1 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli Ranking and rating: Woodo or Science? Andrea Saltelli,

Slides:



Advertisements
Similar presentations
Synthetic Meta-Index of Sustainable Development: A DEA Approach Laurens Cherchye (Catholic University of Leuven, Belgium) Timo Kuosmanen (Wageningen University,
Advertisements

1 Alternative measures of well-being Joint work by ECO/ELSA/STD.
Employment quality in the OECD Better Life Initiative Anne Saint-Martin Meeting of the Group of Experts on Measuring Quality of Employment September.
IREG-4, Astana, 16 June Rickety Numbers Volatility of international rankings of higher education and implications for policy making ANDREA Saltelli.
See ( OECD-JRC handbook on CI The ‘pros’: Can summarise complex or multi-dimensional issues in view of supporting decision-makers.
1 Quality criteria for data aggregation used in academic rankings IREG FORUM on University rankings Methodologies under scrutiny May 2013, Warsaw,
The counterfactual logic for public policy evaluation Alberto Martini hard at first, natural later 1.
Testing the validity of indicators in the field of education The experience of CRELL Rome, October 3-5, 2012, Improving Education through Accountability.
Methodological and Analytical Issues Gaia Dallera 6 June,
Sensitivity Analysis for Observational Comparative Effectiveness Research Prepared for: Agency for Healthcare Research and Quality (AHRQ)
CONCEPTUAL ISSUES IN CONSTRUCTING COMPOSITE INDICES Nadia Farrugia Department of Economics, University of Malta Paper prepared for the INTERNATIONAL CONFERENCE.
CONFERENCE ON SMALL STATES AND RESILIENCE BUILDING Malta, APRIL 2007 " Weighting Procedures for Composite Indicators " Giuseppe Munda European Commission,
A discussion on benchmark assessments both nationally and internationally in order to maintain and improve standards “The difference between school and.
Stéfan Lollivier, Insee 27/06/2012 Improvements in the measurement of quality of life and well-being in France Measuring Well-Being and Fostering the Progress.
Confidence Intervals for Proportions
1-1 Copyright © 2015, 2010, 2007 Pearson Education, Inc. Chapter 18, Slide 1 Chapter 18 Confidence Intervals for Proportions.
Ranking - New Developments in Europe Gero Federkeil CHE – Centre for Higher Education Development The 3rd International Symposium on University Rankings.
Find 8 scholarly articles related to your dependent variable and target population. How does the literature theoretically define your dependent variable?
Ranking universities: The CHE Approach Gero Federkeil CHE – Centre for Higher Education Development International Colloquium “Ranking and Research Assessment.
1 BA 555 Practical Business Analysis Review of Statistics Confidence Interval Estimation Hypothesis Testing Linear Regression Analysis Introduction Case.
Michaela Saisana Second Conference on Measuring Human Progress New York, 4-5 March “Reflections on the Human Development Index” (paper by J. Foster)
Poverty measures: Properties and Robustness
September 26, 2012 DATA EVALUATION AND ANALYSIS IN SYSTEMATIC REVIEW.
The CHE ranking The multi-dimensional way of Ranking Isabel Roessler CHE – Centre for Higher Education Development International Conference “Academic Cooperation.
WORKING PARTY ON NATIONAL ACCOUNTS Paris, 3-5 October 2007 Revisions in Quarterly GDP of OECD Countries: An Update Document STD/CSTAT/WPNA(2007)15 Richard.
Review of Statistical Inference Prepared by Vera Tabakova, East Carolina University ECON 4550 Econometrics Memorial University of Newfoundland.
The Role of Citations in Warwick’s Strategy and Improving Them Nicola Owen (Academic Registrar) Professor Mark Smith (PVC Research: Science and Medicine)
Unido.org/statistics Composite measure of industrial performance for cross-country analysis Shyam Upadhyaya UNIDO The 59th World Statistics Congress Hong.
Animal Welfare EU Strategy Introduction Community Action Plan The Commission's commitment to EU citizens, stakeholders, the EP and.
Slide 1 Estimating Performance Below the National Level Applying Simulation Methods to TIMSS Fourth Annual IES Research Conference Dan Sherman, Ph.D. American.
Lesson 8: Effectiveness Macerata, 11 December Alessandro Valenza, Director, t33 srl.
Gero Federkeil Expert Seminar „Quality Assurance and Accreditation in Lifelong Learning“, Berlin, February 2011 Rankings and Quality Assurance.
TEN-T Experts Briefing, March Annual Call Award Criteria.
VI. Evaluate Model Fit Basic questions that modelers must address are: How well does the model fit the data? Do changes to a model, such as reparameterization,
1 Historical Perspective... Historical Perspective... Science Education Reform Efforts Leading to Standards-based Science Education.
A Bibliometric Comparison of the Research of Three UK Business Schools John Mingers, Kent Business School March 2014.
Infrastructures and ICT. Measurement Issues and Impact on Economic Growth Matilde Mas Universitat de València & Ivie OECD Workshop on Productivity Analysis.
Regular process for global reporting and assessment of the state of the marine environment, including socio-economic aspects Guidance for Authors.
League tables as policy instruments: the political economy of accountability in tertiary education Jamil Salmi and Alenoush Saroyan CIEP, June 2006.
Access to Medicine Index Problem Statement Long-standing debate about: What is the role of the pharmaceutical industry in access to medicines? Where are.
1 Chapter 10: Introduction to Inference. 2 Inference Inference is the statistical process by which we use information collected from a sample to infer.
The European agenda on improving the efficiency of employment and social policies: Bratislava, December 2011 The example of social experimentation.
European Commission DG Joint Research Centre Formal and informal approaches to the quality of information in integrated.
1 The Impact of Low Income Home Owners on the Volatility of Housing Markets Peter Westerheide ZEW European Real Estate Society Conference 2009 Stockholm.
Issues concerning the interpretation of statistical significance tests.
META-ANALYSIS, RESEARCH SYNTHESES AND SYSTEMATIC REVIEWS © LOUIS COHEN, LAWRENCE MANION & KEITH MORRISON.
From description to analysis
Changes in the context of evaluation and assessment: the impact of the European Lifelong Learning strategy Romuald Normand, Institute of Education Lyon,
Measuring Sustainable development: Achievements and Challenges Enrico Giovannini OECD Chief Statistician June 2005.
Methodology for deriving the STRI Hildegunn Kyvik Nordås Alexandros Ragoussis OECD Trade and Agriculture OEC D/TAD Services expert meeting 2 July 2009.
METHODS OF SPATIAL ECONOMIC ANALYSIS LECTURE 06 Δρ. Μαρί-Νοέλ Ντυκέν, Αναπληρώτρια Καθηγήτρια, Τηλ Γραφείο Γ.6 UNIVERSITY.
League tables as policy instruments: the political economy of accountability in tertiary education Jamil Salmi and Alenoush Saroyan 2 nd IREG Meeting Berlin,
Chapter 9: Introduction to the t statistic. The t Statistic The t statistic allows researchers to use sample data to test hypotheses about an unknown.
Measurement Systems for Sustainability Arrow’10 Inclusive wealth – one particular metric Parris & Kates Review 12 indicator initiatives  How do we choose.
Table 3. Merits and Demerits of Selected Water Quality Indices Shweta Tyagi et al. Water Quality Assessment in Terms of Water Quality Index. American Journal.
CORRUPTION PERCEPTIONS INDEX 2012 Published 5 th December 2012 Scores and ranks 176 countries and territories from around the world on the perceived level.
Performance indicators: good, bad, and ugly The report of the Royal Statistical Society working party on Performance Monitoring in the Public Services.
Literature Review: Conception to Completion
Panagiota DIGKOGLOU Jason PAPATHANASIOU
Presented by: Fahrudin Memić Sarajevo, November 2017
Cohesion Policy and Cities
Professor Diana Kornbrot University of Hertfordshire
Regional Policy developments
IMPROVING THE REGIONAL DIMENSION OF EU-SILC
Chapter 7: The Normality Assumption and Inference with OLS
Expert Group on Quality of Life Indicators
DRAFT COMMISSION REGULATIONS IMPLEMENTING EP AND COUNCIL REGULATION No 1338/2008 ON: - CAUSES OF DEATH STATISTICS - ACCIDENTS AT WORK (ESAW) AGENDA.
Prodcom Working Group Item Quality reporting and indicators
LAUNCHING THE 2019 REGIONAL COMPETITIVENESS INDEX RCI 2019
Presentation transcript:

1 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli Ranking and rating: Woodo or Science? Andrea Saltelli, European Commission, Joint Research Centre, Unit of Applied Statistics and Econometrics X Conferenza Nazionale di Statistica Roma 15-16/12/2010

2 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli October June 20061,440 May 20071,900 October 20083,030 September 20094,420 August 20105,240 (5,280 today) Searching “composite indicators” on Scholar Google:

3 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli David Hand, president of the UK Royal statistical Society “League tables […] are an easy target for criticism. […] surgeon can refuse to operate on the difficult cases, schools can refuse to enter those pupils likely to do poor in examinations, health authorities can defer making appointments for some patients, so that the waiting lists look smaller, and so on.”

4 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli

5 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli The Stiglitz report, on page 65, mentions: […] a general criticism that is frequently addressed at composite indicators, i.e. the arbitrary character of the procedures used to weight their various components. Adding: […] The problem is not that these weighting procedures are hidden, non-transparent or non- replicable – they are often very explicitly presented by the authors of the indices, and this is one of the strengths of this literature. The problem is rather that their normative implications are seldom made explicit or justified.

6 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli […] The problem is not that these weighting procedures are hidden, non-transparent or non- replicable – they are often very explicitly presented by the authors of the indices, Disagree: Weighting problems are often not so evident. E.g. Most composite indicators are built by linear aggregation which are almost by definition wrong. Can something be done to alleviate the problem?

7 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli Are ubiquitous Can deceive Can inform How can one tell the good from the bad? A first case study: University ranking

8 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli Rickety Numbers: Volatility of university rankings and policy implications, Michaela Saisana , Béatrice d'Hombres, Andrea Saltelli, To appear on Research Policy ~2010/2011 Sources (I):

9 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli Sources (II):

10 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli Joint OECD-JRC handbook. 5 years of preparation, 2 rounds of consultation with OECD high level statistical committee, finally endorsed March 2008 with one abstention Sources III

11 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli 2010 Rule of Law Index (World Justice Project) 2010 Global Competitiveness Index (WEF) 2010 Regional Competitiveness Index (DG REGIO, JRC) 2010 Multidimensional Poverty Assessment Tool (UN IFAD) 2010/2008/2006 Environmental Performance Index (Yale & Columbia Uni) 2009 Index of African Governance (Harvard Kennedy School) 2008 Product Market Regulation Index (OECD) 2008 European Lifelong Learning Index (Bertelsmann Foundation, CCL) 2007 Alcohol Policy Index (New York Medical College) 2007 Composite Learning Index (Canadian Council on Learning) 2002/2005 Environmental Sustainability Index (Yale & Columbia University) Methodology applied to:

12 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli Sensitivity Analysis

13 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli <<I have proposed a form of organised sensitivity analysis that I call “global sensitivity analysis” in which a neighborhood of alternative assumptions is selected and the corresponding interval of inferences is identified. Conclusions are judged to be sturdy only if the neighborhood of assumptions is wide enough to be credible and the corresponding interval of inferences is narrow enough to be useful.>> Edward E. Leamer, 1990, Let's Take the Con Out of Econometrics, American Economics Review, 73 (March 1983),

14 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli “ One reason these methods are rarely used is their honesty seems destructive; ” Ibidem Tantalus on the Road to Asymptopia Edward E. Leamer, 2010 Journal of Economic Perspectives, 24, (2), 31–46.

15 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli Peter Kennedy, A Guide to Econometrics. Anticipating criticism by applying sensitivity analysis. This is one of the ten commandments of applied econometrics according to Peter Kennedy: <<Thou shall confess in the presence of sensitivity. Corollary: Thou shall anticipate criticism >> The critique of models Uncertainty

16 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli

17 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli Two international university rankings  SJTU ranking  THES ranking  Robustness (uncertainty & sensitivity analysis)

18 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli University rankings are used to judge about the performance of university systems

19 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli Two international university rankings yearly published + Very appealing for capturing a university’s multiple missions in a single number + Allow one to situate a given university in the worldwide context - Can lead to misleading and/or simplistic policy conclusions

20 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli Questions: can we have confidence in university rankings? How much do the university ranks depend on the methodology (weighting scheme, aggregation, indicators)?

21 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli SJTU ranking

22 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli METHODOLOGY 6 indicators Best performing institution =100; score of other institutions calculated as a percentage Weighting scheme chosen by rankers Linear aggregation of the 6 indicators SJTU ranking

23 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli PROS and CONS 6 « objective » indicators Focus on research performance, overlooks other U. missions. Biased towards hard sciences intensive institutions Favours large institutions SJTU ranking

24 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli THES ranking

25 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli METHODOLOGY 6 indicators z-score calculated for each indicator; best performing institution =100; other institutions are calculated as a percentage Weighting scheme: chosen by rankers Linear aggregation of the 6 indicators THES ranking

26 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli PROS and CONS Attempt to take into account teaching quality Two expert-based indicators: 50% of total Subjective indicators, lack of transparency Substantial yearly changes in methodology Measures research quantity THES ranking

27 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli RED=UK (all under the SJTU=THES line…)

28 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli Robustness analysis of SJTU and THES

29 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli  Harvard, Stanford, Berkley, Cambridge, MIT: top 5 in more than 75% of our simulations.  Univ California SF: original rank 18 th but could be ranked anywhere between the 6 th and 100 th position  Impact of assumptions: much stronger for the middle ranked universities SJTU ranking

30 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli  Impact of uncertainties on the university ranks is even more apparent.  M.I.T.: ranked 9th, but confirmed only in 13% of simulations (plausible range [4, 35])  Very high volatility also for universities ranked 10 th -20th position, e.g., Duke Univ, John Hopkins Univ, Cornell Univ. THES ranking

31 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli Thus far: Apart from the top 10 universities, neither the SJTU nor the THES should be used to compare the performance of individual universities. According to SJTU Universities in the US outperform those in Europe – less so for THES but there is a bias toward UK universities

32 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli Can we say more about the relative quality of THES and SJTU? Are these indices coherent? Do the weights given by developers reflect the importance of the variables? Let us try a global sensitivity measure

33 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli Nobel Alumni SJTU

34 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli Using these points we can compute a statistics that tells us: How much (on average) would the variance of SJTU score be reduced if I could fix the variable ‘Alumni with Nobel ’?

35 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli This measure S i shall be our ruler for ‘importance’; example: S i =0.79  I could reduce the variation of the SJTU score by 79% by fixing ‘Nobel Alumni’.

36 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli At this point I can compare the importance of a sub-indicator as given by the nominal weight (assigned by developers) with the importance as measured by S i to test the index for coherence.

37 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli SJTU ranking weightSi

38 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli THES ranking weightSi

39 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli How about pillars In developing composite indicators pillars often represents normative dimensions which are given by design equal weights.

40 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli How about pillars We can now test when this is the case on the real data.

41 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli Index of African governance weight Si

42 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli Sustainable society index (NL) (1/7)*100 (2/7)*100

43 Constructing Composite Indicators: From Theory to Practice ECFIN, November 11-12, 2010 Andrea Saltelli Conclusions -Two methods to test the quality of composite indicators -Uncertainty modeling and propagations (invasive) - Check of coherence (non invasive)