DOES EDUCATIONAL TRACKING AFFECT PERFORMANCE AND INEQUALITY? DIFFERENCIES- IN-DIFFERENCIES EVIDENCE ACROSS COUNTRIES Eric A. Hanushek and Ludger Wößmann.

DOES EDUCATIONAL TRACKING AFFECT PERFORMANCE AND INEQUALITY? DIFFERENCIES-IN-DIFFERENCIES EVIDENCE ACROSS COUNTRIES

Eric A. Hanushek and Ludger Wößmann

Martino AndreaniMtr:751500

APP9/3/2010

STRUCTURE OF THE PRESENTATION

Summary of the results

Strategy of prooving the claim

Merit of the paper

Weackness of the paper

Policy implication

SUMMARY OF THE RESULT

RESEARCH QUESTIONS:

• Does educational tracking affect performance and inequality?

• What is the trade off between these two elements?

WHAT IS TRACKINGIs the manner in wich students are separated by academic ability into

different kinds of schools.

The opposite one is the comprehensive system.

ADVANTAGES:•Homogeneous classrooms permit a focused curriculum and so a maximum learning by all student.•Theachers does not have to worry about different learning ability of the students.•This would lead to an increase in performance for high ability student, and a more adeguate curriculm for low-ability students.

CRITICS:•Lower groups of student will be systematically disadvantaged by slower learning ability.•Presence of other factors that have influence on the learning ability, such as family background and their living context.•Proponents of ungrouped class suggest that heterogeneity might give rise to efficiency gains by a nonlinear relation: high level stundents lose nothing while lower ability students gain through interaction.

SUMMARY OF THE RESULT/3

• Difficulties to analyze such macro issue, because the variation in structure are related to characteristics of the families and of the school choosing.

• Major elements of the institutional structure of schools are choices whose impact is difficult to separate from other influence on achievement.

To deal with this analitical complexity they provide evidence from international experience across country, using the

differences-in-differences method of analysis.

Tracking that operates at the level of nations eliminates all within state variations.

SUMMARY OF THE RESULTS/4

• Results consistently indicate that early tracking increases inequality in achievement.

• The evidence on the level of performance are less certain, there are very little evidences that there are efficiency gains associated with this increased inequality.

Where:=individual achievement of student (i) in grade (g) and country (c)α=country specific intercept У=Impact of tracking T=existence of tracking X=Attributes of family and school

Two problems:β: we do not have sufficient knowledge or data to be confident of any

estimation.T: if every students in the country is subject to tracking, T will be a

constant so we cannot estimate its influence on A.

CROSS-COUNTRY IDENTIFICATION

Finance and operations of school systems, social structure and family backgrounds and other, often unobserved factors, affected the observed outcomes in education.

For a semplification:

STRATEGY FOR PROOVING THE CLAIM

If and also

STRATEGY FOR PROOVING THE CLAIM/2

To detect change that occur between primary school (g: grade before tracking) and later school (g*) we do a simple estimate:

This equation is estimated in a regression framework where mean performance in g* is regressed on mean performance in g along with an indicator for the existance of TrackingIn the differences-in-differences method each country’s own primary school outcome is used as a control for its secondary-school outcome

STRATEGY FOR PROOVING THE CLAIM/4

IMPACT OF EARLY TRACKING

The figure plots the relative standard deviation of scores for country with early tracking (solid line) and countries without early tracking (dashed lines).

•Relative inequality rises in every country with early tracking except Slovak Republic.

•In non tracking country, relative inequality decreases except for Sweden and Latvia.

•The top four countries with higher increases inequality between primary and secondary education, have tracking in their educational systems.

Table below shows the regression analysis of different measures of inequality:

Columns (1) (3) (5): standard deviation of test scores within each country at different percentiles

Columns (2) (4) (6): diff-in-diffs estimates of the effect of early tracking on the three inequality measures

None of the three inequality measures is statistically significantly related to tracking in a simple bivariate analysis. This may be caused by heterogeneity in inequality of the partecipating countries

Diff-in-diffs estimates of the effect of early tracking on the inequality measures which condition on the extent of educational inequality already present in late primary school, before tracking in any country

Point estimates of roughly 0.6 indicate that schools every where tend to reduce the inequality wich was present in primary grades.

Countries that tracked their student before age of 15 shows a statistically significantly larger inequality on the PISA 2003, difference existed in primary school accounted for.Specifically they show a national SD of test scores in secondary school that is one quarter of a cross country SD larger than non tracked.

Differences in differences result unsing the standard deviation as inequality measures.

Inequality in secondary schools is not even statistically significance related to inequality in primary school; this is caused first by the interferences based on the number of country, but the consistency of the repeated test across different years say that the differences between tracked and untracked countries are systematic and sustantial

Shows results of pooling the eight test pairs, which yelds a positive estimate of tracking on inequality with strong statistical power

Quite all estimates of the coefficient on early tracking are positive and four are statistically significance (PISA 2000/02-PIRLS is negative)

TRACKING AN MEAN PERFORMANCE: Does performance converge to a lower or higher level?

Statistically significant lower achievement associated with early tracking

Mathematics results are always lower with early tracking (statistically significance in only one of the three comparisons)

Two of the trhee estimates indicate positive achievement effects from early tracking

In the pooled specification the negative estimate dominates, although with relatively low statistical power

WHO GAINS WHO LOSES?

The most striking finding is that in no case do some student gain at the expenses of others; both high and low achievers lose, or gains, from tracking.

This figure shows lower performers suffer more from early tracking than higher ones.

The net ipact comes from the differential impacts on different parts of the distributions.

FUTURE PRESPECTIVE FOR THE RESEARCH

•Some of the literature has suggested that one channel for increasing inequality is reinforcing the effects of family background.

•If much of the early inequality in achievement is associated with differences in family background, many of the track placements will associated with differences in family background.

•So family background is a divring force in setting track placements even beyond its impact in early achievement levels.

•If we could see the micro data it would be possible to consider more fully the underlyng structural model of achievement that would generate these patterns of aggregate outcomes.

•Extending the dichotomous analysis between tracked and non-tracked systems, there may be heterogeneity in the rigidity of tracked systems.

•Future research may explore the extend to wich allowing mobility across tracks might reduce the negative effects on tracking

MERIT OF THE PAPER• Cross country comparison with the differences in differences

method, reduces the impact of other variables implied in the analysis of the education systems.

• In this way, we can observe the trade off between performances and inequality, which would have been impossible to observein another way.

• Authors have added other variables to the model, such as GDP per capita and total educational expenditure per student by age 15 and they saw that the impact of tracking did not change.

• They try to enter tracking as a linear variable depicting the age at wich a country first tracked its student, and the results show a broadly consistance with results using the simple existance of tracking.

• Authors use some figures and tables that help in the comprehension of the objects studied.

WEACKNESS• The object of the paper is very difficult to study: there are a lot

of factors that influence performance and inequality analizing school systems, like residential place, family background, ability and so on.

• Results on the level of performance, in some cases, are not statistically significance.

• There is not a qualitative analysis. Authors could have interviewed teachers from two types of systems, asking them their opinion about it.

• Standard deviation is an average measure. In the paper there is no indication on the rate of schooling abandon. This rate could be used as a measure of inequality if we consider the drop out as a social problem.

• I tried to observe the proportion of students graduating from upper secondary programmes across analyzed country.

The majority of tracking country alalyzed are above the OECD average 5 non tracking country on 8

analyzed are below the OECD average

POLICY IMPLICATION• It seems incumbent on those advocating early tracking in schools to

identify the potential gains from this. These preliminary result suggest that country lose in term of distribution and also possibly in levels of outcomes.

• A country that already has a tracked system, would consider cost and benefit from changing its educational system. One solution would be to search to reduce negative effects of tracking by rising the age in which tracking starts.

• Policy makers must pay attention to the inequality in the access of this type of school. This is necessary to reduce the impact of family background and residential living

• It is necessary to pay attention on the drop out phenomenon. If comprehensive systems are better in terms of equality and performance they are not so good in term of rate of school abandon.

• A solution would be to create a vocational system for those who are not able to attend comprehensive school. Such school would be locally decentralized to guarantee the domand side of local production.