Impact Evaluation Methods

Slides:



Advertisements
Similar presentations
Methodology and Explanation XX50125 Lecture 2: Experiments Dr. Danaë Stanton Fraser.
Advertisements

Advantages and limitations of non- and quasi-experimental methods Module 2.2.
The World Bank Human Development Network Spanish Impact Evaluation Fund.
Non Experimental Design in Education Ummul Ruthbah.
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.
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.
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.
Impact Evaluation “Randomized Evaluations” Jim Berry Asst. Professor of Economics Cornell University.
Non-experimental methods Markus Goldstein The World Bank DECRG & AFTPM.
Experiments. The essential feature of the strategy of experimental research is that you… Compare two or more situations (e.g., schools) that are as similar.
Evaluating Impacts of MSP Grants Ellen Bobronnikov Hilary Rhodes January 11, 2010 Common Issues and Recommendations.
Chapter 10 Finding Relationships Among Variables: Non-Experimental Research.
Evaluation Designs Adrienne DiTommaso, MPA, CNCS Office of Research and Evaluation.
Agenda: Quasi Experimental Design: Basics WSIPP drug court evaluation Outcomes and Indicators for your projects Next time: bring qualitative instrument.
CJ490: Research Methods in Criminal Justice UNIT #4 SEMINAR Professor Jeffrey Hauck.
What is Research Design? RD is the general plan of how you will answer your research question(s) The plan should state clearly the following issues: The.
Comprehensive Evaluation Concept & Design Analysis Process Evaluation Outcome Assessment.
Impact Evaluation Methods Regression Discontinuity Design and Difference in Differences Slides by Paul J. Gertler & Sebastian Martinez.
Looking for statistical twins
Issues in Evaluating Educational Research
Evaluation Requirements for MSP and Characteristics of Designs to Estimate Impacts with Confidence Ellen Bobronnikov March 23, 2011.
EXPERIMENTAL RESEARCH
Measuring Results and Impact Evaluation: From Promises into Evidence
Quasi Experimental Methods I
Constructing Propensity score weighted and matched Samples Stacey L
Quasi Experimental Methods I
Experimental Research Designs
Lesson 3: Experimental Research Designs
Design (3): quasi-experimental and non-experimental designs
TITLE IV-E WAIVER SITE VISIT
Group Quantitative Designs
March 2017 Susan Edwards, RTI International
Quasi-Experimental Methods
Chapter Eight: Quantitative Methods
Designing an Experiment
Making Causal Inferences and Ruling out Rival Explanations
Experiments and Observational Studies
Impact evaluation: The quantitative methods with applications
Matching Methods & Propensity Scores
11/20/2018 Study Types.
Matching Methods & Propensity Scores
Methods of Economic Investigation Lecture 12
Experiments and Quasi-Experiments
Ten things about Experimental Design
Experimental Design.
Development Impact Evaluation in Finance and Private Sector
Experimental Design.
Quasi-Experimental Design
ABAB Design Ethical considerations
Matching Methods & Propensity Scores
Experiments and Quasi-Experiments
The Nonexperimental and Quasi-Experimental Strategies
Impact Evaluation Methods: Difference in difference & Matching
Evaluating Impacts: An Overview of Quantitative Methods
Explanation of slide: Logos, to show while the audience arrive.
Experiments: Part 2.
Class 2: Evaluating Social Programs
Sampling for Impact Evaluation -theory and application-
Class 2: Evaluating Social Programs
Methods and Approaches to investigate the UK Education System
Study on Method of Mass Communication Research 传播研究方法 (7&8) Dr
Experiment Basics: Designs
Positive analysis in public finance
RESULTS Methodological characteristics
Reminder for next week CUELT Conference.
Misc Internal Validity Scenarios External Validity Construct Validity
Types of Statistical Studies and Producing Data
Presentation transcript:

Impact Evaluation Methods July 9, 2011 Dhaka Sharon Barnhardt, Assistant Professor Institute for Financial Management and Research (IFMR)

Impact evaluation methods Non- or Quasi-Experimental Methods a. Pre-Post Simple Difference Differences-in-Differences Multivariate Regression Statistical Matching Interrupted Time Series Instrumental Variables Regression Discontinuity

Methods to estimate impacts Let’s look at different ways of estimating the impacts using the data from the schools that got a balsakhi Pre – Post (Before vs. After) Simple difference Difference-in-difference Simple or Multivariate regression Randomized Experiment Program participants before program Individuals who did not participate (data collected after program) Same as above, plus: (data collected before and after) Same as above plus: Also have additional “explanatory” variables

1 - Pre-post (Before vs. After) Average change in the outcome of interest before and after the programme Example: when establishing a causal link is not feasible measuring the impact of a govt. run literacy campaign state-wide; where it is difficult to construct a comparison group since everyone in the state is mandated to receive the “treatment” at the same time (Sakshar Bharat Campaign in India) Issues: does not take into account time-trend “response-shift bias” – change in the participant’s metric for answering questions from the pre to the post test Response shift bias is described as a “change in the participant’s metric for answering questions from the pre test to the post test due to a new understanding of a concept being taught (Klatt and Taylor-Powell, 2005).” For example, if a program is designed to teach parents effective discipline for children and some participants don’t know that spanking a child is considered inappropriate, they may report at the beginning of the program (pre test) that they are engaging in age appropriate discipline practices at home. However, if over the course of the program they begin to understand the relationship between age of a child and the “appropriate” discipline techniques, they may also report that they understand age appropriate discipline on the post testt – and thus, their evaluation data would reflect no program impact at all.

1 - Pre-post (Before vs. After) QUESTION: Under what conditions can this difference (26.42) be interpreted as the impact of the balsakhi program? Average post-test score for children with a balsakhi 51.22 Average pretest score for children with a balsakhi 24.80 Difference 26.42

Pre-post Method 1: Before vs. After Impact = 26.42 points? 75 50 25 26.42 points? 2002 2003

2 - Simple difference Example: Issues: A post- programme comparison of outcomes between the group that received the programme and a “comparison” group that did not Example: programme is rolled out in phases leaving a cohort for comparison, even though assignment of treatment is not random if Sakshar Bharat was rolled out in a few districts only at a time Issues: does not take into account differences that exist before the treatment (selection bias)

2 - Simple difference Compare test scores of… With test scores of… Children who got balsakhi Children who did not get balsakhi

2 - Simple difference QUESTION: Under what conditions can this difference (-5.05) be interpreted as the impact of the balsakhi program? Average score for children with a balsakhi 51.22 Average score for children without a balsakhi 56.27 Difference -5.05

What would have happened without balsakhi? Method 2: Simple Comparison Impact = -5.05 points? 75 50 25 -5.05 points? 2002 2003

3 – Difference-in-Differences or Double Difference Comparison of outcome between a treatment and comparison group (1st difference) and before and after the programme (2nd difference) Suitability: programme is rolled out in phases leaving a cohort for comparison, even though assignment of treatment is not random If Sakshar Bharat was rolled out in a few districts only at a time Issues: failure of “parallel trend assumption”, i.e. impact of time on both groups is not similar

3 – Difference-in-Differences Compare gains in test scores of… With gains in test scores of… Children who got balsakhi Children who did not get balsakhi

3 - Difference-in-differences Pretest Post-test Difference Average score for children with a balsakhi 24.80 51.22 26.42

What would have happened without balsakhi? Method 3: Difference-in-differences 75 50 25 26.42 2002 2003

3 - Difference-in-differences Pretest Post-test Difference Average score for children with a balsakhi 24.80 51.22 26.42 Average score for children without a balsakhi 36.67 56.27 19.60

What would have happened without balsakhi? Method 3: Difference-in-differences 75 50 25 19.60 6.82 points? 26.42 2002 2003

3 - Difference-in-differences QUESTION: Under what conditions can 6.82 be interpreted as the impact of the balsakhi program? Pretest Post-test Difference Average score for children with a balsakhi 24.80 51.22 26.42 Average score for children without a balsakhi 36.67 56.27 19.60 6.82

4- Simple/Multivariate Regression Change in outcome in the treatment group controlling for observable characteristics requires theorizing on what observable characteristics may impact the outcome of interest besides the programme Issues: how many characteristics can be accounted for? (omission variable bias) requires a large sample if many factors are to be controlled for

4 - Regression: controlling for pretest

Impact of Balsakhi - Summary Method Impact Estimate (1) Pre-post 26.42* (2) Simple Difference -5.05* (3) Difference-in-Difference 6.82* (4) Regression 1.92 * Statistically significant at the 5% level

Impact of Balsakhi - Summary Method Impact Estimate (1) Pre-post 26.42* (2) Simple Difference -5.05* (3) Difference-in-Difference 6.82* (4) Regression 1.92 (5) Randomized Experiment 5.87* * Statistically significant at the 5% level

Impact of Balsakhi - Summary Method Impact Estimate (1) Pre-post 26.42* (2) Simple Difference -5.05* (3) Difference-in-Difference 6.82* (4) Regression 1.92 (5) Randomized Experiment 5.87* *Statistically significant at the 5% level Bottom Line: Which method we use matters!

Another Example: Jaunpur Study 280 villages in Jaunpur, UP Intervention: Provided information to communities on education status and responsibilities of VECs Encouraged to test children and create community report cards to assess status of education Trained volunteers in villages to conduct after-school reading classes

Jaunpur Study Different impact estimates on reading level scores Method Impact Estimate (1) Pre-post 0.6* (2) Simple Difference -0.68* (3) Difference-in-Difference 0.31* (4) IV Regression 0.88* (5) Propensity Score Matching -0.03

Jaunpur Study Different impact estimates on reading level scores Method Impact Estimate (1) Pre-post 0.6* (2) Simple Difference -0.68* (3) Difference-in-Difference 0.31* (4) IV Regression 0.88* (5) Propensity Score Matching -0.03 (6) Randomized Experiment 0.06*