Understanding education in South Africa
Inequality, outcomes and resources
Nic Spaull
Equal Education (UCT)
5 March 2013
Outline
① Overview of theoretical links between schooling
and the labour market
② What does the data say?
③ Are resources the main constraint?
④ Conclusions & Recommendations
2
Not all schools are born equal
?
Pretoria Boys High School
SA public schools?
3
Inequality - SA
Why has there
been so little
progress in
equalizing the
income
distribution in SA?
4Murray Leibbrandt & Ingrid Woolard & Arden Finn & Jonathan Argent, 2010. "Trends in South African Income Distribution and
Poverty since the Fall of Apartheid," OECD Social, Employment and Migration Working Papers 101, OECD Publishing.
Education and
inequality?
Quality of
education
Duration
of
education
Type of
education
SA is one of the
top 3 most
unequal
countries in
the world
Between 78%
and 85% of
total inequality
is explained by
wage
inequality
Wages
• IQ
• Motivation
• Social
networks
• Discrimination
Does schooling mitigate or
propagate inequality?
6
LOW
ability/motivat
ion
HIGH
ability/motivat
ion
High
SES
Low
SES
Early Childhood
Development (ECD)
Home
background
School
Labor
Market
Post-school
Society stratified
on wealth/SES
Society stratified on
ability/motivation
Fair system where labor market outcomes are
largely a function of ability, motivation, talent etc.
7
LOW
ability/motivat
ion
HIGH
ability/motivat
ion
High
SES
Low
SES
Early Childhood
Development (ECD)
Home
background
School
Labor
Market
Post-school
Society stratified
on wealth/SES
Society stratified on
wealth/SES
UNfair system where labor market outcomes are
largely a function of parental wealth & education
8
High productivity jobs
and incomes (17%)
•
•
•
Mainly professional,
managerial & skilled jobs
Requires graduates, good
quality matric or good
vocational skills
Historically mainly white
Type
Labour Market
University/
FET
•
•
•
•
Vocational training
Affirmative action
•
-
High SES
background
+ECD
Minority
(20%)
Big demand for good
schools despite fees
Some
scholarships/bursaries
Unequal
society
Majority
(80%)
Low quality
secondary
school
Low SES
background
Often manual or low skill
jobs
Limited or low quality
education
Minimum wage can exceed
productivity
Low quality
primary
school
Attainment
•
High
quality
primary
school
-
Low productivity jobs &
incomes
•
Type of institution
(FET or University)
Quality of institution
Type of qualification
(diploma, degree etc.)
Field of study
(Engineering, Arts etc.)
Some motivated, lucky or
talented students make the
transition
Quality
•
•
High
quality
secondary
school
cf. Servaas van der Berg – QLFS 2011
9
OK that’s the theory but
what does the data say?
NB: “Without data you are just another person
with an opinion” - Schleicher
10
The impact of SES on reading/maths
•
Indication of wasted human
capital potential (see
Schleicher, 2009)
550
SEY MAU
SWA
KEN
ZAN
BOT
500
SACMEQ
ZIM
NAM
SOU
UGA MOZ
LES
450
In South Africa socioeconomic
status largely determines
outcomes (with a very small
number of exceptions – see
newspapers for examples)
TAN
ZAM
MAL
400
•
Almost 40% of SA student
reading achievement can be
explained by socioeconomic
status (31 assets, books,
parental education) alone.
Average SACMEQ reading score
•
600
(SACMEQ III – 2007 Gr 6)
0
10
20
30
40
Percentage of variance in performance explained by household socioeconomic status (r-squared X 100)
Spaull, 2013
Gr 1 - Gr 2 - Gr 3 – Gr 4 – Gr 5 – Gr 6 – Gr 7 – Gr 8 – Gr 9 - Gr 10 – Gr 11 – Gr 12
Foundation Phase
Intermediate Phase
Senior Phase
FET Phase
Taylor, 2011 (based on NSES)
12
Gr 1 - Gr 2 - Gr 3 – Gr 4 – Gr 5 – Gr 6 – Gr 7 – Gr 8 – Gr 9 - Gr 10 – Gr 11 – Gr 12
Foundation Phase
Intermediate Phase
Senior Phase
FET Phase
prePIRLS 2011 Benchmark Performance by Test Language
47
Xitsonga
53
53
Tshivenda
47
24
siSwati
0
0
76
0.25
Setswana
34
66
0.1
Sesotho
36
64
0.1
57
Sepedi
43
29
isiZulu
71
38
isiXhosa
0.8
0.4
62
31
isiNdebele
0
69
0.2
English
10
90
19
Afrikaans
12
88
15
South Africa
29
Did not reach
High International Benchmark
These grade 4 children cannot read
71
6
Low International benchmark
Advanced International benchmark
Intemediate International Benchmark
Howie & van Staden (2012)
13
Foundation Phase
Intermediate Phase
Senior Phase
Middle-income countries
Quintile 1
Quintile 2
Quintile 3
Quintile 4
Quintile 5
Independent
Russian Federation
Lithuania
Ukraine
Kazakhstan
Turkey
Iran, Islamic Rep. of
Romania
Chile
Thailand
Jordan
Tunisia
Armenia
Malaysia
Syrian Arab Republic
Georgia
Palestinian Nat'l Auth.
Macedonia, Rep. of
Indonesia
Lebanon
Botswana (Gr 9)
Morocco
Honduras (gr 9)
South Africa (Gr 9)
Ghana
TIMSS 2011 Science score
Gr 1 - Gr 2 - Gr 3 – Gr 4 – Gr 5 – Gr 6 – Gr 7 – Gr 8 – Gr 9 - Gr 10 – Gr 11 – Gr 12
FET Phase
600
560
520
480
440
400
360
320
280
240
200
South Africa (Gr9)
14
Insurmountable learning deficits
Gradients of achievement in the EASTERN Cape and in Quintile 5 (National)
13
12
Desired goal
Performance below “on-track” line creates
increasing gradient of expectation
12
11
10
9
e
lin
k
ac
-tr
n
O
8
7
Projected matric performance: Eastern Cape
Projected matric performance: Quintile 5
6
6
5
5
4
0
Gr1
Gr2
SACMEQ III Eastern Cape
SACMEQ III Quintile 5
1
NSES EC
NSES Quintile 5
2
Initial
conditions
NSES EC
NSES Quintile 5
3
3
Gr3
Gr4
Gr5
Gr6
ck
-tra
line
TIMSS 2011 Eastern Cape
TIMSS 2011 Quintile 5
Off
4
NSES EC
NSES Quintile 5
Effective grade level
9
f
e o ble
n
Zo oba
ss
pr
im rogre
p
Gr7
Actual grade
Gr8
Gr9
Gr10
Gr11
NB: Key assumption,
0.5 SD of national
learning achievement
is equivalent to one
grade level of learning
-agreement from
TIMSS/PIRLS
Gr12
C.f. Lewin (2007: 8)
Spaull 2013
Spaull, 2013
Insurmountable learning deficits
Gradients of achievement in the WESTERN Cape and in Quintile 5 (National)
13
12
Desired goal
Performance below “on-track” line creates
increasing gradient of expectation
12
11
10
9
e
lin
k
ac
-tr
n
O
7
6
6
f
Of
5
e
lin
k
c
a
-tr
Projected matric performance: Western Cape
Projected matric performance: Quintile 5
8
5
NSES WC
NSES Quintile 5
NSES WC
NSES Quintile 5
SACMEQ III Western Cape
SACMEQ III Quintile 5
Gr3
Gr4
Gr5
Gr6
3
3
2
Initial
conditions
1
0
Gr1
Gr2
TIMSS 2011 Western Cape
TIMSS 2011 Quintile 5
4
4
NSES WC
NSES Quintile 5
Effective grade level
9
Gr7
Actual grade
Gr8
Gr9
Gr10
Gr11
NB: WC has
relatively high % of
Q5 schools thus it
should be more
convergent by
construction.
Gr12
C.f. Lewin (2007: 8)
Spaull 2013
Spaull, 2013
Matric pass rate
Media sees only this
What are the root
causes of low and
unequal achievement?
MATRI
C
Pre-MATRIC
HUGE learning deficits…
17
Gr 1 - Gr 2 - Gr 3 – Gr 4 – Gr 5 – Gr 6 – Gr 7 – Gr 8 – Gr 9 - Gr 10 – Gr 11 – Gr 12
Intermediate Phase
Senior Phase
Matric
Of 100 students that enroll in grade 1
approximately 50 will make it to matric, 40
will pass and 12 will qualify for university
Inequality
•
•
•
Subject combinations differ between rich and
poor – differential access to higher education
Maths / Maths-lit case in point
Are more students taking maths literacy
because THEY cannot do pure-maths, or
because their TEACHERS cannot teach puremaths?
Number of students
•
Grade 10 (2 years earlier)
Grade 12
Those who pass matric
Pass matric with maths
Proportion of matrics taking mathematics
• Grade 12 – Various
• Roughly half the cohort
____________________________________
Underperformance
FET Phase
1200000
60%
1000000
50%
800000
40%
600000
30%
400000
20%
200000
10%
0
0%
Matric 2008 (Gr Matric 2009 (Gr Matric 2010 (Gr Matric 2011 (Gr
10 2006)
10 2007)
10 2008)
10 2009)
18
Proportion of matrics (%)
Foundation Phase
Are more resources the answer?
19
Resources?
SACMEQ III (2007) Grade 6 performance by
school library (41% library access)
School has a library
School has no library
600
550
Does this make
sense? What are the
500
450
ways that
400
library-
access
improves
reading/maths
performance?
350
300
250
200
150
100
50
0
Reading
Maths
20
Resources
SACMEQ III (2007) Grade 6 performance by
school sports ground (35% access)
School has a sports ground
School has no sports ground
550
Does this make
sense? What are the
500
450
400
ways that
350
ground
300
sports
access
improves
reading/maths
performance?
250
200
150
100
50
0
Reading
Maths
21
SACMEQ III (2007) Grade 6 performance by
school website (18% access)
School has a website
School has no website
650
Does this make
sense? What are the
600
550
500
ways that website
access
improves
reading/maths
performance?
450
400
350
300
250
200
150
100
50
0
Reading
Maths
22
Library
0.09
0.32
0.29
0.61
0.90
Q1
Q2
Q3
Q4
Q5
Sports Ground
0.5
0.57
0.52
0.77
0.99
Website
0.02
0.09
0.15
0.15
0.50
SACMEQ III (2007)
Grade 6 performance
by school library (41%
library access)
SACMEQ III (2007) Grade
6 performance by school
sports ground (35%
access)
SACMEQ III (2007)
Grade 6 performance
by school website (18%
School has a library
School has a sports ground
School has a website
School has no library
School has no sports ground
School has no website
600
650
600
550
500
450
400
350
300
250
200
150
100
50
0
550
550
500
500
450
450
400
400
350
350
300
300
250
250
200
200
150
150
100
100
50
50
0
0
Reading
Reading
Maths
access)
Maths
Reading
Maths
23
Correlation is not causation
• Given that apartheid discriminated along racial, linguistic
and spatial dimensions, it is unsurprising that these are all
correlated.
• Cannot draw causal inference from bivariate statistics (need
multivariate analysis)
• Serious equalization of public expenditure on education yet
almost no change in outcomes since the transition?
– Yes there is significant pvt spending at the top, but isolating the
lack of resources is an easy target but NOT the most important
one. Looking at things like teacher content knowledge,
curriculum coverage, accountability and capacity  more
difficult but more important.
– For example, teacher content knowledge….
24
Gr 1 - Gr 2 - Gr 3 – Gr 4 – Gr 5 – Gr 6 – Gr 7 – Gr 8 – Gr 9 - Gr 10 – Gr 11 – Gr 12
Foundation Phase
Mean
Intermediate Phase
Senior Phase
Lower bound confidence interval (95%)
FET Phase
Upper bound confidence interval (95%)
950
Maths-teacher mathematics score
900
KEN
850
Q5-SOU
ZIM
800
SWA
750
LES ZAM
700
650
600
MOZ
MAL
SOU NAM
SEY
TAN UGA
BOT
Q4-SOU
Q3-SOU
Q2-SOU
Q1-SOU
ZAN
25
Gr 6 Teacher Content Knowledge - see McKay & Spaull (2013)
Teacher knowledge
SACMEQ III (2007)  401/498 Gr6 Mathematics teachers
7
Correct answer
(7km):
38% of Gr 6
Maths teachers
SACMEQ Maths
teacher test Q17
Correct
1
23%
2
22%
Quintile
3
38%
4
40%
5
74%
Avg
38%
2 education
systems
26
Accountability
“Without an explicit chain of accountability, student outcomes cannot improve on a national level.
While it is indisputable that the government should provide every school with the basic infrastructure
required (water, sanitation, buildings, textbooks etc.) and support teachers and principals, one
cannot absolve schools of responsibility for dysmal performance because
they do not have libraries or science laboratories. Under the right circumstances these
However,
libraries and laboratories are not a prerequisite for basic school functionality and
adequate learning outcomes – as many excellent under-resourced schools prove. How is it possible that two
do improve learning outcomes, and ultimately they should be provided to all schools.
equally poor schools with socioeconomically similar students perform at vastly different levels – one
dysfunctional, the other excellent? Commenting on the impact of the influential “Coleman Report” in
America, Coleman explains that one of the main impacts of the report was to shift the policy-consensus
in the United States towards educational outcomes rather than educational inputs (see below) –
something which is also needed in South Africa.
“The long range impact of the report will probably be to strengthen the move toward
evaluating schools in terms of their results rather than their inputs...School
superintendents and educators have been reluctant to measure schools by how well the
students do. Whether or not they admit it, they feel that the primary variation in
student performance is not what the schools are doing but what the child comes to
school with” (Coleman, 1972, p. 13).
Excerpt from Spaull 2013 CDE report “South Africa’s Education Crisis”
27
Conclusions & Implications
The 2 binding constraints to progress in
education South Africa are
Low
quality
education
Low
social
mobility
1. Lack of accountability
2. Lack of capacity
At all levels:
school/district/province/national.
Hereditary
poverty
Persistent patterns of poverty
and privilege
Lack of capacity means that
teachers/administrators/managers lack
the ability/skills to fulfill their job
descriptions
Lack of accountability means that there
are no tangible consequences for
nonperformance
28
Use your time at UCT wisely
29
Further reading
• www.nicspaull.com/research
• CDE 2013 report on education “South Africa’s
Education Crisis” (end of March)
• Low Quality as a Poverty Trap report
http://resep.sun.ac.za/index.php/projects/ )
• Improving Education Quality in South Africa (report to
the NPC for the NDP)
http://resep.sun.ac.za/index.php/projects/ )
30
Thanks.
Comments & questions…
Website: nicspaull.com/research
Email:
nicholasspaull@gmail.com
Twitter: @NicSpaull
31
.005
Kernel Density of Literacy Score by Race (KZN)
.006
.004
Density
.003
.002
.002
0
20
40
60
Literacy score (%)
Black
Indian
80
0
0
0
.001
.005
.01
.015
kdensity reading test score
.004
.02
U-ANA 2011
100
0
0
200
White
Asian
400
reading test score
600
200
800
Poorest 25%
Second wealthiest 25%
English/Afrikaans schools
African language schools
400
600
Learner Reading Score
800
1000
Second poorest 25%
Wealthiest 25%
.025
PIRLS / TIMSS / SACMEQ / NSES / ANA / Matric… by Wealth / Language / Location / Dept…
Kernel Density of School Literacy by Quintile
.01
.02
Density
.015
.01
0
0
0
Density
.03
.02
.04
U-ANA 2011
.005
Density
.008
Bimodality – indisputable fact
0
20
40
60
Numeracy score 2008
Ex-DET/Homelands schools
80
Historically white schools
100
20
40
60
Average school literacy score
Quintile 1
Quintile 3
Quintile 5
80
100
Quintile 2
Quintile 4
32
Gr 1 - Gr 2 - Gr 3 – Gr 4 – Gr 5 – Gr 6 – Gr 7 – Gr 8 – Gr 9 - Gr 10 – Gr 11 – Gr 12
Intermediate Phase
Senior Phase
FET Phase
0
.002
• Grade 6 – Numeracy and literacy
• 392 schools, 9071 students
____________________________________
Density
SACMEQ 2007
.004
.006
.008
Foundation Phase
Underperformance
•
•
27% of students functionally illiterate
SA performs worse than many low-income
African countries (Tanzania, Kenya, Swaziland,
Zimbabwe)
No improvement between SACMEQ II (2000)
and SACMEQ III (2007)
Although majority (98%) of students are
enrolled, sometimes almost no learning
Inequality
•
•
Large differences between quintiles (see table
later)
Large inequalities in maths teacher content
knowledge
200
400
600
Learner Reading Score
Poorest 25%
Second wealthiest 25%
Mean
Lower bound confidence interval (95%)
800
1000
Second poorest 25%
Wealthiest 25%
Upper bound confidence interval (95%)
950
Maths-teacher mathematics score
•
•
0
900
KEN
850
ZIM
UGA
TAN
SWASEY
BOT
NAM
MALSOU
LESZAMMOZ
800
750
700
650
600
Q5-SOU
Q4-SOU
Q3-SOU
Q2-SOU
Q1-SOU
ZAN
33
Gr 6 Teacher Content Knowledge - see McKay & Spaull (2013)
Gr 1 - Gr 2 - Gr 3 – Gr 4 – Gr 5 – Gr 6 – Gr 7 – Gr 8 – Gr 9 - Gr 10 – Gr 11 – Gr 12
Intermediate Phase
•
•
Avg Q1/Q2 Gr9 student is 3yrs (4yrs) worth of
learning behind the average Q5 student in
maths (science)
Avg Gr 9 student in ECA is 2yrs worth of
learning behind avg Gr9 student in GAU
*Data now available for download
443
433
352
276
275
264
285
1995
1999
2002
2002
Grade 8
2011
Grade 9
332
2011
260
243
244
268
1995
1999
2002
2002
TIMSS
middleincome
country
Gr8
mean
TIMSS Mathematics
600
560
520
480
440
400
360
320
280
240
200
Grade 8
2011
Grade 9
2011
TIMSS
middleincome
country
Gr8
mean
TIMSS Science
Middle-income countries
Quintile 1
Quintile 2
Quintile 3
Quintile 4
Quintile 5
Independent
Underperformance
•
76% of Gr9 students had not acquired a
basic understanding about whole
numbers, decimals, operations or basic
graphs (i.e. had not reached low int.
benchmark)
•
Avg. Gr 9 SA student is 2yrs (2.8yrs)
behind the average Gr8 student from a
middle income country in maths
(science)
•
Contrary to popular belief, even South
Africa’s “top” schools do not perform
well by international standards…
Inequality
480
440
400
360
320
280
240
200
160
120
80
40
0
FET Phase
Russian Federation
Lithuania
Ukraine
Kazakhstan
Turkey
Iran, Islamic Rep. of
Romania
Chile
Thailand
Jordan
Tunisia
Armenia
Malaysia
Syrian Arab Republic
Georgia
Palestinian Nat'l Auth.
Macedonia, Rep. of
Indonesia
Lebanon
Botswana (Gr 9)
Morocco
Honduras (gr 9)
South Africa (Gr 9)
Ghana
• Grade 9 – Maths and science
• 285 schools, 11969 students
____________________________________
TIMSS score
TIMSS 2011
Senior Phase
TIMSS 2011 Science score
Foundation Phase
34
South Africa (Gr9)
Poverty & Privilege paper
• If there are indeed two education systems in
SA and not one – as the background research
seems to suggest - where/how do we draw the
line between one and the other?
35
See Taylor (2012)
Matric
1200000
• More students making it to grade 10 but
not more making it to matric
•
Partially due to less repetition at lower grades
• LARGE differences in the ability of
provinces to “convert” grade 1
enrolments into matric passes
• Why are more students taking maths
literacy?
1000000
800000
600000
400000
200000
0
grade 10
Grade 12
The ratio of grade 2 enrolments ten years prior to matric to matric passes by province
See Taylor (2012)
SACMEQ III (Spaull & Taylor, 2012)
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
2
6
44
9
13
18
25
26
52
53
61
39
58
45
30
14
12
18
11
11
27
17
2
8
34
50
54
8
5
1
11
62
50
31
19
37
13
7
3
15
Literacy
Enrolled and acquired higher order reading skills (Levels 6-8) by grade 6
Enrolled and acquired basic reading skills (Levels 3-5) by grade 6
Enrolled but functionally illiterate (Levels 1-2) by grade 6
Never enrolled or dropped out prior to Grade 6
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
1
0
29
34
59
3
2
2
46
50
53
50
10
10
5
58
64
77
11
15
8
11
13
71
51
44
12
8
14
7
37
11
34
11
39
24
2
8
Numeracy
Enrolled and acquired higher order numeracy skills (Levels 6-8) by grade 6
Enrolled and acquired basic numeracy skills (Levels 3-5) by grade 6
Enrolled but functionally innumerate (Levels 1-2) by grade 6
Never enrolled or dropped out prior to grade 6
11
5
Background: Data
SACMEQ
 Southern and Eastern African Consortium for Monitoring Educational Quality
 14 participating countries
 SACMEQ II (2000), SACMEQ III (2007)
 Background survey
 Testing :
o Gr 6 Numeracy
o Gr 6 Literacy
o HIV/AIDS Health knowledge
SACMEQ III: South Africa
 9071 Grade 6 students
 1163 Grade 6 teachers
 392 primary schools
•
See SACMEQ website for research
•
(Spaull, 2011)
0
• Grade 6 [2007]
• Data: SACMEQ
.002
Density
Socioeconomic
Status
.004
.006
.008
Two school systems not one?
0
200
400
600
Learner Reading Score
Poorest 25%
Second wealthiest 25%
800
1000
Second poorest 25%
Wealthiest 25%
39
.006
.004
0
.002
Latest data? ANA?
Teacher knowledge
Teacher absenteeism
Textbook access
Literacy/numeracy rates
Grade repetition
Parental education
Homework frequency
Density
•
•
•
•
•
•
•
•
.008
Corroborating evidence?
0
200
400
600
Learner Reading Score
Poorest 25%
Second wealthiest 25%
800
1000
Second poorest 25%
Wealthiest 25%
40

In most government
reports outcomes and
inputs are not usually
reported by quintile,
only national averages

41
Implications for reporting and modeling??
42
0
.002
Density
Government reporting – means are
misleading
.004
.006
.008
Do the ends justify the means?
0
200
400
600
Learner Reading Score
Poorest 25%
Second wealthiest 25%
800
1000
Second poorest 25%
Wealthiest 25%
SACMEQ III Reading scores: Mean – Median
SACMEQ Standard deviation approx 100
35
31
30
25
20
15
15
10
14
12
10
17
16
10
7
5
5
2
0
0
-5
-10
BOT
KEN
LES
MAL
MOZ NAM
SOU
SWA
TAN
UGA
ZAM
ZAN
ZIM
-6
43
Teacher knowledge...
 Q6: 53%
correct (D)
Q9: 24% correct (C)
English Q9: 57% correct (D)
44
Modelling student performance
• Two data-generating processes. Little reason
to believe there is the same underlying DGP
• Split samples
– Wealthiest 25% of schools
– Poorest 75% of schools
• Which coefficients are large & significant
across the two regressions?
45
11/29 variables common
5/27 variables common
All of these things have serious
ramifications for the labour market
and thus income inequality…
48
40%
30%
20%
Working-Age Population
All Youth
Youth with Less than Matric
With Matric
Youth With Diploma
Youth With Degree
2011
2010
2009
2008
2007
2006
2005
2004
2003
2002
2001
2000
1999
1998
1997
1996
1995
0%
With Less Than Matric
With Matric
With Diploma
With Degree
Composition of 18 - 24-year-olds by highest level of education completed (Van Broekhuizen, 2013)
2011
2010
2009
2008
2007
2006
2005
2004
2003
2002
2001
2000
1999
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
1998
Demand-Supply
interaction between
labour market and
schooling
50%
1997
•
60%
10%
Employment rates of
youth are stagnant at
20% since 1995
For 18-24yr matriconly or less-thanmatric doesn’t make
much of a difference
70%
1996
•
80%
1995
•
Since 2007 more
youth have matric
Proportion of youth with Qualification
•
Employment/LFA Rate for
18 - 24 -year -olds
Percentage of youth in employment by highest educational attainment (Van Broekhuizen, 2013)
Percentage of unemployed 18-24-yr olds that have never worked (Van Broekhuizen, 2013)
50%
40%
30%
20%
10%
All Youth
Less Than Matric
Matric
2011
0%
2010
Durations of youth
unemployment have
been increasing since
1995
60%
2009
•
In 2011 more than
70% of youth in
unemployment had
never been employed
before
70%
2008
•
Percentage of unemployed
18-24-year-olds
80%
Diploma
Degree
40%
35%
•
Proportion of jobseeking youth
unemployed for more
than 3 years has
increased from 42% in
2008 to 50% in 2011
30%
25%
20%
15%
10%
5%
0%
2008
< 6 months
2009
6months - 1 year
2010
1 - 3 years
2011
3 - 5 years
Percentage of unemployed 18-24-year-olds by duration of unemployment
(Van Broekhuizen, 2013)
> 5years
ECD
Gr 1 - Gr 2 - Gr 3 – Gr 4 – Gr 5 – Gr 6 – Gr 7 – Gr 8 – Gr 9 - Gr 10 – Gr 11 – Gr 12
Foundation Phase
Foundation Phase
Intermediate Phase
Intermediate Phase
Senior Phase
Senior Phase
FET Phase
FET Phase
Mother-tongue instruction
De facto / De jure ?
Primary school
High school
Main drop-out zone
51
Conclusions & Implications
We have 2 education systems in South Africa
1. Reporting education statistics in SA
– Averages are uniquely misleading in SA – the average child does not exist in any
meaningful sense
– Report educational statistics by quintile in addition to province
– You can’t solve a problem that doesn’t officially exist
2. Modelling educational performance in SA
–
Modelling a single education system when there are two can lead to spurious results
3. Policy differentiation
–
–
–
Policies suited to one system are not necessarily suited to the other
Don’t interfere with high-performing schools If it aint broke don’t (try) fix it
LITNUM intervention in WC  Blanket approach
52
When faced with an exceedingly low and
unequal quality of education do we….
A) Increase accountability {US model}
• Create a fool-proof highly specified, sequenced curriculum (CAPS/workbooks)
• Measure learning better and more frequently (ANA)
• Increase choice/information in a variety of ways
B) Improve the quality of teachers {Finnish model}
• Attract better candidates into teaching degrees  draw candidates from the
top (rather than the bottom) of the matric distribution
• Increase the competence of existing teachers (Capacitation)
• Long term endeavor which requires sustained, committed, strategic,
thoughtful leadership (something we don’t have)
C) All of the above {Utopian model}
•
Perhaps A while we set out on the costly and difficult journey of B??
53
More research on…
• Things we don’t really understand:
– The impact of LOLT switching at grade 4 – should we
be teaching in mother-tongue for longer?
– Why districts are so dysfunctional
– How to do in-service teacher training (existing
programs don’t work)
– The existing system of incentives facing national and
provincial bureaucrats/politicians and how to change
this
– What the SA version of an accountable system of
education looks like Nic’s PhD
54
References
•
•
•
•
•
•
•
•
•
•
Fleisch, B. (2008). Primary Education in Crisis: Why South African schoolchildren underachieve in reading and mathematics. Cape
Town. : Juta & Co.
Hoadley, U. (2010). What doe we know about teaching and learning in primary schools in South Africa? A review of the classroombased research literature. Report for the Grade 3 Improvement project of the University of Stellenbosch. Western Cape Education
Department.
Hungi, N., Makuwa, D., Ross, K., Saito, M., Dolata, S., van Capelle, F., et al. (2011). SACMEQ III Project Results: Levels and Trends in
School Resources among SACMEQ School Systems. Paris: Southern and Eastern Africa Consortium for Monitoring Educational
Quality.
Ross, K., Saito, M., Dolata, S., Ikeda, M., Zuze, L., Murimba, S., et al. (2005). The Conduct of the SACMEQ III Project. In E. Onsomu,
J. Nzomo, & C. Obiero, The SACMEQ II Project in Kenya: A Study of the Conditions of Schooling and the Quality of
Education. Harare: SACMEQ.
Shepherd, D. (2011). Constraints to School Effectiveness: What prevents poor schools from delivering results? Stellenbosch
Economic Working Papers 05/11. [PIRLS]
Spaull, N. (2011a). A Preliminary Analysis of SACMEQ III South Africa.Stellenbosch Economic Working Papers.
Spaull, N. (2011). Primary School Performance in Botswana, Mozambique, Namibia and South Africa. Paris: Southern and Eastern
African Consortium for Monitoring Educational Quality (SACMEQ) Working Paper no.8.
Spaull, N. 2012 Equity & Efficiency in South African primary schools : a preliminary analysis of SACMEQ III South Africa Masters
Thesis. Economics. Stellenbosch University
Taylor, S. (2011). Uncovering indicators of effective school management in South Africa using the National School Effectiveness
Study.Stellenbosch Economic Working Papers 10/11, 1-51. [NSES]
Van der Berg, S., Burger, C., Burger, R., de Vos, M., du Rand, G., Gustafsson, M., Shepherd, D., Spaull, N., Taylor, S., van
Broekhuizen, H., and von Fintel, D. (2011). Low quality education as a poverty trap. Stellenbosch: University of Stellenbosch,
Department of Economics. Research report for the PSPPD project for Presidency.
55
Performance of quintile five schools in TIMSS 2003 Maths
.004
0
.002
Density
.006
.008
– see Taylor MST (2011)
0
200
400
Grade 8 mathematics score
South Africa Quintile 5
Chile Quintile 5
Singapore Quintile 5
600
800
Chile
Singapore
Even Q5 schools in SA perform at a comparatively low level
57
Description of Range on 500
levels
point scale
Level 1
Pre-reading
< 373
Skills
Level 2
Emergent
reading
Matches words and pictures involving prepositions and abstract concepts; uses cuing
systems (by sounding out, using simple sentence structure, and familiar words) to
interpret phrases by reading on.
373  414
[1]
Level 3
Basic reading
Level 4
Reading
for
meaning
Level 5
Interpretive
reading
Level 6
Inferential
reading
Level 7
Analytical
reading
Matches words and pictures involving concrete concepts and everyday objects. Follows
short simple written instructions.
See Ross et al. (2005, p. 95).
414  457
Interprets meaning (by matching words and phrases, completing a sentence, or matching
adjacent words) in a short and simple text by reading on or reading back.
457  509
Reads on or reads back in order to link and interpret information located in various parts
of the text.
509  563
Reads on and reads back in order to combine and interpret information from various parts
of the text in association with external information (based on recalled factual knowledge)
that “completes” and contextualizes meaning.
563  618
Reads on and reads back through longer texts (narrative, document or expository) in order
to combine information from various parts of the text so as to infer the writer’s purpose.
618  703
Locates information in longer texts (narrative, document or expository) by reading on and
reading back in order to combine information from various parts of the text so as to infer
the writer’s personal beliefs (value systems, prejudices, and/or biases).
Level 8
Critical reading
703+
Locates information in a longer texts (narrative, document or expository) by reading on
and reading back in order to combine information from various parts of the text so as to
infer and evaluate what the writer has assumed about both the topic and the
characteristics of the reader – such as age, knowledge, and personal beliefs (value
systems, prejudices, and/or biases).
58
Literacy
See Ross et al. (2005) Kenya Report
59
Description of levels
Level 1
Pre-numeracy
Level 2
Emergent numeracy
Range on 500
point scale
< 364
Applies single step addition or subtraction operations. Recognizes simple shapes. Matches
numbers and pictures. Counts in whole numbers.
364  462
Applies a two-step addition or subtraction operation involving carrying, checking (through
very basic estimation), or conversion of pictures to numbers. Estimates the length of
familiar objects. Recognizes common two-dimensional shapes.
Level 3
Basic numeracy
462  532
Level 4
Beginning numeracy
532  587
Level 5
Competent numeracy
587  644
Level 6
Mathematically
skilled
Level 7
Concrete
solving
Level 8
Abstract
solving
Skills
644  720
problem
720  806
problem
> 806
Translates verbal information presented in a sentence, simple graph or table using one
arithmetic operation in several repeated steps. Translates graphical information into
fractions. Interprets place value of whole numbers up to thousands. Interprets simple
common everyday units of measurement.
Translates verbal or graphic information into simple arithmetic problems. Uses multiple
different arithmetic operations (in the correct order) on whole numbers, fractions, and/or
decimals.
Translates verbal, graphic, or tabular information into an arithmetic form in order to solve
a given problem. Solves multiple-operation problems (using the correct order of
arithmetic operations) involving everyday units of measurement and/or whole and mixed
numbers. Converts basic measurement units from one level of measurement to another
(for example, metres to centimetres).
Solves multiple-operation problems (using the correct order of arithmetic operations)
involving fractions, ratios, and decimals. Translates verbal and graphic representation
information into symbolic, algebraic, and equation form in order to solve a given
mathematical problem. Checks and estimates answers using external knowledge (not
provided within the problem).
Extracts and converts (for example, with respect to measurement units) information from
tables, charts, visual and symbolic presentations in order to identify, and then solves
multi-step problems.
Identifies the nature of an unstated mathematical problem embedded within verbal or
graphic information, and then translate this into symbolic, algebraic, or equation form in
60
order to solve the problem.
Source: (Hungi, et al., 2010)
Numeracy
See Ross et al. (2005) Kenya Report
61
Grade 6 Numeracy
(V-ANA 2011)
Correct answer (90 litres):
32% of Gr 6 students
Verification ANA 2011
Gr6 Numeracy
(Quest 25.1)
Quintile
1
2
3
4
5
Total
Wrong
74%
75%
70%
68%
50%
68%
Right
26%
25%
30%
32%
50%
Total
100%
100%
100%
100%
100%
32%
62
100%
60
40
0
• What proportion of students
are reaching higher order
competency levels?
20
Percent
80
100
Student competency levels
5
4
3
2
1
Quintiles of School Socio-economic Status
Reading Competency Levels
L1 - Pre Reading
L2 - Emergent Reading
L3 - Basic Reading
L4 - Reading for Meaning
L5 - Interpretive Reading
L6 - Inferential Reading
L7 - Analytical Reading
L8 - Critical Reading
63
Comparing SA & Botswana
Source: Spaull, 2011
Two school systems not one
Ex-department
• Grade 4 [2008]
• Data: NSES
•
(Taylor, 2011)
65
(Spaull & Taylor, 2012)
(Hanushek & Woessmann, 2008)
School system “success” - literacy
100%
6
2
9
13
90%
26
25
18
80%
70%
44
61
45
50
30
31
19
20%
18
27
10%
12
14
11
11
2
0%
Zambia
Malawi
54
62
40%
39
50
58
50%
30%
34
52
53
60%
37
Lesotho
Uganda
South Africa
17
13
8
7
Zimbabwe
Namibia
3
15
5
Tanzania
Enrolled and acquired higher order reading skills (Levels 6-8) by grade 6
Enrolled and acquired basic reading skills (Levels 3-5) by grade 6
Enrolled but functionally illiterate (Levels 1-2) by grade 6
Never enrolled or dropped out prior to Grade 6
8
Kenya
1
11
Swaziland
School system “success” - numeracy
100%
1
0
2
3
2
8
10
10
5
13
90%
29
80%
34
46
50
70%
53
50
58
60%
64
77
71
50%
40%
59
51
30%
37
44
34
39
20%
24
11
8
10%
12
14
0%
Zambia
Malawi
7
11
Namibia
Lesotho
11
2
Uganda
South Africa
8
Zimbabwe
15
Tanzania
Enrolled and acquired higher order numeracy skills (Levels 6-8) by grade 6
Enrolled and acquired basic numeracy skills (Levels 3-5) by grade 6
Enrolled but functionally innumerate (Levels 1-2) by grade 6
Never enrolled or dropped out prior to grade 6
11
11
Swaziland
5
Kenya
2 education systems
Dysfunctional Schools (75% of schools)
Functional Schools (25% of schools)
Weak accountability
Strong accountability
Incompetent school management
Good school management
Lack of culture of learning, discipline and order
Culture of learning, discipline and order
Inadequate LTSM
Adequate LTSM
Weak teacher content knowledge
Adequate teacher content knowledge
High teacher absenteeism (1 month/yr)
Low teacher absenteeism (2 week/yr)
Slow curriculum coverage, little homework or testing
Covers the curriculum, weekly homework, frequent
testing
High repetition & dropout (Gr10-12)
Low repetition & dropout (Gr10-12)
Extremely weak learning: most students fail
standardised tests
Adequate learner performance (primary and matric)
See Hoadley (2010), Fleisch (2008), Van der Berg et al (2011), Taylor (2012) as a few of many
69
Education
“Education is the great engine of
personal development. It is through
education that the daughter of a
peasant can become a doctor, that the
son of a mineworker can become the
head of the mine, that a child of farmworkers can become the president”
– Nelson Mandela
If we looked at 200 black Grade 1 children
12 years ago and then look at them again
in matric, according to the 2007 results,
only 1 out of the 200 were eligible for a
maths or science degree based on their
matric marks – the corresponding figure
for white children was 15 times higher.
70
Teacher absenteeism
SACMEQ III (2007)
SACMEQ III South Africa
Quintile 1
• What is the distribution
of teacher absenteeism
across school SES
quintiles?
Quintile 4
Quintile 3
Quintile 2
Quintile 5
0
10
20
30
Days absent per year
40
50
excludes outside values
71
PIRLS 2011 Performance compared to
Reference Countries
600
571 568 568 567 558
556 552
531 527
Only Eng/Afr Gr 5 home language
(wealthier) schools participated in PIRLS
2011. (Other countries tested Gr4)
500
450 448 439
430 428 425 421 419
400
391
310
300
200
100
0
72
What does a dualistic school
system look like in practice?
73
Teacher knowledge
SACMEQ III (2007)  401/498 Gr6 Mathematics teachers
7
Correct answer
(7km):
38% of Gr 6
Maths teachers
SACMEQ Maths
teacher test Q17
Correct
1
23%
2
22%
Quintile
3
38%
4
40%
5
74%
Avg
38%
2 education
systems
74
School practices
60
40
0
20
Percent
80
100
• How often do you send pupils home when the teacher is absent?
(SACMEQ 3)
EC
FS
KZN
NC
NW
MPU
NEVER
SOMETIMES
OFTEN
LIM
GAU
WC
(Taylor, 2011)
School practices
60
40
0
20
Percent
80
100
• How often do you leave pupils on their own when the teacher
is absent? (SACMEQ 3)
EC
FS
NW
KZN
NC
LIM
NEVER
SOMETIMES
OFTEN
GAU
MPU
WC
(Taylor, 2011)
School practices
60
40
0
20
Percent
80
100
• How often do you substitute an absent teacher with a qualified
teacher? (SACMEQ 3)
GAU
WC
KZN
FS
EC
NW
NEVER
SOMETIMES
OFTEN
NC
MPU
LIM
(Taylor, 2011)