Differences-in-Differences

Slides:



Advertisements
Similar presentations
MICS4 Survey Design Workshop Multiple Indicator Cluster Surveys Survey Design Workshop Household Questionnaire: Education.
Advertisements

Treatment Evaluation. Identification Graduate and professional economics mainly concerned with identification in empirical work. Concept of understanding.
Advantages and limitations of non- and quasi-experimental methods Module 2.2.
1 Difference in Difference Models Bill Evans Spring 2008.
VIII Evaluation Conference ‘Methodological Developments and Challenges in UK Policy Evaluation’ Daniel Fujiwara Senior Economist Cabinet Office & London.
Impact Evaluation Methods. Randomized Trials Regression Discontinuity Matching Difference in Differences.
Difference in difference models
Differences-in- Differences November 10, 2009 Erick Gong Thanks to Null & Miguel.
ABHISHEK CHAKRAVARTY LECTURE 9 EC336 Economic Development in a Global Perspective.
1 Lecture 20: Non-experimental studies of interventions Describe the levels of evaluation (structure, process, outcome) and give examples of measures of.
Matching Methods. Matching: Overview  The ideal comparison group is selected such that matches the treatment group using either a comprehensive baseline.
TRANSLATING RESEARCH INTO ACTION What is Randomized Evaluation? Why Randomize? J-PAL South Asia, April 29, 2011.
Health Programme Evaluation by Propensity Score Matching: Accounting for Treatment Intensity and Health Externalities with an Application to Brazil (HEDG.
Intergenerational Poverty and Mobility. Intergenerational Mobility Leblanc’s Random Family How does this excerpt relate to what we have been talking about?
Nigeria Impact Evaluation Community of Practice Abuja, Nigeria, April 2, 2014 Measuring Program Impacts Through Randomization David Evans (World Bank)
Non-experimental methods Markus Goldstein The World Bank DECRG & AFTPM.
Differences-in- Differences. Identifying Assumption Whatever happened to the control group over time is what would have happened to the treatment group.
We take a multi-period model of childhood investment, based on Cuhna, Heckman et al (2005), which distinguishes early from late investments. In particular,
THE HEALTH CHALLENGE Sheila Shribman National Clinical Director Children, Young People & Maternity.
Key Health Indicators in Developing Countries and Australia
OVERVIEW OF STUDY DESIGN. COMMUNITY SURVEYS
Child Health.
Differences-in-Differences
Difference-in-Differences Models
Is High Placebo Response Really a Problem in Clinical Trials?
Measuring Results and Impact Evaluation: From Promises into Evidence
Reducing global mortality of children and newborns
Differences-in-Differences
Epidemiologic Measures of Association
Vaccine Efficacy, Effectiveness and Impact
Biostatistics Case Studies 2016
An introduction to Impact Evaluation
Experimental Research Designs
Difference-in-Differences
Quasi-Experimental Methods
The Global Health Initiative of the United Methodist Church
Kevin Croke World Bank Long run educational gains from malaria control in Tanzania: Preliminary results CSAE Conference March 19,
Influenza Vaccine Effectiveness Against Pediatric Deaths:
March 2017 Susan Edwards, RTI International
Nurses' Health Study: Risk of hypertension associated with >1000 µg/day of folate vs
Goals Boosting Diabetes and Pre-Diabetes Screening
Matching Methods & Propensity Scores
2016 AmeriCorps Texas All-Grantee Meeting February 25-26, 2016
11/20/2018 Study Types.
Matching Methods & Propensity Scores
Methods of Economic Investigation Lecture 12
Radiation Protection in Dental Radiology
THE PHILIPPINES EARLY CHILDHOOD DEVELOPMENT PROJECT
Genetics.
Free Distribution or Cost-Sharing
Cross Sectional Designs
CDC Diabetes Stats Estimated percentage of people aged 20 years or older with diagnosed and undiagnosed diabetes, by age group, United States, 2005–2008.
Impact Evaluation Methods
remember to round it to whole numbers
Jeremiah Maller Partner Organization: Operation Smile
Impact Evaluation Toolbox
Matching Methods & Propensity Scores
Implementation Challenges
Impact Evaluation Methods: Difference in difference & Matching
Control of Communicable Diseases and IHR
Sampling for Impact Evaluation -theory and application-
Applying Impact Evaluation Tools: Hypothetical Fertilizer Project
Part 2: Defining Geographic Areas Frank Porell
Adjusted Probability of ITN Use by Risk Group, Senegal, 2010Abbreviations: ITN, insecticide-treated net; WRA, (non-pregnant) women of reproductive age.Pregnant.
Adjusted Probability of ITN Use by Risk Group, Liberia, 2011Abbreviations: ITN, insecticide-treated net; WRA, (non-pregnant) women of reproductive age.Pregnant.
Welcome! Please get out objectives #9-17 for a stamp.
Sample Sizes for IE Power Calculations.
Tuberculosis is an infection that is transmitted though airborne particles. It is an uncommon infection in Canada, but is still seen in indigenous populations.
Unit 4: The Biosphere Human Populations.
Columbia University, Department of Biostatistics
Presentation transcript:

Differences-in-Differences

Identifying Assumption Whatever happened to the control group over time is what would have happened to the treatment group in the absence of the program.

Identifying Assumption Whatever happened to the control group over time is what would have happened to the treatment group in the absence of the program. Pre Post

Identifying Assumption Whatever happened to the control group over time is what would have happened to the treatment group in the absence of the program. Effect of program using only pre- & post- data from T group (ignoring general time trend). Pre Post

Identifying Assumption Whatever happened to the control group over time is what would have happened to the treatment group in the absence of the program. Effect of program using only T & C comparison from post-intervention (ignoring pre-existing differences between T & C groups). Pre Post

Identifying Assumption Whatever happened to the control group over time is what would have happened to the treatment group in the absence of the program. Pre Post

Identifying Assumption Whatever happened to the control group over time is what would have happened to the treatment group in the absence of the program. Effect of program difference-in-difference (taking into account pre-existing differences between T & C and general time trend). Pre Post

Uses of Diff-in-Diff Simple two-period, two-group comparison very useful in combination with randomization, matching, or RD Can also do much more complicated “cohort” analysis, comparing many groups over many time periods

Cohort Analysis Example #1 Bleakley’s paper on malaria eradication in the Americas Treated vs Control – those who were (were not) children in malaria endemic regions Pre vs Post – DDT spraying “In both absolute terms and relative to the comparison group of non-malarious areas, cohorts born after eradication had higher income and literacy as adults than the preceding generations.”

Cohort Analysis Example #2 Frankenburg et. al.’s paper on expansion in health services (midwives) Treated vs Control – those who were (were not) children in areas that eventually received midwives Pre vs Post – government placement of midwives in villages “[T]he nutritional status of children fully exposed to a midwife during early childhood is significantly better than that of their peers of the same age and cohort in communities without a midwife. These children are also better off than children measured at the same age from the same communities, but who were born before the midwife arrived.”

Robustness Checks If possible, use data on multiple pre-program periods to show that difference between treated & control is stable Not necessary for trends to be parallel, just to know function for each If possible, use data on multiple post-program periods to show that unusual difference between treated & control occurs only concurrent with program Alternatively, use data on multiple indicators to show that response to program is only manifest for those we expect it to be (e.g. the diff-in-diff estimate of the impact of ITN distribution on diarrhea should be zero)