Opportunities, Challenges and Recent Trends in Meta-Analysis: A Journal Editor’s Perspective Andreas Schwab, Ph.D. Associate Professor Iowa State University Visiting Senior Fulbright Scholar Institut Teknologi Bandung June 24, 2014 .Copyright © 2014 Andreas Schwab -- All rights reserved. Personal Background Institutional Affiliations US Senior Fulbright Scholar Visiting faculty member Institute of Technology Bandung since September 2013 Associate Professor of Management College of Business at Iowa State University Personal Background Involvement in Methodological Change Initiatives Alternatives to Null-Hypothesis Significance Tests PDWs at Annual Academy of Management Conference Schwab et al. (2011) Organization Science Numerous invited presentations Bayesian Statistics: Potential Advantages and Applications PDWs at Annual Academy of Management Conference Personal Background Editorial Expertise Contributing Editor for Research Methodology and Statistics at Entrepreneurship, Theory and Practice (5Y- ImpF: 2.2) Member Editorial Boards Organization Science (5.5) Strategic Entrepreneurship Journal (3.1) Journal of Small Business Management (2.1 ) Special Issue Editor Organization Studies "Temporary Organizations“ (3.2) IJESB “Sustainable Entrepreneurial Eco-Systems” Whom else you would like to be here? Ray Ragby Baylor University Danny Miller HEC Montreal Angelo S. DeNisis Tulane University Andreas Schwab Iowa State University Senior Editorial Board of Entrepreneurship Theory and Practice Today's Agenda Introduction into Meta Analysis (brief) History Effect Size vs Statistical Significance Meta-Analytic Thinking and Scientific Progress Implications for Academic Journals (speculative) Submission Process Review Process Publication Process If I get to excited or confusing, slow me down with a question! What to expect from today's session Motivation to consider meta analysis as a useful methodology for your own research Appreciation of opportunities of meta analysis for accumulation of knowledge Anticipate and contribute to related appropriate changes in publication process Brief History of Meta Analysis Behavioral Sciences Gene Glass, University of Colorado Boulder Presentation to American Educational Research Association (1976) Quantitative methodology to aggregate findings from multiple prior empirical studies Brief History of Meta Analysis Medical Sciences Thomas Chalmers , US physician and medical administrator Quantitative approaches to systematically combine findings from available studies (mid-1970) Institutionalization of systematic collection, archiving and analysis of all controlled clinical trials (Cochrane Collaboration) Accumulative Nature of Scientific Progress Any single study is not enough … Statistical power issues Measurement issues Unknown boundary conditions Dynamic nature of research context … scientific progress depends on the accumulation of knowledge across studies. Traditional “Qualitative” Aggregation of Prior Empirical Findings Key Limitations of “Traditional” Review Articles Conflicting findings Complex differences between studies Large number of relevant studies Current Solution: Focus on a few “high quality” studies No general agreement on “quality” criteria Only move from “one study” to “few studies” Meta Analysis Quantitative integration of findings in prior related empirical studies Estimates standardized effect size Estimates ES credibility intervals, prediction intervals and confidence intervals Options to account for: Sampling error Measurement errors Other artifacts and causal factors Meta Analysis Potential Advantages Draws on all relevant studies Increases statistical power Can capture design and empirical context related moderating effects Focus on effect sizes (not stat. significance) Focus on credibility, prediction, and confidence intervals (not p-values) Type of Effect Size Measure for Meta Analysis Effect sizes measures estimate A measure of associational strength Comparable across studies Independent of sample size With a computable standard error Alternative Measures include: Bivariate correlation coefficient Partial correlation coefficient Others Effect Sizes for Meta Analysis: Unique vs Standardized Dialectic tension between alternative approaches to effect size evaluations Adjust to specific empirical research question, research design and empirical context versus Use similar measures across studies for the aggregation of findings across studies Solution: Estimate, interpret, and report both! Weighting of Effect Sizes Effect sizes in some studies more precise than in others Alternative indices of precision Sample size weight Inverse variance weight They provide a statistical basis for estimating Standard error of the mean effect size Confidence intervals Homogeneity tests Comprehensive Meta-Analytic Study: Alternative Key Models Several relevant estimates of linear model Overall size of an effect (HOMA ) Mean simple (or partial) correlation coefficient Homogeneity of ES estimates to determine need for moderator investigation Probing for publication bias Effects of moderators (MARA) measurement method, research context, estimation procedure, model specification Entire theoretical models (MASEM analysis) Probing for Publication Bias Funnel Graphs Scatter diagram of precision (e.g., sample size, 1/SE) vs. estimated effect (e.g., r) Estimates vary randomly and symmetrically around true effect in the absence of publication bias Inverted funnel shape is dictated by predictable normal distribution of chancebased heteroskedasticity Funnel Plots: An Example Publication Bias vs Questionable Research Practices Likely upward bias in mean effect sizes and correlations (Bakker et al., 2012; Levine et al. 2009) Publication bias more likely for small sample studies (lack of power) Questionable research practices prevalent and serious threat in management research (Simmons et al. 2011) Statistical significance of <5% offers only weak false-positives protection Starbuck, 1988). (Webster & Recommended literature Cumming, Geoff (2011): Understanding the new statistics: Effect sizes, confidence intervals, and meta-analysis. Routledge, New York. (Good fundamental introduction) Meta Analysis Literature: Books Cooper & Hedges (2009), The handbook of research synthesis and meta-analysis. (Great book; combination of very relevant papers) Lipsey & Wilson (2001), Practical metaanalysis. (Good introductory text) Stanley & Doucouliagos (2012), Meta- regression in economics and business. (Most advanced) Meta Analysis Literature: Journal Articles Aguinis, Pierce, Bosco, Dalton and Dalton (2010) Debunking myths and urban legends about meta-analysis. Organizational Research Methods, 14, 306-331. Cortina (2003), Apples and oranges (and pears, Oh my!): The search for moderators in meta-analysis, Organizational Research Methods, 6, 415-439. Geyskens, Krishnan, Steenkamp and Cunha (2009) A review and evaluation of meta-analysis practices in management research. Journal of Management, 35, 393-419. Stanley (2001), Wheat from chaff: Meta-analysis as quantitative literature review, Journal of Economic Perspectives, 15, 131150. Stanley and Jarrell (2005), Meta-regression analysis: A quantitative method of literature surveys, Journal of Economic Surveys, 19, 299-308. Meta Analysis: Key Challenges Current limitations of Meta Analysis applications Key information not reported in prior empirical studies Key design and context differences not reported/recognized in prior empirical studies Focus on single highly standardized effect size measures across studies (e.g., bivariate correlations) Need to further improve methods and methods understanding (requiring often problematic conversions) Increasing Number of Meta Analyses in Entrepreneurship Research Figure 1 Number of Meta-Analytic Studies in Leading Entrepreneurship and Management Journals (1990-2013) 18 Management Journals 16 14 12 10 8 6 Entrepreneurship Journals 4 2 0 1990 1992 1994 1996 1998 2000 2002 Years 2004 2006 2008 2010 2012 Meta-Analytic Thinking and Scientific Progress Promising approach to theory development Move away from “single study” theory test approaches Embrace iterative development and adjustment of models and theory … and goes far beyond the concept of “replication studies” Meta-Analysis and Replication Studies Replication studies are not enough! (Schmidt 2014) Exact future replications only probes for sampling error effects Future studies that differ in a few key features can probe for other moderating effects (e.g., measurement error, boundary conditions) Accumulation of knowledge in pool of studies Reasonably heterogeneous > homogeneous Opportunities, Challenges and Recent Trends in Meta-Analysis How can academic journals help? One Journal Editor’s Perspective My Personal Convictions (Biases) Combination of single study empirical theory tests using statistical significance is detrimental to scientific progress Inherent power problems Statistical significance flawed indicator for importance of effects for theory or practice Meta Analysis has potential to support meaningful iterative model and theory development across studies Improves upon straight replications Improves upon conceptual review articles Meta Analysis Opportunities and Publication Process Potential Implication for Academic Publication Process Stronger focus on meta-analysis compared to traditional review articles Provide necessary information for future meta analyses Proposal to Support Meta Analyses in Entrepreneurship Journals Changes in all stages of academic publication process Pre-Submission Review Copy Editing and Distribution After-Print Review-Oriented Journals Review Journals should (primarily) publish meta analyses Encourage collaborations among phenomenon experts and meta-analysis experts Develop editorial board with metaanalytical expertise Journals Publishing Original Empirical Studies Journals should collect the necessary information for future meta analyses Effect size Credibility, Prediction and Confidence intervals Relevant study design features Measurement, sampling process, data coding Empirical context features Solution Adjust submission guidelines (= new requirements) Reinforce requirements during review and copyediting stage Journals Publishing Original Empirical Studies Archive and distribute MA-relevant information for future meta analyses using Printed article Text Appendix Online storage and retrieval systems Systematic Detailed Open-access Solution: Online journal archives! Publication Bias: Journal Archives are not enough Need for institutionalized collection, archiving and dissemination of prior empirical studies similar to Cochrane Collaboration Government (NSF) Professional associations (AOM) Research foundations (Kauffmann) Publishing in Top Entrepreneurship Journals: Opportunities, Challenges and Recent Trends Questions and Thoughts . Opportunities, Challenges and Recent Trends in Meta-Analysis: A Journal Editor’s Perspective Further Thoughts and Ideas? . Opportunities, Challenges and Recent Trends in Meta-Analysis: A Journal Editor’s Perspective Thank You! . Opportunities, Challenges and Recent Trends in Meta-Analysis: A Journal Editor’s Perspective Andreas Schwab, Ph.D. Associate Professor Iowa State University Visiting Senior Fulbright Scholar Institut Teknologi Bandung June 24, 2014 . Entrepreneurship, Theory and Practice Aims and Scope of ETP Leading ENT scholarly journal (others JBV, ETP, SEJ) Original conceptual and empirical papers that advance the field of entrepreneurship Official journal of "United States Association for Small Business and Entrepreneurship" (USASBE) Entrepreneurship, Theory and Practice Review Process Structure Single action editor 'Double-blind' review process Three independent reviewers Three-month review cycle Rarely (= never) straight acceptance 1 - 3 rounds of respectfully worded and constructive criticism Acceptance rate < 10%