Pre-Analysis Plans: Pros and Cons of Limiting Discretion in the Analysis of RCTs Donald P. Green Columbia University.

Slides:



Advertisements
Similar presentations
Critical Reading Strategies: Overview of Research Process
Advertisements

Evidence Battles in Evaluation: How can we do better? Mel Mark Penn State University.
National Human Resources for Health Observatory HRH Research Forum Dr. Ayat Abuagla.
Progress on Registration in Political Science Jeremy M. Weinstein (on behalf of Experiments in Governance and Politics – EGAP)
Registration in political science reflections and proposals Macartan Humphreys.
Pre-analysis plans Module 8.3. Recap on statistics If we find a result is significant at the 5% level, what does this mean? – there is a 5% or less probability.
Psychological Methods
Basic Research Methods CSE EST ISE 323 Spring 2012 Tony Scarlatos.
1 Moderators of Treatment Effects in the General Medicine Literature: Looking for Improvement Nicole Bloser, MHA, MPH University of California, Davis June.
Dynamic Treatment Regimes, STAR*D & Voting D. Lizotte, E. Laber & S. Murphy LSU ---- Geaux Tigers! April 2009.
Dynamic Treatment Regimes, STAR*D & Voting D. Lizotte, E. Laber & S. Murphy ENAR March 2009.
Chapter 13: Descriptive and Exploratory Research
Dynamic Treatment Regimes, STAR*D & Voting D. Lizotte, E. Laber & S. Murphy Psychiatric Biostatistics Symposium May 2009.
Regression Discontinuity (RD) Andrej Tusicisny, methodological reading group 2008.
Research Proposal Development of research question
Scientific method - 1 Scientific method is a body of techniques for investigating phenomena and acquiring new knowledge, as well as for correcting and.
Checklists for Reporting Social Science Experiments Alan Gerber Dave Doherty Conor Dowling.
Psychology and Scientific Research. Experimental Science Definition: inquiry in seeking facts and the search for truth through testing of theories and.
Chapter 4 Principles of Quantitative Research. Answering Questions  Quantitative Research attempts to answer questions by ascribing importance (significance)
8-10% of AP Exam. » Does sleeping less than seven hours a day reduce how long you will live? » Do violent video games make people more aggressive? » Can.
AICT5 – eProject Project Planning for ICT. Process Centre receives Scenario Group Work Scenario on website in October Assessment Window Individual Work.
Funded through the ESRC’s Researcher Development Initiative
Moving from Development to Efficacy & Intervention Fidelity Topics National Center for Special Education Research Grantee Meeting: June 28, 2010.
Bayesian Hierarchical Models of Individual Differences in Skill Acquisition Dr Jeromy Anglim Deakin University 22 nd May 2015.
NSW Curriculum and Learning Innovation Centre Draft Senior Secondary Curriculum ENGLISH May, 2012.
Background to Adaptive Design Nigel Stallard Professor of Medical Statistics Director of Health Sciences Research Institute Warwick Medical School
Scientific Inquiry Mr. Wai-Pan Chan Scientific Inquiry Research & Exploratory Investigation Scientific inquiry is a way to investigate things, events.
Main issues Effect-size ratio Development of protocols and improvement of designs Research workforce and stakeholders Reproducibility practices and reward.
Research Study Design. Objective- To devise a study method that will clearly answer the study question with the least amount of time, energy, cost, and.
A Comparison of 42 Local, National, and International HIA Guidelines Andrew L. Dannenberg, MD, MPH Katherine Hebert, MCRP Arthur M. Wendel, MD, MPH Sarah.
PolMeth2009: Freedman Panel Regression Adjustments to Experimental Data: Do David Freedman’s Concerns Apply to Political Science? Donald P. Green Yale.
A Brief Overview of Core Topics
Measuring Efficiency CRJS 4466EA. Introduction It is very important to understand the effectiveness of a program, as we have discovered in all earlier.
1 Critical Review of Published Microarray Studies for Cancer Outcome and Guidelines on Statistical Analysis and Reporting Authors: A. Dupuy and R.M. Simon.
1 f02kitchenham5 Preliminary Guidelines for Empirical Research in Software Engineering Barbara A. Kitchenham etal IEEE TSE Aug 02.
Introduction to Earth Science Section 2 Section 2: Science as a Process Preview Key Ideas Behavior of Natural Systems Scientific Methods Scientific Measurements.
1 f02laitenberger7 An Internally Replicated Quasi- Experimental Comparison of Checklist and Perspective-Based Reading of Code Documents Laitenberger, etal.
Guidelines for Clinical Trial Registration.. Background. In 2005 the International Committee of Medical Journal Editors (ICMJE) announced that in order.
1 Monitoring indicators of children ’ s victimization in school Mona Khoury-Kassabri and Rami Benbenishty Paul Baerwald School of Social Work and Social.
Chapter 2 What is Evidence?. Objectives Discuss the concept of “best available clinical evidence.” Describe the general content and procedural characteristics.
Le parc japonais est beau et calme La fille japonaise est belle mais bavarde Ritsurin Park, Takamatsu.
Title Page The title page is the first page of your psychology paper. In order to make a good first impression, it is important to have a well-formatted.
Session 6: Other Analysis Issues In this session, we consider various analysis issues that occur in practice: Incomplete Data: –Subjects drop-out, do not.
Developing an evaluation of professional development Webinar #2: Going deeper into planning the design 1.
Unit 11: Evaluating Epidemiologic Literature. Unit 11 Learning Objectives: 1. Recognize uniform guidelines used in preparing manuscripts for publication.
1 Usability Analysis n Why Analyze n Types of Usability Analysis n Human Subjects Research n Project 3: Heuristic Evaluation.
Practical Steps for Increasing Openness and Reproducibility Courtney Soderberg Statistical and Methodological Consultant Center for Open Science.
Webinar on increasing openness and reproducibility April Clyburne-Sherin Reproducible Research Evangelist
April Center for Open Fostering openness, integrity, and reproducibility of scientific research.
Evidence-Based Mental Health PSYC 377. Structure of the Presentation 1. Describe EBP issues 2. Categorize EBP issues 3. Assess the quality of ‘evidence’
Practical Steps for Increasing Openness and Reproducibility Courtney Soderberg Statistical and Methodological Consultant Center for Open Science.
BERKELEY INITIATIVE FOR TRANSPARENCY IN THE SOCIAL SCIENCES Garret Christensen, Research Fellow BITSS and Berkeley Institute for Data Science.
Sara Bowman Center for Open Science | Promoting, Supporting, and Incentivizing Openness in Scientific Research.
STA248 week 121 Bootstrap Test for Pairs of Means of a Non-Normal Population – small samples Suppose X 1, …, X n are iid from some distribution independent.
Statistical Analysis Plans EEF Evaluators’ Conference 2016 Dr Ben Styles Head of NFER’s Education Trials Unit.
Chapter 12 Quantitative Questions and Procedures.
Analysis for Designs with Assignment of Both Clusters and Individuals
Distinguish between an experiment and other types of scientific investigations where variables are not controlled,
Principles of Quantitative Research
Center for Open Science: Practical Steps for Increasing Openness
Statistical Approaches to Support Device Innovation- FDA View
Transparency increases the credibility and relevance of research
A Registry of Efficacy & Effectiveness Studies Supported under grant IES R305U to the Society for Research on Educational Effectiveness Jessaca.
Dorothy V. M. Bishop University of
Module 4 Finding the Evidence: Individual Trials
AICT5 – eProject Project Planning for ICT
Regulatory Perspective of the Use of EHRs in RCTs
Some Further Considerations in Combining Single Case and Group Designs
Methodology & Open Science
Presentation transcript:

Pre-Analysis Plans: Pros and Cons of Limiting Discretion in the Analysis of RCTs Donald P. Green Columbia University

Overview of talk What is a registry? What are pre-analysis plans? Arguments in favor of pre-analysis plans What do pre-analysis plans look like in practice? Critiques and resistance A proposal for reducing the burden that pre- analysis plans impose on researchers: “standard operating procedures” An example SOP Next steps

Distinguishing between pre-registration and pre-analysis plans Pre-registration describes a study in a public, searchable format but need not make any commitments with regard to analysis Usually registration occurs prior to the launch of an RCT, but the AEA website allows for retrospective registration Pre-analysis plans lay out specific coding and analysis rules (e.g., the script to analyze the data) – Prior to the launch of the study or prior to obtaining outcome measures?

Social Science Registries: Source APSA Roundtable

Arguments in favor of registration Facilitates meta-analysis Assembling a more comprehensive set of studies lessens the “file-drawer problem” – Assuming that one can ascertain the findings obtained by registered but unpublished studies

Arguments in favor of pre-analysis plans for RCTs (and prospective observational studies) Limiting discretion (and cherry-picking) Example: Casey, Glennerster, and Miguel (2012) “Reshaping Institutions: Evidence On Aid Impacts Using A Preanalysis Plan” JPE Modest overall results but many opportunities for cherry-picking, naïve p-values that change markedly after correcting for multiple comparisons Simonson: inventory of model specifications

Simonsohn’s inventory as applied to Bertrand & Mullainathan (2004)

Arguments in favor of pre-analysis plans for RCTs (and prospective observational studies) Allows readers to more accurately gauge which analyses were planned and which were derived ex post Does NOT preclude exploratory analysis but makes the distinction between exploratory and confirmatory more transparent

More arguments in favor Forces the researchers (and their funders?) to do more careful thinking up front Example: Humphreys et al. evaluation of community driven reconstruction in the Democratic Republic of Congo – Authors did a dry run with funders using simulated data so that everyone was clear ex ante of what the analyses would be and how they would be interpreted in light of priors

Potentially changes the journal review process Upcoming issue of Comparative Political Studies, with articles accepted based on their designs, pending results – Similar plan for J. of Experimental Political Science Lingering issue of monitoring and enforcement “Badges” issued by third-party groups? Initiative underway coordinated by the Center for Open Science

Critiques and resistance Kills innovation Produces damaging inflexibility Creates an unaccountable gatekeeper Will not stop fraud Exposes researchers to threat of being scooped Only meaningful for prospective analyses Creates unnecessary bureaucratic hurdles for scholars Replication is a better use of finite resources Generate inequalities across scholars and methods

What do pre-analysis plans look like in practice? Currently no standard format, although work is underway to create an R package Browsing the major registries suggests that the level of detail varies widely Some include code, while others offer a verbal characterization of measurement and modeling choices Foreshadows the push for SOP

A proposal: Standard Operating Procedures SOP: a set of default procedures that a researcher adopts when there is a gap in the pre-analysis plan (or no pre-analysis plan at all) Date-stamped and public, with versioning May be adapted to suit other labs’ substantive and methodological requirements (e.g., our SOP says nothing about biomedical procedures)

An Example: the SOP for our lab Lin, Green, and Coppock (2015) SOP SOP Safety net procedures rather than a manual of best practices, in part because best practices are often subject to debate (e.g., optimal covariate selection)

SOP contents and omissions Included: Covariates, noncompliance, attrition, estimation under clustered assignment or assignment with varying treatment probabilities, hypothesis testing – Blinded juries for handling idiosyncratic issues (e.g., covariates discovered after the fact) – Included: Coding and analysis choices that are specific to voter mobilization research Omitted: Multiple comparisons, multiple treatment arms…our SOP will grow over time

Next steps Reducing the burden of writing pre-analysis plans through defaults and automation Rebutting common misunderstandings about pre-analysis plans (e.g., exploratory analysis is forbidden) Continued reflection on sticky institutional questions such as monitoring and adjudication