Comparability of NSC mathematics scores and former SC HG

AMERU
African Micro Economic Research Unit
Comparability of NSC mathematics
scores and former SC HG
mathematics scores:
How consistent is the
signal across time?
AMERU
School leaving examination scores
signal the University as to applicants’:



academic potential.
skills set – particularly in language based
and quantitative areas of study.
preparedness for tertiary level education.
AMERU
The Conversion Key
Higher Grade SC to NSC
Percentage
80 –100
70 – 79
60 – 69
50 – 59
40 – 49
30 – 39
20 – 29
10 – 19
0-9
HG
A
B
C
D
E
F
G
H
I
NSC
7
6
5
4
3
2
1
1
1
Source: Wits Admission Office, 2008
AMERU
Observed Pass rates 2006 - 2010
2006
2007
2008
2009
2010
Enrolled
Students
Pass
rate
(%)
Enrolled
Students
Pass
rate
(%)
Enrolled
Students
Pass
rate
(%)
Enrolled
Students
Pass
rate
(%)
Enrolled
Students
Pass
rate
(%)
ECON
1000
1480
67
1353
46
1437
52
2100
36
1675
45
APPM
1004
959
66
901
63
892
64
1235
46
792
63
Weighted
Average
Seven
courses
63
57
55
42
48
AMERU
Research indicates that NSC mathematics
may be inflated by 20% in the lower ranges
Nel and Kirsten (2009) Stellenbosch
University access test used as a benchmark

Schoër et al (2010) Witwatersrand
Standardised pre-test and Economics I Test 1
used to convert NSC scores to HG SC scores

AMERU
SC HG 2006 vs NSC 2009 and 2010

Standard of NSC matric maths paper changed from
2009 to 2010 (Parliamentary Monitoring Group, 2010)

Comparing 2006 cohort to both NSC cohorts
controls for specific characteristics.

2006 wrote HG SC maths in 2005 - embedded in
previous curriculum
AMERU
The Measuring Instrument
Computational Maths I – Test 1





Computational maths I compulsory for all first year
commerce students.
Written three weeks into the semester.
New work based on school work.
Highly correlated with matric maths marks
Test 1 kept standard across the years.
–
–
–
numbers changed but questions the same
test kept confidential
questions re-ordered to promote accessibility
AMERU
Test 1 scores
APPM1004 Test1 comparison: score distribution (2006, 2009, 2010)
20%
% of Students
15%
2006
10%
2009
2010
5%
0%
<10
<20
<30
<40
<50
<60
Test1 score (%)
<70
<80
<90
<=100
AMERU
Delimiting the data – excluding:




students >21 years
repeating students
foreign students
students who did not write matric in South
Africa the year prior to entering University
AMERU
Comparing delimited cohorts
AMERU
Using propensity score matching to
create a more accurate conversion key.

Students matched from 2006 to 2009 & 2010
by:
 Test 1 mark
 Gender
 Race
AMERU
HG vs NSC
for Propensity Score Matches
NSC Mathematics %
Scores
HG vs NSC Mathematics Conversion
(Propensity Score Match)
90
80
2009 (NSC)
2010 (NSC)
70
60
50
40-49 50-59 60-69 70-79 80-89 90-100
HG Mathematics % Scores (2006)
AMERU
Corrected Conversion Key
Percentage
Category
80 – 100%
70 – 79%
60 – 69%
50 – 59%
40 – 49%
Higher Grade
A
B
C
D
E
NSC
(DoE key)
7
6
5
4
3
NSC
(Propensity
Score match)
7
7
6
6
5
AMERU
Comparison of deflated scores
AMERU
Test 1 score variance by maths score
category
2009
2006
2010
AMERU
Conclusion



NSC marks generally inflated by 15% - 20%.
20% in lower ranges and 15% in upper
ranges.
University courses that rely on a matric
maths mark as a signal may consider
increasing the required cut off mark.
NSC matric maths scores may be more
precise predictors of performance than HG
scores of 2006 cohort when deflated.