Growth mindset tempers the effects of poverty onacademic achievement Susana Claroa,1, David Pauneskub, and Carol S. Dweckb,1 a Graduate School of Education, Stanford University, Stanford, CA 94305-3001; and bDepartment of Psychology, Stanford University, Stanford, CA 94305 Contributed by Carol S. Dweck, May 25, 2016 (sent for review September 27, 2015; reviewed by Ross A. Thompson and Timothy D. Wilson) Two largely separate bodies of empirical research have shown have now shown that growth mindset plays a causal role in that academic achievement is influenced by structural factors, such achievement. These field experiments, including two blinded, as socioeconomic background, and psychological factors, such as randomized, controlled studies conducted with over 1,500 par- students’ beliefs about their abilities. In this research, we use a ticipants each, have shown that targeted interventions can help nationwide sample of high school students from Chile to investi- students start to develop a growth mindset and that such inter- gate how these factors interact on a systemic level. Confirming ventions can lead to higher achievement for students facing prior research, we find that family income is a strong predictor greater adversity (10–14). of achievement. Extending prior research, we find that a growth However, because previous research was conducted with un- mindset (the belief that intelligence is not fixed and can be de- representative samples and lacked socioeconomic data, it has veloped) is a comparably strong predictor of achievement and that been impossible for researchers to address fundamental ques- it exhibits a positive relationship with achievement across all of tions about the relationship between mindset and socioeconomic the socioeconomic strata in the country. Furthermore, we find that achievement gaps. Is the relationship between mindset and ac- students from lower-income families were less likely to hold a ademic achievement a lawful pattern that can be observed reliably growth mindset than their wealthier peers, but those who did across an entire nation, and is it strong enough to be practically hold a growth mindset were appreciably buffered against the del- meaningful when measured against canonical structural factors, eterious effects of poverty on achievement: students in the lowest like family income? Is there evidence that economic disadvantage 10th percentile of family income who exhibited a growth mind- reinforces the fixed mindset? Finally, is a fixed mindset even more set showed academic performance as high as that of fixed mindset deleterious to economically disadvantaged students because they students from the 80th income percentile. These results suggest that must overcome greater obstacles to succeed? We systematically students’ mindsets may temper or exacerbate the effects of economic investigate these questions for the first time, to our knowledge, disadvantage on a systemic level. using a national dataset containing all 10th graders in Chile. mindset | academic achievement | income | inequality | education equality Materials and Methods This work uses a dataset of all 10th grade public school students in Chile to S ocioeconomic background is one of the strongest, best address these questions on a national scale. The Chilean Government ad- ministers standardized tests to measure the mathematics and language skills established predictors of academic achievement (1, 2). It is of all 10th graders in the country every other year. It also surveys each well-known that economic disadvantage can depress students’ academic achievement through multiple mechanisms, including reduced access to educational resources, higher levels of stress, Significance poorer nutrition, and reduced access to healthcare (3–5). None- theless, students with the same economic background clearly vary in This study is the first, to our knowledge, to show that a growth their academic outcomes, and researchers have long suggested that mindset (the belief that intelligence is not fixed and can be de- students’ beliefs, such as locus of control, may temper or exacerbate veloped) reliably predicts achievement across a national sample of the effects of economic disadvantage on academic achievement (6– students, including virtually all of the schools and socioeconomic 9). However, there has been a lack of clarity as to what these beliefs strata in Chile. It also explores the relationship between income are or how they interact with structural factors, like economic dis- and mindset for the first time, to our knowledge, finding that advantage, on a systemic level. The current research identifies a students from lower-income families were less likely to hold a belief—students’ mindset about intelligence—that is systematically growth mindset than their wealthier peers but that those who did associated with economic disadvantage and moderates its effects on hold a growth mindset were appreciably buffered against the achievement. Importantly, it is also a belief that is potentially deleterious effects of poverty on achievement. These results amenable to change (10–14). suggest that mindsets may be one mechanism through which Numerous studies have found that students fare better if they economic disadvantage can affect achievement. believe that their intellectual abilities can be developed—a belief Author contributions: S.C. designed research; S.C. and C.S.D. performed research; S.C. and called growth mindset—than if they believe that their intellectual D.P. analyzed data; S.C., D.P., and C.S.D. wrote the paper; and S.C., D.P., and C.S.D. abilities are immutable—a belief called fixed mindset (15). These designed figures. studies have documented numerous ways in which mindsets in- Reviewers: R.A.T., University of California, Davis; and T.D.W., University of Virginia. fluence behaviors that impact academic achievement (16–18). The authors declare no conflict of interest. Students with a fixed mindset tend to avoid situations in which Data deposition: The original datasets used for this study belong to the Chilean Depart- they might struggle or fail because these experiences undermine ment of Education. Applications to access these datasets should be done at www. their sense of their intelligence. In contrast, students who have a agenciaeducacion.cl/simce/bases-de-datos-nacionales/ for the SIMCE datasets. The JUNAEB dataset can be downloaded directly from their site at junaeb.cl/wp-content/ growth mindset tend to see difficult tasks as a way to increase uploads/2013/02/PRIORIDADES-2012-B%C3%81SICA-MEDIA-COMUNA-CON-IVE-SINAE- their abilities (11) and seek out challenging learning experiences OFICIAL.xlsx. that enable them to do so (16, 17). As a consequence, students 1 To whom correspondence may be addressed. Email:
[email protected] or dweck@ who have a growth mindset tend to earn better grades than stanford.edu. students who hold a fixed mindset (11, 17, 18), especially in the This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. face of difficulty. Additionally, a number of field experiments 1073/pnas.1608207113/-/DCSupplemental. 8664–8668 | PNAS | August 2, 2016 | vol. 113 | no. 31 www.pnas.org/cgi/doi/10.1073/pnas.1608207113 343) in a composite ual school.486 0. measured students’ mindsets about the 0.8% of variance (r = 0. We found that the relationship be- tween mindset and achievement could be observed across the Student-level variables socioeconomic spectrum and even when controlling for an ex- Mindset 0. canonical predictors of academic Thus. effect per school. Column 4 in Table 2 further adds school- 2010 level variables.11. against canonical structural factors.29.336 series of hierarchical linear regression models (21) controlling education for all available canonical predictors of achievement (1. PNAS | August 2. such as the mindset item and completed at least one standardized test (n = 168.g. P < 0. 19). the model included a school-level random com- dent-level socioeconomic predictor explained 11. categorized as having a growth mindset. those who disagreed or strongly disagreed were on language test scores and 0. Among school-level association between mindset and achievement for each of 2. the relationship between student mindsets and achievement was An additional model was created to assess the reliability of the comparably strong and held across all students in Chile. School language 0.301 0.392 public schools. suggest that. geographic region.119. SES. whereas a continuous standardized the other way around.13 SDs on mathematics test scores.001 for mathematics).427 0. 31 | 8665 .077– ance). books. and the schools represent 98% of all 2. are provided in SI Materials and Methods. such as agreement with the statements “I am listed in Table S1.Table 1. level control variables. the estimate student. on average.338 0. including the socioeconomic level of the school.. telligent. such as family income and parents’ education.289 shows.525 0. SE = 0. “intelligence is student who changes from having a fixed mindset to a mixed something that cannot be changed very much” and “you can learn new mindset or from a mixed mindset to a growth mindset is 0.001). and each school. smart.402 0.326 Table 2 presents the results. mother’s and father’s edu- SES quintile 0. School mathematics 0.500 cations. These students for reverse causation. tent with prior findings (1. Unconditional Pearson correlations between key mindset”—was again on par with this variable (explaining 26. P < 0.and school- The difference between these correlations was statistically sig. As Fig. students who subscribed to a growth mindset outperformed Natural log of 0. The descriptive dents’ self-assessments of their intelligence and their ability in statistics for variables on the whole population and the analytical sample are each subject.516 mathematics and language scores without any covariates. That is. These beliefs included stu- missing data are available in SI Materials and Methods. we estimate a positive nificant: Fisher’s r to Z = 2. 19. Consis. A detailed description of controls for a variety of beliefs and expectations that could play a the population as well as the imputation methods that were used for role in this reverse causal process. such as believing themselves to be in- the items. 22). accomplished students.336). that can be observed reliably across an entire nation and whether The relationship between mindsets and achievement remained it is strong enough to be practically meaningful when measured highly significant when controlling for these factors (B = 0.553 for mathematics and language. but you can’t change a person’s intelligence”) were categorized as having a fixed mindset.138.528 0. The details. the academic growth associated with a ments suggesting that intelligence cannot be changed (i. 113 | no.625 0.001 for language and B = 0. Details about each variable are in SI Materials and Methods. showing that the relationship be- education tween mindset and test scores still holds in each of these models. the poverty concentration index was schools included in the sample (the 95% plausible value range for the strongest predictor of test scores (explaining 26.275 0.203 and belief that their intellectual ability can grow over time. students who do well may hold score was used in analyses. This difference was not statistically significant: Average Fisher’s r to Z = 1. and the school’s 2010 average mathematics and language test scores.261. ponent for the mindset coefficient as well as all student.6% variables and standardized achievement test scores of the variance). Importantly.58.331 0.” We also Results controlled for student’s and parents’ expectations of the stu- First.e.001 for language and B = 0.292 0. and come to the conclusion that intelligence can be developed. the model accounted for almost all of the variability of scores between schools (93–95% depending on the subject) and 36–44% of the total variance. like family income. P < 0. socioeconomic status.. 2016 | vol.203.578 school enrollment. and the top stu.and school-level factors.343 tensive list of important student.412 −0. family income. we sought to determine whether the relationship between dent’s academic attainment and the degree to which the stu- mindset and academic achievement constitutes a lawful pattern dent liked each subject area and thought that it was important.” “I am better than the majority of my classmates on mathematics tests. and those who were uncertain We also considered the possibility of reverse causation—per- were categorized as having a mixed mindset. were themselves as doing well simply observe their academic growth correlated with test scores in our sample (Table 1). including Poverty index −0. the presence of household assets (e. respectively).512 gender. P < 0. each student’s family.326 0. Students who agreed or strongly agreed with state. A categorical system was used haps doing well in school leads to a growth mindset rather than in graphical presentations for clarity.3% (r = 0. we sought to determine the robustness and general- Variable Language Mathematics mathematics izability of this relationship. School-level variables To start. SE = malleability of intelligence using a short version of the standard instrument 0. 1 Family income 0.02 (20). To calculate a range of plausible values for the mindset average of mathematics and language scores. type of administration. column 2 in Table 2 shows this analysis for standardized Average mindset 0. language and Second.002.520 −0. whereas the average mindset at the school—or “school 0.319 their peers at each family income level.610 and family structure.2 SDs things.333 0. P = 0. ethnic origin.508 0.339 socioeconomic variables.001 for mathematics). P = 0. Student relationship between mindset and achievement across each individ- mindset explained 11.2% of vari. covariates included. It is plausible that these positive The analyses include all public school students who answered at least one self-perceptions could lead to other positive beliefs. PSYCHOLOGICAL AND COGNITIVE SCIENCES 2010 urbanicity. we ran the previous model and added represent 75% of all 10th graders from Chile’s public schools. computer). to our knowledge. the association between mindset and language per school was 0. Through this analysis.249 0.002. To test n = 168. Father years of 0. including the Spanish translation of positive self-perceptions.” and “I do well in language arts. our effect is not because of the fact that students who see achievement.269 0. Furthermore.171. mindset family income remained a highly significant predictor of achievement across a Mother years of 0. however. Column in school 3 in Table 2 controls for student-level characteristics.513 0. average class size.226 0. The 2012 student survey for of the mindset effect on achievement remained significant (B = the first time. With all of these important All values reported are significant (P < 0. The estimated coefficients used by Dweck (15). and the 95% plausible value range for the association between Claro et al. we note that mixed mindset students consistently fell in between the two other groups.001 and B = −0. fixed mindset. those likely to endorse a fixed mindset as students from the top-income with a growth mindset achieved at higher levels than those with a families and schools (Fig. Details are in mindset. Language and mathematics test scores predicted by mindset score (standardized) when controlling for student.552 168.Fig.203 168.200). Second.003 0. we tested whether a fixed mindset was even more to our knowledge. we tested the respectively.org/cgi/doi/10. A simple correlation revealed that students’ mindsets and SI Materials and Methods.552 No.01.018.002 0. At the Consistent with prior experimental studies. which is shown in SI Materials and Methods. lacking the resources of holds true systemically—across an entire nation’s socioeconomic higher-income students. Column 3 adds student-level controls.339 2. Full list of controls is available in SI Materials and Methods. indeed. mindset and mathematics per school was 0.and school-level variables Test score predicted Test score predicted Test score predicted by mindset and by mindset with by mindset and both student. conversely.020. P < 0. that a growth mindset may help Table 2. of schools 2.138* SE 0. However. and B shows mathematics scores.552 168.339 Each column describes a maximum likelihood hierarchical linear model with students nested in schools. and solid lines represent students with fixed mindset.214* 0. P < as a function of income as well as the relationship between income 0. P < 0.203 No.and Variables no other controls student-level variables school-level variables Language score Mindset regression coefficient 0. these results show for the first time. our results show extremes. family income were. A shows language scores. Column 2 presents mindset standardized regression coefficients without any control variables. .206* 0. of students 168.002 Student controls Included Included School controls Included No. to succeed. First. of schools 2. For clarity.001 for language and mathematics.pnas.146* 0. 1. that this relationship is comparably strong harmful to the academic achievement of economically disadvan- with that between family income and achievement and that it taged students because those students.203* SE 0. Table S2) possibility that economic disadvantage and the limited structural suggested that lower income magnifies the deleterious effects of opportunities associated with it could themselves reinforce a fixed a fixed mindset or. would need to overcome greater obstacles spectrum and across virtually all of its schools. Column 4 adds school-level controls. Furthermore.001). Table S1. 2).339 2. 8666 | www. A negative interaction between family income (stan- A final series of models investigated the prevalence of mindsets dardized) and mindset in predicting test scores (B = −0.140* 0. linked (r = 0.038–0.1073/pnas. Dashed lines represent students with growth mindset. Average standardized mathematics and language test scores for students with growth and fixed mindsets by family income decile. of students 168.002 0. only fixed mindset and growth mindset (not mixed mindset) students are included.17.339 Mathematics score Mindset regression coefficient 0. students from the lowest-income families were twice as that.1608207113 Claro et al. for students with the same observable characteristics. and achievement as a function of mindset.339 2. *Regression coefficients are P < 0.203 168.339 2.002 Student controls Included Included School controls Included No.002 0. Thanks to lowest-income Chilean students were twice as likely as the highest. 8. Reardon SF (2011) The widening of the socioeconomic status achievement gap: New Psychol 38(2):113–125. J Exp Soc 2. Lefkowitz MM. with decades of research and our own data. We To be clear. at every a growth mindset is a substitute for systemic efforts to alleviate socioeconomic level. Aronson J. 14). Gordon NH (1980) Childhood depression. Proc Natl Acad Sci USA 106(16):6545–6549. This difference is illustrated in Fig. and adult working 12. Murnane RJ (Russell Sage Foundation. 13. Mueller CM. which has found that a fixed mindset is more de- Discussion bilitating (and a growth mindset is more protective) when individuals The results of this study speak to researchers. Coleman JS. and Development locus of control. 2016 | vol. research on psychological factors can help growth mindset is more likely to enjoy higher academic achievement. like document for the first time. Loeb. Schamberg MA (2009) Childhood poverty. J Pers 32(3):355–370. strikingly. Our research shows that. Shores. Coleman JS (1966) Equal schools or equal students? Public Interest (4):70–75. Brooks-Gunn J. This research was funded. the impact of structural inequalities on achievement and future for any two students with equal characteristics. schools. Shear. These findings also document for the first time. We thank the team at the Agencia de Calidad at a relationship between mindsets and economic disadvantage. An intervention to reduce the effects of stereotype threat. students from low-income families (the lowest in part by leading low-income students to believe that they cannot 10%) who had a growth mindset showed comparable test scores grow their intellectual abilities. Washington. we are not suggesting that structural factors. As such. and 15. vation and performance. S. Lara. 31 | 8667 . on a national scale a income inequality or disparities in school quality. Yeager DS. 4. J Appl Psychol 82(2):221–234. Rather. pp 91–116. The the Chilean Department of Education for their valuable support. Evans GW. 3. Dweck CS (2007) Implicit theories of intelligence predict children’s life chances. that economic disadvantage may lead to poorer academic outcomes. Paunesku D. Tesiny EP. demic underachievement. to our knowledge. Some income deciles are missing because on the questionnaires. we are sug- stant a panoply of socioeconomic and attitudinal factors. L. Philadelphia). Blackwell LS. Brady. part. Future Child 24(1):41–59. test anxiety. In other words. 1. PSYCHOLOGICAL AND COGNITIVE SCIENCES mitigate the negative effects of economic deprivation on academic were more likely to endorse a fixed mindset. ready for scaling: The case of the growth mindset during the transition to high school. ACKNOWLEDGMENTS. parents were not offered an income choice corresponding to that decile. serve that. Good C (2002) Reducing the effects of stereotype threat on Printing Office. income students to report a fixed mindset. B. B. S. K. Kraft. Duncan GJ (1997) The effects of poverty on children. Valentino. pp 1066–5684. Trzesniewski KH. achievement across an adolescent transition: A longitudinal study and an intervention. Budge. Wilson. Donovan. B. D. are less important robust relationship between students’ mindsets about intelligence than psychological factors. where we ob. Thompson for helpful input. (Psychology. Butterfield EC (1964) Locus of control. New York). (2015) Mind-set interventions are a scalable treatment for aca- 5. Such claims would stand at odds consistently outperform those who do not—even after holding con. these robust. The observation that mindset is a with fixed mindset students whose families earned 13 times more more important predictor of success for low-income students than (80th percentile). gesting that structural inequalities can give rise to psychological lationship between mindset and achievement holds true across all of inequalities and that those psychological inequalities can reinforce Chile’s schools and across all levels of family income. evidence and possible explanations. for their high-income peers is novel. Aronson J. 2. Personality. However. Dee. chronic stress. 16. L. (2016) Using design thinking to make psychological interventions 7. Fried CB. must overcome significant barriers to succeed (13. Whither opportunity? Rising inequality. Child Dev 78(1):246–263. those who hold more of a growth mindset poverty and economic inequality. et al. Reardon. reactions to frustration. Psychol Sci 26(6):784–793. 6. J Educ Psychol 108(3):374–391. only fixed mindset and growth mindset (not mixed mindset) students are included. Castellano. Percentage of students with fixed and growth mindsets as a function of family income. J Pers Soc Psychol 75(1):33–52. Rock DA (1997) Academic success among students at risk for school failure. Dweck CS (1998) Praise for intelligence can undermine children’s moti- ment attitudes1. 113 | no. Percentages do not add up to 100 at each income decile because. and achieve. illuminate one set of processes through which economic disad- suggesting that the benefit of having a growth mindset holds widely. family income.Fig. (1966) Equality of Educational Opportunity (US Government 10. and their mindset was an T. for clarity. this finding does suggest achievement. nation-level correlations are com. although it is consistent with prior research. R. and R. Future Child 7(2):55–71. The re. Thompson RA (2014) Stress and child development. 9. Yeager. the one endorsing a opportunity. K. J Nerv Ment Dis 168(12):732–735. DC). and 11. Inzlicht M (2003) Improving adolescents’ standardized test performance: memory. PNAS | August 2. cymakers interested in understanding equality of opportunity. et al. 14. et al. in even stronger predictor of success for these low-income students. 1. T. vantage leads to academic underachievement and reveal ways to Furthermore. Finn JD. Good C. M. African American college students by shaping theories of intelligence. Dweck CS (2000) Self-Theories: Their Role in Motivation. to our knowledge. Quay. educators. Claro et al. more effectively support students who face additional challenges plemented by multiple prior randomized field experiments showing because of their socioeconomic circumstances. Families reported their income by selecting one income range from a list. S. J Appl Dev Psychol 24(6):645–662. Nor are we saying that teaching students and their academic performance. and poli. we note that the percentage of mixed mindset students consistently fell in between the two other groups. M. by Raikes Foundation Grant 227 (to the Stanford University Project Although existing data cannot explain why low-income students for Education Research That Scales). eds Duncan GJ. that a growth mindset has a causal impact on achievement (10–14). Educ Eval Policy Anal 33(2):119–137. 20. CA). 21.pdf. J Educ Psychol 88(3):397–407. 8668 | www. Romaguera P. Stipek D. Repetto A (2011) The effectiveness of private voucher education: Evidence 18.org/cgi/doi/10.agenciaeducacion. Bryk AS (2002) Hierarchical Linear Models: Applications and Data tional functioning in middle school: The role of implicit theories. 3rd Ed. NJ). 2nd Ed. Thousand Oaks. 22.1608207113 Claro et al. Lara B. mance.cl/wp-content/uploads/ in the generation of school quality information. Upper Saddle River. 2015. Romero C. Emotion 14(2): Analysis Methods (Sage Publications. Master A. Paunesku D. Dweck CS. Accessed July 20. Raudenbush SW. J Dev Econ 84(1):61–75. from structural school switches. . Gralinski JH (1996) Children’s beliefs about intelligence and school perfor. 2013/02/Metodologia-de-Construccion-de-Grupos-Socioeconomicos-SIMCE-2012. Mizala A. Gross JJ (2014) Academic and emo.1073/pnas.pnas. Available at www.17. Zar JH (1996) Biostatistical Analysis (Prentice Hall. Agencia de Calidad (2013) Metodología de Construcción de Grupos Socioeconómicos 19. 23. 227–234. Mizala A. Urquiola M (2007) Socioeconomic status or noise? Tradeoffs Pruebas SIMCE 2012.