Janeli Kotzé 20 September 2016 School choice is everything School quality is heterogeneous and unequally distributed in South Africa Attending a school which performs better on observed measures of quality has a significant causal effect on the academic performance of children ID: 524126
Download Presentation The PPT/PDF document "Performing above expectations" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Performing above expectations
Janeli Kotzé20 September 2016Slide2
School choice is everything
School quality is heterogeneous and unequally distributed in South
Africa
Attending a school which performs better on observed measures of quality has a significant causal effect on the academic performance of children
This emphasises the importance of school choice, especially for poorer learnersQuintile 5 schools are the obvious choice but there are numerous barriers to entry for poor learners (school fees, geographic areas)Enrolling considerably more poor learners in these schools is not feasible as an approach to addressing systemic inequalities in learning.
More effort is required to disrupt systemic dysfunctionality among the majority of poorer schools to provide increased learning opportunities for poor learners. Slide3
Three Research Questions:
How many poor schools perform
consistently
above expectation?
How much learning do poor learners gain by going to these schools? What common factors are associated with these schools?ie: How many are there? What is the benefit of going to them? Which common factors are associated with this
benefit?Slide4
Data
School level Universal ANA data panel (2012, 2013 and 2014)
Track average school performance from 2012 – 2014
Possible to match 17 139 Primary Schools across the 3 years
Learner level Panel (2012, 2013 and 2014; using 2013 V-ANA)Matched to School Level Panel to classify the schoolsPossible to classify 872 (of the 987) schools
Only 30% of learners Matched2011 School Monitoring Survey linked to 2012 ANA school performanceMatched to School Level Panel to classify schoolsPossible to match 1366 of the 1557 primary schools 87.73% Matched
Slide5
Some Caveats about Working with ANA data
Pro’s
- Population based
- Compare sub-groups within a grade and year
Con’s- Possible teacher, school or district level cheating- Not comparable across grades and years. - Linking learners across years are complicated since there is no unique identifierSlide6
Defining An above expectation school
1. Derive a single
indicator of school
performance:
Learner numeracy and literacy scores were averaged for each gradeThese scores were averaged to arrive at a composite measure of performance per grade. Finally the scores were averaged across the grades to create a final composite measure of school performance.
This approach thereby gives equal weight to both numeracy and literacy, and to each grade within a school.2. Define schools as weak performing/ above average performing:Schools that consistently perform at least at the level of the TIMSS low international benchmark.
Schools
that consistently perform amongst the top 25% of all Quintile 1 - 3 schools, excluding small schools Slide7
Defining An above expectation school
Reference group for
the TIMSS low international benchmark group is specifically white and Indian learners of the appropriate age, and not Quintile 5 schoolsSlide8
above expectation schools
Only 6 schools
Only 10 schoolsSlide9
How much learning do poor learners gain?
0.8Slide10
Education production function:
Lagged value-added model with treatment variable:
Estimation Framework
The modelSlide11
Education production function:
Lagged value-added model with treatment variable:
True achievement without measurement error
Inputs such as previous learning
Cumulative shocks to learner productivity
All past and present inputs:
α = input coefficient
Catch all variable: controls for unobserved inputs or endowments
β = persistence coefficient
Treatment Variable: Attending a Well Performing Quintile 1 – 3 School
Estimation Framework
The modelSlide12
Top 25% excl. small schools
Low International Benchmark
Quintile 5
A
B
C
A
B
C
A
B
C
Treatment
0.58***
0.58***
0.57***
0.80***
0.80***
0.81***
0.74***
0.75***
0.63***
0.08
0.08
0.08
0.09
0.09
0.1
0.07
0.07
0.08
Persistence Parameter
0.34***
0.34***
0.33***
0.35***
0.35***
0.33***
0.39***
0.39***
0.37***
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
N
4282
4282
4282
3713
3713
3713
4422
4422
4422
R-squared
0.2850.2890.3170.2370.2420.2730.4040.4070.427Clusters316316316277277277375375375Learner ControlsXXXXXXXXXHousehold ControlsXXXXXXSchool Level ControlsXXX
Basic Value-added Model
Quintile 1-3 learners in different quality schools:Slide13
Robustness ChecksSlide14
School Monitoring Survey
Six Themes:
G
eneral
school characteristicsAccountability systemsSchool governanceSchool managementTeacher
training Provincial and district supportWhich factors could be driving higher performance among poor schools?
Verification ANA
Four Themes:
General school characteristics
Accountability
systems
General
principal and teacher
characteristics
Teacher
training
activities
Classroom practices
Predicting overall school performance
Predicting grade 3 mathematics performance
Predicting school fixed effectsSlide15
School Monitoring Survey
Six Themes:
G
eneral
school characteristicsAccountability systemsSchool governanceS
chool managementTeacher training Provincial and district supportWhich factors could be driving higher performance among poor schools?
S
chool quintile
Ex-department
Small School
Learner-teacher ratio
0.03***Slide16
School Monitoring Survey
Six Themes:
G
eneral
school characteristicsAccountability systemsSchool governanceSchool management
Teacher training Provincial and district supportWhich factors could be driving higher performance among poor schools?
Bureaucratic
: 1. Distance from district office
2. Number of district visits
Market
: Number of neighbouring schools in
10 km radius.
Professional
: Number of Quintile 5 schools in
10 km radius.
0.00***
-0.00*
(including district fixed effects)Slide17
School Monitoring Survey
Six Themes:
G
eneral
school characteristicsAccountability systemsSchool governanceSchool management
Teacher training Provincial and district supportWhich factors could be driving higher performance among poor schools?
Number of SGB Functions filled
0.01***Slide18
School Monitoring Survey
Six Themes:
G
eneral
school characteristicsAccountability systemsSchool governanceSchool management
Teacher training Provincial and district supportWhich factors could be driving higher performance among poor schools?
Number of educators absent
Has an improvement plan
Has an academic improvement plan
Number of academic reports
Has an updated Gr3 class register
Has an LTSM Register
0.03***
-0.02***
0.02***Slide19
School Monitoring Survey
Six Themes:
G
eneral
school characteristicsAccountability systemsSchool governanceSchool management
Teacher training Provincial and district supportWhich factors could be driving higher performance among poor schools?
Self initiated training
School teacher training
External teacher trainingSlide20
School Monitoring Survey
Six Themes:
G
eneral
school characteristicsAccountability systemsSchool governanceSchool managementTeacher
training Provincial and district support
Which factors could be driving higher performance among poor schools?
Received less money than expected
Has a letter stating learner allocation for 2010
Has a letter stating learner allocation for
2011
Has a letter stating learner allocation for
2012
Has at least one vacant position
Subject advisor: Checked curriculum coverage
Subject advisor: Checked
lesson planning
Subject advisor:
Gives advice on teaching
Subject advisor:
Assists with content knowledge
District support index
District monitoring index
-0.02***
0.05***
0.05***
-0.02***
0.02***
-0.01***
0.01***Slide21
Which factors could be driving higher performance among poor schools?
Verification ANA
Four Themes:
General school characteristics
Accountability
systems
General
principal and teacher
characteristics
Teacher
training
activities
Classroom Practices
Parents support the school process
School
LoLT
is an African Language
School has a library
Gr3
Math
School F.E.
1.79*
0.07
-3.37***
0.12
0.51
-0.16Slide22
Which factors could be driving higher performance among poor schools?
Verification ANA
Four Themes:
General school characteristics
Accountability
systems
General
principal and teacher
characteristics
Teacher
training
activities
Classroom Practices
Teacher was observed by an official
Teacher was observed by principal
Teacher was observed by peer
Number neighbours in 10km radius
Number Q5 in 10 km radius
Satisfaction with district support index
Gr3
Math
School F.E.
4.02**
0.47**
-4.17
-0.08
-0.35
-0.43
0.04*
0.00
-0.21***
0.00
1.06*
0.00Slide23
Which factors could be driving higher performance among poor schools?
Verification ANA
Four Themes:
General school characteristics
Accountability
systems
General
principal and teacher
characteristics
Teacher
training
activities
Classroom Practices
Principal over 50 years
Principal has a university degree
Principal has a college diploma
Principal is male
Average teacher age
% Female teachers
Gr3
Math
School F.E.
-0.71
-0.04
-1.26
0.67**
-0.38
0.62*
-0.26
-0.10
-0.13
0.00
0.88
0.25Slide24
Which factors could be driving higher performance among poor schools?
Verification ANA
Four Themes:
General school characteristics
Accountability
systems
General
principal and teacher
characteristics
Teacher
training
activities
Classroom Practices
CAPS through Department
In-Service through school
In-Service externally
In-Service through the Department
Gr3
Math
School F.E.
2.67*
0.34**
0.9
-0.04
2.59*
0.39**
-1.52
0.07Slide25
Which factors could be driving higher performance among poor schools?
Verification ANA
Four Themes:
General school characteristics
Accountability
systems
General
principal and teacher
characteristics
Teacher training activities
Classroom Practices
Teacher covered 90% of curriculum
Number of hours teacher teaches
% administer weekly class test
% administer weekly oral tests
% mark homework regularly
% mark classwork regularly
Gr3
Math
School F.E.
1.56
-0.19
-0.07
0.01
-0.59
0.01
-0.76
-0.33
13.78***
0.24
-0.33
0.09Slide26
Good School Management
Has an LSTM register
Academic Improvement Plan
Supportive district
Isomorphic MimicrySupportive rather than mere monitoring
Bureaucratic accountabilityPrincipal observing lessonsSummary