Title page 1

Contents 1

Abstract 2

1. Introduction 3

2. Research Design and Data 8

2.1. The Education System in Denmark 8

2.2. Survey Design 9

2.3. Procedures and Data 11

3. Results 12

3.1. Characterizing Mental Models of High School Success 12

3.2. Perceived Drivers of Success Gaps: Evidence from Textual Data 16

3.3. Origins of Mental Models 19

3.3.1. Role of the social environment 20

3.3.2. Intra-family transmission 22

3.4. Decision-Relevance of Mental Models: Correlational Evidence 24

3.5. Causal Evidence on the Relevance of Mental Models 27

3.6. Mechanisms 31

4. Conclusion 34

References 35

Online Appendix 41

A. Additional Figures and Tables 42

B. Theoretical Framework 60

C. Coding scheme 65

D. Survey Instructions: Main Survey 72

D.1. Survey of Students 72

D.2. Survey of Parents 87

E. Survey Instructions: Causal Survey 102

E.1. Initial Survey of Female Students (with Treatment) 102

E.2. Follow-up Survey of Female Students (without Treatment) 112

F. Survey Instructions: Mechanism Survey (Students) 114

Tables 21

Table 1. Social environments predict mental models 21

Table 2. Correlations of population-level beliefs within households 23

Table 3. Mental models, self-perception, and choices: correlational evidence 26

Table 4. Effects of information treatment on mental models, self-perception, and choices 30

Figures 13

Figure 1. Actual high school success rates across groups 13

Figure 2. Students' perceptions of gender and SES success gaps 14

Figure 3. Perceived drivers of differences in high school success 17

Figure 4. Effect of a better female vs. male performance on perceived fundamentals 33

Appendix Tables 52

Table A.1. Summary statistics and representativeness of the population 52

Table A.2. Correlates of perceived population-level differences 53

Table A.3. Perceived drivers and quantitative mental models 54

Table A.4. Correlations of self-beliefs within households 55

Table A.5. Robustness: register-based definition of SES 56

Table A.6. Sample characteristics and treatment balance in pre-registered and pooled samples 57

Table A.7. Persistence of treatment effects over time 58

Table A.8. Sample characteristics of mechanism sample 59

Table C.1. Overview of the coding scheme 65

Appendix Figures 42

Figure A.1. Primary school grades and labor market outcomes by track choice 42

Figure A.2. Outline of main survey 43

Figure A.3. Perceived and actual distribution of elementary school grades (math) 44

Figure A.4. Survey interface core module 45

Figure A.5. Students' perceptions of gender success gaps by grade 46

Figure A.6. Students' perceptions of SES success gaps by grade 47

Figure A.7. Parents' perceptions of gender and SES success gaps 48

Figure A.8. Parents' perceived drivers of differences in high school success 49

Figure A.9. Survey interface causal survey 50

Figure A.10. Actual high school success by gender for the 2002 cohort 51

Figure A.11. Definition of belief groups for experimental analysis 51