A presentation by Tahir Andrabi as part of the Societal and Community Influences across the Life Course panel discussion at the International Symposium on Cohort and Longitudinal Studies in Developing Contexts, UNICEF Office of Research - Innocenti, Florence, Italy 13-15 October 2014
The Universal GTM - how we design GTM and dataLayer
Parental perceptions and educational attainment
1. PARENTAL PERCEPTIONS
AND EDUCATIONAL
ATTAINMENT
Tahir Andrabi
October 2014
UNICEF International Symposium on Cohort and
Longitudinal studies
Florence, Italy
2. Pakistan debate
• girls education and gender bias
•Preferences, norms and income disparities
• access
•Largely Ignores variation within girls,
within households
3. Education and rate of return
• Expected rate of return
• Rob Jensen, Nguyen
• Parental investments and decision-making
• Becker and Tomes, Cunha and Heckman
• Perceptions on the market and child ability
(and preferences)
• Mansky 2004
4. What we do
• Use LEAPS endpoints of 7 year
longitudinal study: 5 rounds:
•2004, 05, 06, 07 and 11
•Maternal rating of perceived intelligence
of children 5-15 in year 2004.
•How intelligent is ____?
• 5 point Likert scale (converted to 3
categories: Above Average, Average,
Below Average)
5. What we do, contd.
•Examine effect of intelligence rating on
educational and marital status in year
2011
•Household fixed effects
• Within gender variation
6. What we find
• Perceptions matter
• for both boys and girls: enrollment,
educational attainment
• Girls likelihood of being married
• The difference between low ranked girls
and high ranked girls is greater than the
gender gap in enrollment (roughly 10
percentage points)
7. The LEAPS Study
• A comprehensive look at educational environment
of children
• Representative 120 villages in 3 districts in Punjab
• Linked school and household study
• 1800+ randomly selected households
• 800 schools, 4000 teachers
• 25,000 children tested in schools
• 2 cohorts in schools
• 5 waves: 2004, 05, 06, 07, 11
8. Attrition
• HH level attrition 13.5% over 7 years
• Mostly households migrated
• Member level: we know what happened to the
missing member.
• Marriage
• Migrated, etc.
• Can find out something about their
educational attainment if they left at the time
of the 2007 survey.
9. Perceptions
• Conceptual Discussion:
• Ex ante, ex post, endogenous?
• Mechanisms
• Leave till the end
• Are they based on some objective criterion?
10. Are perceptions accurate?
Test Scores and Intelligence
(1) (2) (3)
VARIABLES urdu_theta_mle eng_theta_mle math_theta_mle
Intelligence:
Average
0.261 0.424** 0.176
(0.194) (0.213) (0.184)
Intelligence:
Above Average
0.779*** 0.834*** 0.500***
(0.197) (0.212) (0.183)
Constant -1.133*** -1.609*** -1.044***
(0.253) (0.349) (0.329)
Observations 813 813 813
R-squared 0.286 0.335 0.300
Notes:OLS regression with standard errors clustered at the village level; village fixed effects and
child age indicator dummy variables. The omitted category for dummy variable is Perceived
Intelligence--Low Robust standard errors in parentheses, clustered at the village level ***
p<0.01, ** p<0.05, * p<0.1
11. Perceived Intelligence, by Gender
3000
2500
2000
1500
1000
500
0
Boy Girl
Above Average
Average
Below Average
12. Perceived Intelligence
by Maternal Education
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
No Yes
Mother Educated?
Above Average
Average
Below Average
13. Are uneducated mothers different in their
perceptions?
0
-1 -.5
How Accurate are Parental Perceptions?
Child Perceived Intelligence vs. Test Scores
Below Average Average Above Average Below Average Average Above Average
Mother Not Educated Mother Educated
Source: LEAPS Household Survey 2004
14. Enrollment Status 2011 and Intelligence 2004
(1) (2) (3) (4) (5) (6)
VARIABLES enrolled11 enrolled11m enrolled11 enrolled11m enrolled11 enrolled11m
enrolled in 2004 enrolled in 2004
Male, Below
Average
0.121*** 0.134*** 0.130*** 0.144*** 0.149** 0.178***
(0.0443) (0.0402) (0.0430) (0.0394) (0.0601) (0.0586)
Male, Average 0.164*** 0.164*** 0.157*** 0.164*** 0.156*** 0.178***
(0.0391) (0.0319) (0.0385) (0.0317) (0.0476) (0.0442)
Male, Above
Average
0.200*** 0.205*** 0.189*** 0.202*** 0.197*** 0.223***
(0.0430) (0.0380) (0.0431) (0.0386) (0.0523) (0.0496)
Female, Average 0.0579 0.0646** 0.0547 0.0655** 0.0598 0.0757*
(0.0360) (0.0287) (0.0355) (0.0285) (0.0438) (0.0402)
Female, Above
Average
0.133*** 0.127*** 0.128*** 0.127*** 0.138*** 0.147***
(0.0390) (0.0325) (0.0389) (0.0326) (0.0477) (0.0433)
Enrolled 04 0.0365 0.0136
(0.0270) (0.0230)
Constant 0.682*** 0.662*** 0.665*** 0.653*** 0.750*** 0.727***
(0.0460) (0.0403) (0.0487) (0.0427) (0.0554) (0.0549)
Observations 4,081 4,462 4,056 4,436 3,246 3,421
R-squared 0.653 0.644 0.655 0.645 0.679 0.672
Notes: OLS regression girls aged 5-15 years in 2004. Standard errors clustered at village level with household
fixed effects and child age indicator dummy variables. The omitted categories for dummy variables are Girl and
Perceived Intelligence--Low. Robust standard errors in parentheses, clustered at the village level. *** p<0.01, **
p<0.05, * p<0.1
15. Educational Attainment and Intelligence
(Middle: 8th grade and above)
(1) (2) (3)
VARIABLES middle middle middle
Age 2011 12-17 Age 2011 12-17 Age 2011 12-17
Male, Below Average 0.0376 0.0395 0.0587
(0.0803) (0.0910) (0.130)
Male, Average 0.164** 0.156** 0.191
(0.0667) (0.0760) (0.119)
Male, Above Average 0.227*** 0.270*** 0.300**
(0.0766) (0.0776) (0.116)
Female, Average 0.101 0.133* 0.189
(0.0708) (0.0766) (0.116)
Female, Above Average 0.177** 0.194** 0.223**
(0.0788) (0.0798) (0.112)
Enrolled 04 0.421***
(0.0571)
Constant -0.109 0.0312 0.453***
(0.0687) (0.0842) (0.112)
Observations 2,509 2,471 1,950
16. Marital Status and Intelligence
(1) (2) (3) (4)
VARIABLES married married married married
age
2011>=16
age
2011>=16
age 2011>=16
Enrolled in 2004
Male, Below Average -0.157*** -0.281*** -0.272*** -0.190**
(0.0300) (0.0613) (0.0625) (0.0879)
Male, Average -0.151*** -0.258*** -0.237*** -0.179**
(0.0287) (0.0587) (0.0598) (0.0819)
Male, Above Average -0.150*** -0.224*** -0.200*** -0.155*
(0.0301) (0.0584) (0.0592) (0.0826)
Female, Average -0.0189 -0.0560 -0.0457 -0.0423
(0.0315) (0.0602) (0.0607) (0.0874)
Female, Above Average -0.0380 -0.104* -0.0881 -0.0561
(0.0323) (0.0612) (0.0625) (0.0811)
Enrolled 04 -0.0755**
(0.0354)
Constant 0.0838*** 0.210*** 0.260*** 0.146*
(0.0262) (0.0567) (0.0629) (0.0806)
17. Further Work
• Experimentally move rate of return through information
• See what happens to parental aspirations about their girls’
education
• And differential effect based on girls perceptions
18. Discussion
• Discussion
• Efficiency, equity
• Room for Policy: government vs households
• Perceptions
• Ex ante, ex post, endogenous
Notas del editor
10% difference in gender, note small initial enrollment effect because of delayed enrollment.