Inequality of Opportunity in Educational Achievements

Transcription

1 Inequality of Opportunity in Educational Achievements Cross-Country and Intertemporal Comparisons P. Luongo University of Bari Inequality Measurement, trend, impacts and policies UNU-WIDER Conference September 2014 P. Luongo (University of Bari) Inequality of Educational Opportunity September / 28

2 Outline 1 Aim & Motivation 2 Model 3 Data 4 Results 5 Conclusions 6 Appendix P. Luongo (University of Bari) Inequality of Educational Opportunity September / 28

3 Research questions Aim & Motivation 1 Does the country ranking change when we switch the focus of the analysis from average test scores to fairness? 2 Is there any country that outperform in both the level and the degree of fairness? 3 There has been any change in the strength of the association between socio-economic characteristics and students performances? P. Luongo (University of Bari) Inequality of Educational Opportunity September / 28

4 Aim & Motivation Motivations Education influences labour market participation, civic engagement, health status, earnings, social mobility, etc. (Blau & Kahn, 2005; Hanushek & Woessmann, 2010; among others). Intergenerational persistence in educational achievements (Marks, 2005; Macdonald et al. 2010; Ermisch et al. 2012) Inequality in educational attainments (Thomas et al. 2001; Morrison & Murtin, 2007) Inequality in educational achievements (Brown et al., 2007; Micklewright et al. 2007) P. Luongo (University of Bari) Inequality of Educational Opportunity September / 28

5 Aim & Motivation IEOp Existing evidences de la Vega & Lekuona (2013): PISA 2009 Gamboa & Waltenberg (2011) PISA 2006 & 2009, LAC Ferreira & Gignoux (2011) PISA 2006 What s new? 1 PISA Changes over time (PISA 2003, 2006, 2009, 2012) 3 How do the less advantaged students perform? P. Luongo (University of Bari) Inequality of Educational Opportunity September / 28

6 Model Model Adapt to our framework the idea of measuring fairness through an ordered pair (Roemer, 2013): EduOpp = (W EEOp, IEOp) W EEOp : focuses on worst-off students IEOp: looks at the whole sample P. Luongo (University of Bari) Inequality of Educational Opportunity September / 28

7 Model Outcome function Test scores (s) depend only on circumstances (c) and effort (e) s i = f (c, e) c used to partition students into K (j = 1,..., K ) types e correspond to the rank π occupied by each student in its own type distribution of test scores v j (π): level of s for individuals in type j occupying the rank π P. Luongo (University of Bari) Inequality of Educational Opportunity September / 28

8 Model W EEOp W EEOp = 1 0 min j (π)dπ (1) Class-ranked situations: W EEOp corresponds to the average score of the worst-off students (Roemer, 2013) Not class-ranked situations: W EEOp corresponds to the left-hand envelope of the distribution of CDFs (Roemer, 2013) Empirically this involves the estimation, for each country, of each type-specific CDF and their envelopes P. Luongo (University of Bari) Inequality of Educational Opportunity September / 28

9 Model IEOp Ex-Ante Approach: IEOp measured as between type inequality in mean outcome Parametric procedure s i = βk i + ε i Index of Inequality IEOp = var(k, ˆβ) var(y) P. Luongo (University of Bari) Inequality of Educational Opportunity September / 28

10 Data Dataset OECD PISA 2003: 41 countries PISA 2006: 57 countries PISA 2009: 74 countries PISA 2012: 65 countries Domains Mathematics Science Reading P. Luongo (University of Bari) Inequality of Educational Opportunity September / 28

11 Data Sample 2 stages sampling procedure Students aged between 15 years and 3 months and 16 years and 3 months enrolled in grade 7 or higher Raw test scores (s) scaled by using IRT and then standardized s i = ˆµ + ˆσ σ (x i µ) where x i is the test score of student i, ˆµ = 500 and ˆσ = 100 are the arbitrary (final) grand mean and SD P. Luongo (University of Bari) Inequality of Educational Opportunity September / 28

12 Data Available Data PISA contains information on: Schools policies and practices Students background Students motivation Students learning style P. Luongo (University of Bari) Inequality of Educational Opportunity September / 28

13 Data Some empirical issues EduOpp = (W EEOp, IEOp) W EEOp focuses on the worst-off type: the omission of relevant circumstances determines a measure of social welfare UPWARD biased Intuition: when a new circumstance is added there is at least one additional type-distribution, conditional to a given value of the new circumstance, which is going to be at its left IEOp looks at the whole population: the omission of relevant circumstances determines a measure of inequality which is DOWNWARD biased; some variation is left unexplained and attributed to effort. As # of K W EEOp and IEOp P. Luongo (University of Bari) Inequality of Educational Opportunity September / 28

14 Data Variables Students circumstances 12types Gender Parental level of education 1 ISCED ISCED 4 3 ISCED 5 Parental job classification 1 White collar 2 Blue collar P. Luongo (University of Bari) Inequality of Educational Opportunity September / 28

15 W EEOp in Reading, 2012 Results SVK QRS BGR MNE KAZ QAT PER ARE JOR ROU CZE SRB ARG ISR MYS TUN SVN URY COL BRA LVA IDN CHL LTU RUS SWE HUN MEX CRI THA GRC NOR HRV POL QUB BEL LUX ISL FRA DNK QUA ITA TUR FIN LIE EST ESP JPN NZL CHE PRT KOR NLD QUC CAN GBR TAP USA DEU AUS IRL MAC VNM SGP HKG QCN WEOp Reading Author's elaborations on OECD PISA 2012 P. Luongo (University of Bari) Inequality of Educational Opportunity September / 28

16 IEOp in Reading, 2012 Results MAC HKG KOR JPN QUA IDN GBR NLD CAN NOR ISL MEX MYS ESP TUN IRL SRB USA QRS ITA KAZ RUS COL LIE AUS ARG EST VNM SWE SGP QUC NZL BRA QCN CZE TAP QUB CRI DNK CHE FIN HRV FRA LUX BEL QAT TUR GRC PRT POL URY ROU DEU ARE PER CHL HUN ISR LTU SVN LVA MNE JOR SVK THA BGR IEOp Reading Author's elaborations on OECD PISA 2012 P. Luongo (University of Bari) Inequality of Educational Opportunity September / 28

17 Results Average performance and IEOp in Reading, 2012 Average score in Reading BGR SVK THA JOR LVA ISR HUN SVN LTU CHL ARE ROU MNE QCN HKG SGP JPN KOR QUC FIN CAN QUB TAP IRL NZL DEU POL EST BEL AUS LIE NLD FRA CHE VNM NOR GBR MAC DNK USA LUX CZE QUA PRT SWE ITA ESP ISL HRV QRS GRC OECD TUR RUS SRB URY CRI MEX BRA TUN COL ARG MYS QAT IDN KAZ PER IEOp Reading Author's elaboration on OECD PISA P. Luongo (University of Bari) Inequality of Educational Opportunity September / 28

18 Results Is there any outperforming country? WEOp Reading a d BGR THA JOR SVK LVA MNE LTU ISR HUN CHL DEU ROU ARE PER GRC TUR FIN FRA BEL LUX IEOp Reading Author's elaboration on OECD PISA 2012 SVN PRT POL URY QAT HRV QCN OECD CRI BRA CZE COL ARG KAZ IRL AUS TAP QUC USA CHE NZL ESTLIE ITA DNK QUB SGP VNM SWE RUS ESP TUN MYS SRB QRS ISL NOR MEX.1 CAN NLD GBR IDN QUA KOR JPN HKG b MAC c.05 P. Luongo (University of Bari) Inequality of Educational Opportunity September / 28

19 Results Geographical Pattern IEOp Reading WEEOp Reading 2012 Western Europe Eastern Europe South America North Africa and Asia North America P. Luongo (University of Bari) Inequality of Educational Opportunity September / 28

20 Results Changes over time, WEOp Reading IEOp Reduction IDN DEU MEX GBR LIE ITACHE IRL POL BEL HKG NLD AUS NOR ISL HUN KOR LUX CZE TUN TUR JPN FRA OECD ESP DNK MAC SWE CAN FIN PRT USA RUS BRA THA NZL URY GRC IEOp Increase IEOp Reding Author's elaboration on OECD PISA LVA WEOp Increase WEOp Reduction P. Luongo (University of Bari) Inequality of Educational Opportunity September / 28

21 Limits of the analysis Conclusions Due to the omission of relevant circumstances the two components risk to be biased, so caution is necessary in interpreting the results. PISA involves only students who do not drop out and have not repeated too many grades. With these caveats in mind... P. Luongo (University of Bari) Inequality of Educational Opportunity September / 28

22 Conclusions Conclusion High heterogeneity across countries in terms of both levels and degree of fairness in education The strength of the association between parental background and students test scores tends to be higher in Reading than in Math and Science This association is, on average, lower in countries that perform better in average test scores There aren t countries that outperform in both dimensions of fairness W EEOp tends to be higher and IEOp lower in some Asiatic countries, in North America and in Western European countries where also variability is lower P. Luongo (University of Bari) Inequality of Educational Opportunity September / 28

23 Conclusions Conclusions (ctd.) Eastern European countries occupy an intermediate position in terms of IEOp Between 2003 and 2012 in W EEOp has been accompanied by in IEOp Few countries moved toward lower degree of IEOp all the while improving the performances of the less advantaged students. Most of them, with the exceptions of Indonesia and Mexico, are Western European. P. Luongo (University of Bari) Inequality of Educational Opportunity September / 28

24 Conclusions THANK YOU! P. Luongo (University of Bari) Inequality of Educational Opportunity September / 28

25 Appendix W EEOp in Mathematics, 2012 SVK MNE QRS PER QAT JOR ISR COL SRB BGR ARG CZE TUN BRA IDN KAZ ROU CHL ARE CRI URY ALB HUN LTU RUS MYS MEX GRC HRV NOR THA QUB SVN POL QUA FRA LVA TUR SWE QUC LUX DNK USA NZL GBR LIE IRL BEL ITA ESP AUS FIN ISL CAN CHE PRT JPN EST DEU NLD TAP VNM KOR MAC SGP HKG QCN WEOp Math Author's elaborations on OECD PISA 2012 P. Luongo (University of Bari) Inequality of Educational Opportunity September / 28

26 W EEOp in Science, 2012 Appendix QRS MNE KAZ QAT CZE PER BGR JOR SRB ARE ISR ROU IDN TUN ARG COL BRA ALB URY MYS SVN CRI MEX LTU CHL GRC NOR RUS HUN SWE LUX DNK QUB QUA FRA THA LVA HRV ISL LIE QUC POL CHE USA BEL TUR GBR NZL CAN ITA JPN NLD PRT IRL ESP AUS FIN TAP KOR DEU SGP MAC EST VNM HKG QCN WEOp Science Author's elaborations on OECD PISA 2012 P. Luongo (University of Bari) Inequality of Educational Opportunity September / 28

27 Appendix IEOp in Mathematics, 2012 IEOp Math ISL FIN QRS RUS NOR EST KOR SRB CAN HKG SWE GBR IDN NLD HRV MEX QUA MYS JPN ITA MNE QAT TUR CZE AUS LVA VNM IRL CHE JOR LTU SGP THA DNK QCN USA ARG GRC ESP SVN TUN QUC COL TAP ARE DEU ROU BEL BRA FRA NZL POL LIE CRI PRT QUB HUN URY SVK LUX PER BGR CHL ISR Author's elaborations on OECD PISA 2012 P. Luongo (University of Bari) Inequality of Educational Opportunity September / 28

28 IEOp in Science, 2012 Appendix KOR ISL HKG KAZ SRB EST NOR CAN FIN HRV QRS IDN VNM MYS JPN SWE TUR TUN ITA MEX RUS NLD QUA CZE GBR LVA THA IRL AUS ESP LTU COL USA MNE DNK SVN ARG QCN QAT CRI BRA CHE QUC POL GRC SGP TAP DEU NZL ROU PRT BEL FRA ARE JOR QUB LIE URY HUN CHL PER ISR LUX SVK BGR IEOp Science Author's elaborations on OECD PISA 2012 P. Luongo (University of Bari) Inequality of Educational Opportunity September / 28