational bar- not a “math person”—they behave in ways that can make this true, riers. Doing that, we believe, is easier when combining theoretical for example, by studying less rather than more or by avoiding expertise with a design-based approach (Razzouk & Shute, 2012; future math challenges they might learn from. By changing initial also see Bryk, 2009; Yeager & Walton, 2011). Design thinking is construals and behaviors, psychological interventions can set in “problem-centered.” That is, effective design seeks to solve pre- motion recursive processes that alter students’ achievement into dictable problems for specified user groups (Kelley & Kelley, the future (see Cohen & Sherman, 2014; Garcia & Cohen, 2012; 2013; also see Bryk, 2009; Razzouk & Shute, 2012). Our hypoth- Walton, 2014; Yeager & Walton, 2011). esis is that this problem-specific customization, guided by theory, Although promising, self-administered psychological interven- can increase the likelihood that an intervention will be more tions have not often been tested in ways that are sufficiently effective for a predefined population. relevant for policy and practice. For example, rigorous randomized We apply two design traditions, user-centered design and A/B trials have been conducted with only limited samples of students testing. Theories of user-centered design were pioneered by firms within schools—those who could be conveniently recruited. These such as IDEO (Kelley & Kelley, 2013; see Razzouk & Shute, studies have been extremely useful for testing of novel theoretical 2012). The design process privileges the user’s subjective perspec- claims (e.g., Aronson, Fried, & Good, 2002; Blackwell, Trzesni- tive—in the present case, 9th grade students. To do so, it often ewski, & Dweck, 2007; Good, Aronson, & Inzlicht, 2003). Some employs qualitative research methods such as ethnographic obser- studies have subsequently taken a step toward scale by developing vations of people’s mundane goal pursuit in their natural habitats methods for delivering intervention materials to large samples via (Kelley & Kelley, 2013). User-centered design also has a bias the Internet without requiring professional development (e.g., toward action. Designers test minimally viable products early in Paunesku et al., 2015; Yeager et al., 2014). However, such tests are the design phase in an effort to learn from users how to improve limited in relevance for policy and practice because they did not them (see Ries, 2011). Applied to psychological intervention, this attempt to improve the outcomes of an entire student body or entire can help prevent running a large, costly experiment with an inter- subgroups of students. vention that has easily discoverable flaws. In sum, our aim was to There is not currently a methodology for adapting materials that acquire insights about the barriers to students’ adoption of a were effective in initial experiments so they can be improved and growth mindset during the transition to high school as quickly as made more effective for populations of students who are facing possible, using data that were readily obtainable, without waiting particular issues at specific points in their academic lives. We seek for a full-scale, long-term evaluation. to develop this methodology here. To do so, we focus on students not at diverse schools but at a similar developmental stage, and who User-centered design typically does ask users what they therefore may encounter similar academic and psychological chal- desire or would find compelling. Individuals may not always have lenges and may benefit from an intervention