Nigeria Impact Evaluation Community of Practice Abuja, Nigeria, April 2, 2014 Measuring Program Impacts Through Randomization David Evans (World Bank)

Slides:



Advertisements
Similar presentations
AFRICA IMPACT EVALUATION INITIATIVE, AFTRL Africa Program for Education Impact Evaluation Muna Meky Impact Evaluation Cluster, AFTRL Slides by Paul J.
Advertisements

The World Bank Human Development Network Spanish Impact Evaluation Fund.
The World Bank Human Development Network Spanish Impact Evaluation Fund.
The World Bank Human Development Network Spanish Impact Evaluation Fund.
Knowing if the RBF mechanism is working Incorporating Rigorous Impact Evaluation into your HRBF program Sebastian Martinez World Bank.
Advantages and limitations of non- and quasi-experimental methods Module 2.2.
#ieGovern Impact Evaluation Workshop Istanbul, Turkey January 27-30, 2015 Measuring Impact 1 Non-experimental methods 2 Experiments Vincenzo Di Maro Development.
The World Bank Human Development Network Spanish Impact Evaluation Fund.
Impact Evaluation Click to edit Master title style Click to edit Master subtitle style Impact Evaluation World Bank InstituteHuman Development Network.
Chapter 11 What Works and What Doesn’t. Are Hospitals Good for You? From Angrist and Pischke, Mostly Harmless Econometrics.
The counterfactual logic for public policy evaluation Alberto Martini hard at first, natural later 1.
Assessing Program Impact Chapter 8. Impact assessments answer… Does a program really work? Does a program produce desired effects over and above what.
Impact Evaluation Methods. Randomized Trials Regression Discontinuity Matching Difference in Differences.
Who are the participants? Creating a Quality Sample 47:269: Research Methods I Dr. Leonard March 22, 2010.
Impact Evaluation: The case of Bogotá’s concession schools Felipe Barrera-Osorio World Bank 1 October 2010.
Clinical Trials Hanyan Yang
SAMPLING AND STATISTICAL POWER Erich Battistin Kinnon Scott Erich Battistin Kinnon Scott University of Padua DECRG, World Bank University of Padua DECRG,
1 Randomization in Practice. Unit of randomization Randomizing at the individual level Randomizing at the group level –School –Community / village –Health.
Matching Methods. Matching: Overview  The ideal comparison group is selected such that matches the treatment group using either a comprehensive baseline.
AADAPT Workshop Latin America Brasilia, November 16-20, 2009 Non-Experimental Methods Florence Kondylis.
Measuring Impact: Experiments
Global Workshop on Development Impact Evaluation in Finance and Private Sector Rio de Janeiro, June 6-10, 2011 Mattea Stein Quasi Experimental Methods.
Quasi Experimental Methods I Nethra Palaniswamy Development Strategy and Governance International Food Policy Research Institute.
Assessing the Distributional Impact of Social Programs The World Bank Public Expenditure Analysis and Manage Core Course Presented by: Dominique van de.
CAUSAL INFERENCE Shwetlena Sabarwal Africa Program for Education Impact Evaluation Accra, Ghana, May 2010.
The days ahead Monday-Wednesday –Training workshop on how to measure the actual reduction in HIV incidence that is caused by implementation of MC programs.
The World Bank Human Development Network Spanish Impact Evaluation Fund.
LAURA RALSTON, ECONOMIST, CCSD FINDING TRUE PROGRAM IMPACTS THROUGH RANDOMIZATION.
Africa Impact Evaluation Program on AIDS (AIM-AIDS) Cape Town, South Africa March 8 – 13, Causal Inference Nandini Krishnan Africa Impact Evaluation.
Impact Evaluation Designs for Male Circumcision Sandi McCoy University of California, Berkeley Male Circumcision Evaluation Workshop and Operations Meeting.
The World Bank Human Development Network Spanish Impact Evaluation Fund.
AFRICA IMPACT EVALUATION INITIATIVE, AFTRL Africa Program for Education Impact Evaluation David Evans Impact Evaluation Cluster, AFTRL Slides by Paul J.
Applying impact evaluation tools A hypothetical fertilizer project.
Non-experimental methods Markus Goldstein The World Bank DECRG & AFTPM.
What is randomization and how does it solve the causality problem? 2.3.
Measuring Impact 1 Non-experimental methods 2 Experiments
Africa Impact Evaluation Program on AIDS (AIM-AIDS) Cape Town, South Africa March 8 – 13, Steps in Implementing an Impact Evaluation Nandini Krishnan.
Africa Program for Education Impact Evaluation Dakar, Senegal December 15-19, 2008 Experimental Methods Muna Meky Economist Africa Impact Evaluation Initiative.
Using Propensity Score Matching in Observational Services Research Neal Wallace, Ph.D. Portland State University February
Cross-Country Workshop for Impact Evaluations in Agriculture and Community Driven Development Addis Ababa, April 13-16, 2009 Steps in Implementing an Impact.
Implementing an impact evaluation under constraints Emanuela Galasso (DECRG) Prem Learning Week May 2 nd, 2006.
Randomized Assignment Difference-in-Differences
David Evans World Bank Joint work with Brian Holtemeyer and Katrina Kosec (IFPRI) July 9, 2015.
Bilal Siddiqi Istanbul, May 12, 2015 Measuring Impact: Non-Experimental Methods.
Social Experimentation & Randomized Evaluations Hélène Giacobino Director J-PAL Europe DG EMPLOI, Brussells,Nov 2011 World Bank Bratislawa December 2011.
What is Impact Evaluation … and How Do We Use It? Deon Filmer Development Research Group, The World Bank Evidence-Based Decision-Making in Education Workshop.
Africa Impact Evaluation Program on AIDS (AIM-AIDS) Cape Town, South Africa March 8 – 13, Randomization.
Impact Evaluation for Evidence-Based Policy Making Arianna Legovini Lead Specialist Africa Impact Evaluation Initiative.
Measuring causal impact 2.1. What is impact? The impact of a program is the difference in outcomes caused by the program It is the difference between.
Impact Evaluation Methods Randomization and Causal Inference Slides by Paul J. Gertler & Sebastian Martinez.
Development Impact Evaluation Initiative Innovations in investment climate reforms Paris, Nov 13, 2012 In collaboration with the Investment Climate Global.
Cross-Country Workshop for Impact Evaluations in Agriculture and Community Driven Development Addis Ababa, April 13-16, Causal Inference Nandini.
Kenya Evidence Forum - June 14, 2016 Using Evidence to Improve Policy and Program Designs How do we interpret “evidence”? Aidan Coville, Economist, World.
Measuring Results and Impact Evaluation: From Promises into Evidence
Quasi Experimental Methods I
Quasi Experimental Methods I
Explanation of slide: Logos, to show while the audience arrive.
Quasi-Experimental Methods
Development Impact Evaluation in Finance and Private Sector
1 Causal Inference Counterfactuals False Counterfactuals
Implementation Challenges
Randomization This presentation draws on previous presentations by Muna Meky, Arianna Legovini, Jed Friedman, David Evans and Sebastian Martinez.
Impact Evaluation Methods: Difference in difference & Matching
Evaluating Impacts: An Overview of Quantitative Methods
Randomization This presentation draws on previous presentations by Muna Meky, Arianna Legovini, Jed Friedman, David Evans and Sebastian Martinez.
Impact Evaluation Designs for Male Circumcision
Sampling for Impact Evaluation -theory and application-
Applying Impact Evaluation Tools: Hypothetical Fertilizer Project
Steps in Implementing an Impact Evaluation
Presentation transcript:

Nigeria Impact Evaluation Community of Practice Abuja, Nigeria, April 2, 2014 Measuring Program Impacts Through Randomization David Evans (World Bank)

Objective 2  Evaluate the causal impact of a program or an intervention on some outcome  Examples  How much did free distribution of bednets decrease malaria incidence?  Which of two supply chain models was most effective at eliminating drug shortages?

Counterfactual Criteria 3 Treated & comparison groups…  Have identical average characteristics (observed & unobserved)  The only difference is the treatment  Therefore the only reason for any difference in outcomes is the treatment  Key question: What would participant look like if she hadn’t received the program?

Perfect Experiment 4 1. Identify target beneficiaries 2. Clone them! Identical on the outside (observable) Identical on the inside (unobservable) Chief Ahun We’re both middle- aged chiefs We both love to take up new health interventions! Chief Batun

Perfect Experiment 5 Give the intervention to one set of clones Ahun Batun

Perfect Experiment 6 Observe some time later Because the groups are identical (inside & out), the difference is due to the bednets! Ahun Batun

Back to Reality 7 What would Batun look like if he didn’t receive the bednet? Room For Improvement Control Groups  Before – After  Participants – Non Participants ???

RFI: Before-After BEFORE BEDNETS 6 malaria episodes in 6 months AFTER BEDNETS 2 malaria episodes in 6 months What else might be going on besides the bednets? Seasonal differences Rising incomes: Households invest in other measures Too many other factors! Impact of bednets = ???

RFI: Before-After  Important to monitor before-after  Insufficient to show impact of program  Too many factors changing over time  Example of cash transfers in Nicaragua!  Counterfactual: What would have happened in the absence of the project, with everything else the same

RFI: Participants vs Non-Participants 10  Compare recipients of a program to  People who were not eligible for the program  People who chose not to enroll in the program Impact of clinic births? What else might explain the difference?

RFI: Participants vs Non-Participants 11 Observable differences  Income  Education Unobservable differences  Heard rumor about hospitals  Neighbor available to care for other children Mercy Patience

RFI: Participants vs Non-Participants 12 How much of difference is because of clinic? Impact of clinic births = ??? Impact of clinic births Other factors!

Selection bias 13  People who choose to join the program are different!  If we cannot account completely for those differences in our data…  We never can  How do you capture attitudes toward health systems? Initiative?  …then our comparison will not show the true impact of the program

What should we do? Gold standard: Randomized experimental design

Randomized Experimental Design 15  Randomly assign potential beneficiaries to be in the treatment or comparison group  Treatment and comparison have the same characteristics (observed and unobserved), on average, so…  Any difference in outcomes is due to treatment

Why Randomization Works 16  Randomization with two doesn’t work!  But differences average out in a big sample  On average, same number of Ahuns and Patiences  Observable AND unobservable  Result: Measure true impact of program Comparison Treatment Comparison Treatment

Random Sample or Assignment? RANDOM SAMPLE  Select randomly who to gather data on  Gives unbiased average of the group NOT of impact  If take random sample of group: Half women, half men – Sample should be about ½ women, ½ men RANDOM ASSIGNMENT  Randomly assign who gets the program  Gives unbiased estimate of program impact  Why?  Treatment & comparison are IDENTICAL (on average) 17 T T C C

LET’S RANDOMIZE! Identify the eligible participants What is the impact of receiving a new car on body-mass index?

LET’S RANDOMIZE! Generate a random number for each one

LET’S RANDOMIZE! Generate a random number for each one

LET’S RANDOMIZE! Re-order based on the random number

LET’S RANDOMIZE! Assign the first ten to receive cars I really wanted a car!

LET’S RANDOMIZE! Check for balance across treatment & control Treatment # Drs: 4 Control # Drs: 6 Total observations 21 Not a very big sample! How close? 2/3

LET’S RANDOMIZE! 24 What if we had 500 observations? Treatment # Drs: 114 Control # Drs: 116 How close? 98/100

Is there more? 25  That’s a simple way to randomize  Works with BIG samples  You can help randomization by stratifying  Randomize within each sub-group  Ensure that each group is equally represented

Stratifying Identify the characteristic(s) to stratify on

Stratifying Sort on those characteristics

Stratifying Generate those random numbers

Stratifying Sort on those characteristics

Stratifying Assign treatment within each sub-group Result: Equal doctors in each group

Can we really randomize? 31  Randomization does not mean denying people the benefits of the project  Usually existing constraints in project roll-out allow randomization  Randomization often the fairest way to allocate treatment

Use Staggered Roll-out 32 Roll-out to 200 clinics Roll-out to 200 more clinics Roll-out to 400 more clinics Jan 2014 July 2014Jan 2015 Randomize the order in which clinics receive program Compare Jan 2014 group to Jan 2015 group at end of first year

Some groups must get the program 33

Vary treatment INTENSITY OF TREATMENT Malaria Information Campaign 100 villages Malaria Information Campaign + SMS Reminders 100 villages NATURE OF TREATMENT Radio campaign 100 villages Newspaper campaign 100 villages 34

Randomization is often the fairest 35 Watch the movie! [link]link Randomization of an early child development program in Côte d’Ivoire

What if randomization is impossible? 36  Think again: It often is possible on some level, and it’s the best way to get a clear measure of impact  Always begin the IE with imagining what the ideal would look like  With a national policy  Use randomization to test implementation

Key takeaway #1 The single best way to evaluate the unbiased average impact of an activity is by randomizing treatment. 37

Key takeaway #2 It is more ethical to test programs rigorously before universally implementing them than it is to use scarce public resources to implement a universal program with uncertain benefits. 38

Key takeaway #3 Randomization is more flexible than you think:  It does not require withholding of benefits.  It can take advantage of necessary staggered roll-out.  It can test different reforms or packages of services across groups at the same time (so all receive at least some package). 39

Let’s randomize 40

Thank you! 41

BONUS SLIDES

43

LET’S RANDOMIZE! 44 What if we had 100 observations? Treatment # Drs: 26 Control # Drs: 22 Total observations 21 Not a very big sample!