1
Public and private schools and PISA performance
One issue in Latin American education is the relative success of public to private schools. The new results from PISA provide further information. The term Latin America is, in this context the countries that submitted to PISA in 2006 and 2009 and consist of Argentina, Brazil, Chile, Colombia, Mexico, and Uruguay. The tables include similar data for the OECD average and Canada for comparisons.
The declining municipal/public sector in secondary education
There is a considerable variety of educational systems, measured by the proportion of students attending public or private schools (or at least participating in PISA); however there is only one tendency for all systems, the growing importance of the private sector. Table 1 rank orders the systems by the proportion of public school students in 2006 from Canada with 93 percent to Chile with 46.9 percent for the same year. In the three following years pupils attending public schools have diminished from a range of minus 7.5 percent (Brazil) to minus 0.5 percent (Canada). The average for all OECD countries has declined from 85.6 percent to 81.9 percent in that period. The rank order for the two years would be slightly altered by the fact that Brazil now has fewer pupils in public schools than Mexico. All the other rankings remain the same.
Table 1 Changes to public enrolment, 2006-2009
Percent public school students (by enrolment)2006 / 2009 / Change
Canada / 93.0 / 92.5 / -0.5
Brazil / 92.4 / 84.9 / -7.5
Mexico / 89.7 / 88.5 / -1.2
OECD average / 85.6 / 81.9 / -3.7
Uruguay / 84.9 / 82.1 / -2.8
Colombia / 82.7 / 80.0 / -2.7
Argentina / 67.5 / 63.9 / -3.6
Chile / 46.9 / 40.6 / -6.3
The dichotomy between public and private spheres is imposed by the data available for 2009. Whereas schools in 2006 were subject to a threefold classification – public, private and government assisted – the latter category does not appear to be available in the on-line database used here for the 2009 figures. This appears to be the product of collapsing the private and government assisted categories into a more general category – private. Hence to make the data comparable it was necessary to do the same with the 2006 survey scores by calculating the proportions of student in government assisted and private schools, and multiplying the reading, science and math scores by these proportions and then adding them to achieve a synthetic (in terms of 2006) new augmented ‘private’ category. These calculations are used in the following tables. Needless to say the 2009 figures, without this distinction, are not as rich, and it might be added that only the public category for this year remains unchanged and so fully deserving our trust.
Public schools and national performance
Overall the PISA results are very encouraging, comparing 2009 to 2006 not least because public schools, while not overtaking private schools, have shown considerable gains. The following table attempts to summarize the gains made by school systems and those of their public schools. The PISA test consists of three subjects and the horizontal shows in how many (out of three) there have been system gains when 2009 is compared to 2006. Thus Mexico has shown system wide gains in all three, whereas Canada has shown a gain in one (counting an equal score as a gain). The vertical dimension shows how many public schools showed a gain; following the above example, Mexico’s public schools showed gains in all three, whereas Canada showed no public school gains. Most important four countries, including Chile, showed gains in two subjects, but their public schools showed gains in all. The OECD average shows that all schools increased in two subjects but that public schools improved in one.
Table 2 Gains by system and public school.
Gains / All schoolsPublic schools / 3 / 2 / 1 / None
3 / Mexico / Brazil, Colombia,
Argentina, Chile
2
1 / OECD Av, Uruguay
None / Canada
How large were these gains and their alternative losses? If it is supposed that a large gain or loss is considered to be an improvement of 10 points or more, (an arbitrary number), and a large loss is a fall in 10 points or more, then the pattern is as shown in table 3.
Table 3 Large gains and losses by school type.
Large gains > 10 / Large losses < 10All / Public / Private / All / Public / Private
Canada / M
Brazil / R, M / S,R,M / R
Mexico / R, M / R, M / M
OECD Average
Uruguay / R / R / S
Colombia / R, M / R / S,R,M
Argentina / R, M / R
Chile / M / S,R,M / S
Thus Brazil showed large system gains in two areas, but public schools in three, a record only matched by Chilean public schools, an interesting dimension given that the system itself only made a large gain in mathematics. In general Latin American private schools showed fewer large gains than public schools and noticeably in the case of Uruguay and Brazil in different subjects than public schools. No system, apart from Chile in science, showed large losses. The limitations of this table are evident when considering Mexico, where the system showed large gains in reading and math, as did public schools, but private schools showed a large loss in mathematics. Hence it can inferred that it is the improvement in public schools that is driving the large system gains in that subject.
Do private schools continue to maintain their score advantage?
Are public schools likely to catch up tp private schools in terms of their average scores? The private school advantage is calculated for each year by subtracting the public from the private score for each subject area. The results are shown in Appendix table A.2 for all countries and subjects in 2006 and 2009. Thus the advantage of Canadian private over public schools in science was 46 points (2006) and 40 points in 2009, thus the private advantage registers a (relative) decline. Private schools,in all countries,scored higher than public schools in all three subject areas in both test years, ranging from an advantage of 109 points in math (Brazil, 2006) to 25 points in math (OECD average) in 2006; and from 118 in reading, (Brazil) to the OECD average in math (28) in 2009. In Latin America, compared to the OECD average, there are substantial differences between private and public school performance with all countries, with the exception of Chile, showing advantages of over 50 points in 2006. Chile is joined by Mexico underneath this cutoff point (less than 50 points advantage) in 2009.
Table 4 uses this data to examine if the advantage of private over public schools is increasing by comparing the static private advantages for 2006 and 2009 and calculating their difference, recalling that in all cases private schools have higher averages than public schools.
Table 4 Private to public advantage between 2006 & 2009
Private to Public Advantage / Increase / Large > 10 / Decrease / Large > 10Canada / 1 / 2 / M
Brazil / 3 / R / 0
Mexico / 0 / 3 / M
OECD Average / 3 / 0
Uruguay / 2 / S / 1
Colombia / 3 / S,R,M / 0
Argentina / 1 / 2
Chile / 2 / 1
Source: Appendix table 2
The table shows that private schools have improved their position on average in all OECD countries (the OECD average) although as the Appendix table shows these are less than two digits. Two countries (Brazil, Colombia) have seen increases in private advantage in the three subject areas, two (Uruguay and Chile) in two areas and Canada and Argentina in one. However it as important to know the strength of these gains and it is Colombia that shows there have been large gains in all subject areas and in Brazil and Uruguay in one. The absence of large gains in Canada, Argentina and Chile is a cause for celebration in that the distances between the types of school are not growing, even though overall averages are. The obverse is the number of decreases in distance between average marks and in particular the large declines in the teaching of mathematics in Canada and Mexico. Indeed this table calls attention to Mexico, in that the private to public advantage has declined in all subject areas. It should be recalled that Mexico is predominantly a public system (see table 1).
Explaining growth
However the assumption that changes in school types (see last column in table 1) might be closely associated with score changes is not born out. Simple regressions using school changes as the independent variable to explain total, public and private changes are very weak. The exercise was only carried out for reading and the scatter diagrams with regression calculations are reproduced as appendix 3, (Diagrams 1a to 2b). Greater strength might be shown if more examples were included.
Summary
Although private schools have grown in all countries, public schools have shown considerable gains in a number of countries. Equally important, a number of countries including Chile have not allowed the private to public advantage to increase, while their overall averages have increased.
Appendix Tables
A.1. Performance changes 2006, 2009
Change in Average score / Change in public score / Change in private scoreScience / Reading / Math / Science / Reading / Math / Science / Reading / Math
Canada / -5 / -3 / 0 / -6 / -3 / -2 / -6 / 8 / -14
Brazil / -4 / 19 / 35 / 18 / 20 / 20 / 5 / 14 / 3
Mexico / 9 / 15 / 10 / 9 / 18 / 16 / -8 / -5 / -13
OECD / 1 / 1 / -2 / 0 / 0 / -3 / 2 / 4 / 3
Uruguay / -1 / 13 / 0 / -5 / 12 / -3 / 10 / -2 / 8
Colombia / -7 / 28 / 32 / 10 / 22 / 8 / 33 / 41 / 25
Argentina / -3 / 24 / 20 / 8 / 25 / 9 / 1 / -3 / -7
Chile / -17 / 7 / 36 / 16 / 11 / 13 / -4 / 1 / 0
A.2. Public and private school relative advantages
Private to public difference2006 / 2009 / Change
Canada
Science / 46 / 40 / -6
Reading / 53 / 61 / 8
Math / 59 / 45 / -14
Brazil
Science / 107 / 112 / 5
Reading / 104 / 118 / 14
Math / 109 / 112 / 3
Mexico
Science / 53 / 45 / -8
Reading / 53 / 48 / -5
Math / 56 / 43 / -13
OECD Average
Science / 27 / 29 / 2
Reading / 26 / 30 / 4
Math / 25 / 28 / 3
Uruguay
Science / 80 / 90 / 10
Reading / 97 / 95 / -2
Math / 81 / 89 / 8
Colombia
Science / 78 / 79 / 1
Reading / 89 / 86 / -3
Math / 77 / 70 / -7
Argentina
Science / 78 / 79 / 1
Reading / 89 / 86 / -3
Math / 77 / 70 / -7
Chile
Science / 43 / 39 / -4
Reading / 45 / 46 / 1
Math / 40 / 40 / 0