1 Managing Threats to Randomization. Threat (1): Spillovers If people in the control group get treated, randomization is no more perfect Choose the appropriate.

Slides:



Advertisements
Similar presentations
II. Potential Errors In Epidemiologic Studies Random Error Dr. Sherine Shawky.
Advertisements

Choosing the level of randomization
A Guide to Education Research in the Era of NCLB Brian Jacob University of Michigan December 5, 2007.
Presented by Malte Lierl (Yale University).  How do we measure program impact when random assignment is not possible ?  e.g. universal take-up  non-excludable.
What could go wrong? Deon Filmer Development Research Group, The World Bank Evidence-Based Decision-Making in Education Workshop Africa Program for Education.
Quasi-Experimental Design
The Bahrain Branch of the UK Cochrane Centre In Collaboration with Reyada Training & Management Consultancy, Dubai-UAE Cochrane Collaboration and Systematic.
Sensitivity Analysis for Observational Comparative Effectiveness Research Prepared for: Agency for Healthcare Research and Quality (AHRQ)
Estimation of Sample Size
WHY LARGE-SCALE RANDOMIZED CONTROL TRIALS? David Myers Senior Vice President IES 2006 Research Conference David Myers Senior Vice President IES 2006 Research.
AP Statistics Chapter 5 Notes.
Validity, Sampling & Experimental Control Psych 231: Research Methods in Psychology.
Program Evaluation Spero Manson PhD
Agenda: Block Watch: Random Assignment, Outcomes, and indicators Issues in Impact and Random Assignment: Youth Transition Demonstration –Who is randomized?
Critical Appraisal of an Article by Dr. I. Selvaraj B. SC. ,M. B. B. S
I want to test a wound treatment or educational program but I have no funding or resources, How do I do it? Implementing & evaluating wound research conducted.
1 Randomization in Practice. Unit of randomization Randomizing at the individual level Randomizing at the group level –School –Community / village –Health.
Intervention Studies Principles of Epidemiology Lecture 10 Dona Schneider, PhD, MPH, FACE.
Moving from Development to Efficacy & Intervention Fidelity Topics National Center for Special Education Research Grantee Meeting: June 28, 2010.
Sociology 3322a. “…the systematic assessment of the operation and/or outcomes of a program or policy, compared to a set of explicit or implicit standards.
Epidemiology The Basics Only… Adapted with permission from a class presentation developed by Dr. Charles Lynch – University of Iowa, Iowa City.
Measuring Impact: Experiments
AADAPT Workshop South Asia Goa, December 17-21, 2009 Nandini Krishnan 1.
1 What are Monitoring and Evaluation? How do we think about M&E in the context of the LAM Project?
Learning Objectives Copyright © 2004 John Wiley & Sons,Inc Primary Data Collection: Experimentation CHAPTER Seven.
Criteria for Assessing The Feasibility of RCTs. RCTs in Social Science: York September 2006 Today’s Headlines: “Drugs education is not working” “ having.
Shawn Cole Harvard Business School Threats and Analysis.
 Internal Validity  Construct Validity  External Validity * In the context of a research study, i.e., not measurement validity.
Evidence-based policies and indicator systems 2006 Conducting effective research and analysis to support policy delivery. The Green Book: Appraisal and.
INDUSTRIAL ECONOMICS, INCORPORATED IEc INDUSTRIAL ECONOMICS, INCORPORATED Measuring Impact of Compliance Assistance on Auto Body Shops using an Experimental.
Group PresentaTion Template
Successful Concepts Study Rationale Literature Review Study Design Rationale for Intervention Eligibility Criteria Endpoint Measurement Tools.
1 Experimental Research Cause + Effect Manipulation Control.
Impact Evaluation “Randomized Evaluations” Jim Berry Asst. Professor of Economics Cornell University.
Research Design ED 592A Fall Research Concepts 1. Quantitative vs. Qualitative & Mixed Methods 2. Sampling 3. Instrumentation 4. Validity and Reliability.
Africa Impact Evaluation Program on AIDS (AIM-AIDS) Cape Town, South Africa March 8 – 13, Steps in Implementing an Impact Evaluation Nandini Krishnan.
1 Study Design Issues and Considerations in HUS Trials Yan Wang, Ph.D. Statistical Reviewer Division of Biometrics IV OB/OTS/CDER/FDA April 12, 2007.
1 Module 3 Designs. 2 Family Health Project: Exercise Review Discuss the Family Health Case and these questions. Consider how gender issues influence.
The Disability Employment Initiative (DEI): Impact Evaluation Design October 21, 2015 Sung-Woo Cho, Ph.D.
TRANSLATING RESEARCH INTO ACTION What and How to randomize? July 9, 2011 Dhaka Raymond Guiteras, Assistant Professor University of Maryland povertyactionlab.org.
CHOOSING THE LEVEL OF RANDOMIZATION. Unit of Randomization: Individual?
Lecture 2: Evidence Level and Types of Research. Do you recommend flossing to your patients? Of course YES! Because: I have been taught to. I read textbooks.
Developing an evaluation of professional development Webinar #2: Going deeper into planning the design 1.
Bilal Siddiqi Istanbul, May 12, 2015 Measuring Impact: Non-Experimental Methods.
Types of Studies. Aim of epidemiological studies To determine distribution of disease To examine determinants of a disease To judge whether a given exposure.
Framing a research question Chitra Grace A Scientist- C (PGDHE) NIE, Chennai RM Workshop for ICMR Scientists 01/11/2011.
Public Finance and Public Policy Jonathan Gruber Third Edition Copyright © 2010 Worth Publishers 1 of 24 Copyright © 2010 Worth Publishers.
Common Pitfalls in Randomized Evaluations Jenny C. Aker Tufts University.
Statistics 100 Lecture Set 4. Lecture Set 4 Chapter 5 and 6 … please read Read Chapter 7 … you are responsible for all of this chapter Some suggested.
IMPACT EVALUATION PBAF 526 Class 5, October 31, 2011.
Randomized Evaluation: Dos and Don’ts An example from Peru Tania Alfonso Training Director, IPA.
Journal Club Curriculum-Study designs. Objectives  Distinguish between the main types of research designs  Randomized control trials  Cohort studies.
Chapter 12 Quantitative Questions and Procedures.
Methods to Handle Noncompliance
Issues in Evaluating Educational Research
Critically Appraising a Medical Journal Article
Threats and Analysis.
Implementation Issues Program roll-out
Title List your Team Members
Impact Evaluation Methods
Implementation Challenges
Title List your Team Members
Group PresentaTion Template
Evidence Based Practice
Sample Sizes for IE Power Calculations.
Monitoring and Evaluating FGM/C abandonment programs
Title List your Team Members
Steps in Implementing an Impact Evaluation
Misc Internal Validity Scenarios External Validity Construct Validity
Types of Statistical Studies and Producing Data
Presentation transcript:

1 Managing Threats to Randomization

Threat (1): Spillovers If people in the control group get treated, randomization is no more perfect Choose the appropriate unit of randomization to minimize the risk Monitor the implementation May be interesting to analyze as such –Measure the spillovers –Spillovers may reflect a good program with a large effect or an important take-up

Spillover: An Example Deworming -Previous studies randomize deworming treatment within schoools -Suppose that deworming prevents transmission of disease, what problems does this create for evaluation? -Suppose externalities are local, how can we measure total impact? 3 Adapted from Threats and Analysis, Shawn Cole, J-PAL.

4

Measuring total impact in the presence of spillovers Design unit of randomization so that it encompasses the spillovers E.g. if we expect externalities are all within the school: -Randomization at the school level allows for estimation of overall impact 5 Adapted from Threats and Analysis, Shawn Cole, J-PAL.

Measuring total impact in the presence of spillovers 6 Adapted from Threats and Analysis, Shawn Cole, J-PAL.

Threat (2): Attrition Failure to collect outcome data from some individuals who were part of the original sample. Random attrition will only reduce a study's statistical power Attrition that is correlated with the treatment may bias estimates. People who benefited from the program left more People who did benefit from the program staid more Even if attrition rates are similar, people who dropped out may be different

Fighting Attrition Track participants after they leave the program. –Collect good information in the baseline on how to follow people (for example the names of neighbors and relatives) Report attrition levels in the treatment and comparison groups and compare attritors with non-attritors using baseline data (when available). Bound the potential bias of your effect

Attrition bias: an example Problem to be addressed: -Too weak (undernourished) children don’t come to school Intervention: -You start a school feeding program and want to do an evaluation: Treatment & Control Group Expectation: -Weak children start going to school more if they live next to a treatment school Outcome of interest -Weight of children who attend school Measurement: -You go to all the schools (T & C) and measure everyone who is in school on a given day -Will the treatment-control difference in weight be over-stated or understated? (due to attrition bias) Adapted from Threats and Analysis, Shawn Cole, J-PAL.

10 Adapted from Threats and Analysis, Shawn Cole, J-PAL.

Attrition Bias: If only children >20kg come to school 11 Treatment-control difference becomes understated! Adapted from Threats and Analysis, Shawn Cole, J-PAL.

Threat (3): Non-compliance Treatment units do not actually take up the program, or are not served –May lead you to underestimate the impact You can try to prevent it by monitoring the implementation closely But you can see the effect on the treated –Multiplying the observed effect by the inverse proportion of people who complied And non-compliance often teaches a lot about the program…

Conclusions If properly designed and conducted, RCTs provide the most credible assessment of the impact of a program: with other methods, difficult to be sure that we control for all selection bias → risk of overestimation of impact Results from RCTs are easy to understand and much less subject to methodological quibbles Credibility + Ease of understanding =>More likely to convince policymakers and funders of effectiveness (or lack thereof) of program 13

Conclusions (cont.) However, in some cases RCTs are simply not feasible Must be aware of the limitations of any method used –What are the implicit assumptions, the potential bias etc. 14