RESEARCH DESIGN IIINTRODUCTION AND COURSE OBJECTIVESCourse Reading ResourcesTOPICS AND READINGSI. Introduction and OverviewII. Causal Logic, Control and ValidityA. Causal Order: First Principles1. The Logic of Causal InferenceSC2, “Preface” and Chapter 1 (pp. 1-18).2. Control: Varieties and ImplicationsData Analysis: varieties of multivariate analysisVijverberg (1997), “The Quantitative Methods Component….”Journal of Policy Analysis and Management.Design Type: nonexperimental; experimental; quasi-experimentalTrochim, “Design” Chap. 6.3. Public Policy Research Design Issues and IllustrationsShadish, William and David Myers (2004), “Research Design Policy Brief” http://www.campbellcollaboration.org/MG/ResDesPolicyBrief.pdfGreene, Jay P., Paul E. Peterson, and Jiangtao Du, “School Choice in Milwaukee: A Randomized Experiment. BF (Chap 21).Peterson, Peter (1999). “Vouchers and Test Scores.” Policy Review, pp. 10ff.2. Timing: Locating Effects in Time4. Time-Mapping Causal ProcessIV. Experimental vs. Econometric EstimationA. Impact Designs: OverviewBF, “Evaluation Designs (Chap. 2).B. Econometric Analysis: Logic and Limitations2. Are Matched Comparison Groups an Adequate Substitute forRandomization?3. Fixed Effects as a Control StrategyFraker and Maynard, op. cit. (see esp. pp. 211-12; 223-24).V. Using Research Design to Improve Econometric EstimatesA. Ex Post Facto Designs1. Design Attributes and Internal ValidityEx Post Facto Design Applicationsa) Designs Lacking Either a Control Group or a Pretestb) Designs That Use Both a Control Group and a Pretestc) Comparative Change Designd) Interrupted Time-Series Designse) Comparative Time-Series Designf) Criterion Population DesignVI. Measurement Issues and Approaches1. Factor Scaling Logic and Techniques2. Measurement Validity and Reliability IssuesTrochim, “Measurement” (Chap 3).O’Regan, K. and M. Wiseman, “Using Birth Weights to Chart the Spatial Distribution of Urban Poverty,” Urban Geography, Vol. 11, 1990, pp. 217-233. [Note: This study illustrates the validation of a new measure of poverty.]VII. Sampling Concepts, Designs, and StrategiesA. Sampling Logic and MethodsTrochim, “Sampling” Chap. 2.Nachimas and Nachimas, “Sampling and Sampling Designs” (Chap 8.) in Research Methods in the Social Sciences, 1987.VIII. Conclusions and OverviewRESEARCH DESIGN IIInstructor: D. A. Hicks Office: GR 3.804Fall 2007 Vox: 972.883.2733Tuesday 4:00-6:45 PM Email: [email protected] AND COURSE OBJECTIVES“[I]n the interplay between design and statistics, designrules!”(Shadish, et al., 2002:xvi)The principal task of all science is to explain. The framework for explanation is the cause-and-effect sequence. The “science” part of social science indicates our dependence on replicable methods as the means of isolating causal sequences fromother influences. We cannot use science to prove something. The most we can ever do is gradually strengthen our confidence in the validity of a causal sequence by eliminating possible alternative explanations. We accomplish this by repeated testing. Testing involves introducing “controls” for alternative explanations suggested by theory or intuition. And it is this process that social science shares with any and all other sciences. Science is an empirical activity; the coin of the realm is data. We derive understanding from data—generally expressed as numbers—by developing them in disciplined ways. The resulting computations – (data analysis) -- involve attempts to characterize and summarize sets of data so that we might identify and trace possible causal relationships between measured attributes (variables) of a sample ofunits of analysis. Inevitably, the quality of the science we do depends on both the quality of the concepts and measurements underlying our data and the integrity of the techniques we use to develop them. The best known way of controlling (“holding constant”) extraneous factors—and thereby eliminating alternative explanations—involves the use of statistical techniques. The field of “econometrics” illustrates the emphasis and effort that have gone into the development of such techniques. As powerful and useful as these techniques can be, however, they focus only on one stage—and a latter one atthat—of the larger research process. Much can be done at prior stages to develop our data in ways that can lessen dramatically our reliance on ex post facto statisticaltechniques to control unwanted influences. The field of “research design” which we will explore together this semester illustrates the emphasis and effort that have gone into the development of such techniques.Research design, then, emphasizes techniques for organizing a research process to achieve variance control prior to – and during—data collection and development, rather than doing so only after-the-fact through the use of data analytic (statistical) techniques. This long statement should now be reasonably clear to you. The focus of research design is on identifying junctures in the research process at which steps can be taken to help clarify cause-effect relationships traced out in complex researchsettings. Research design is all about how to “develop” and organize data in the2first place in ways that can enhance our ability to interpret it later. In that sense, research design is conceptually prior to statistical analysis. The more thoughtful we can be about the quality of the data we develop for subsequent analysis, the fewer compromises we have to accept later when selecting and using analysis tools/techniques.This course has been designed for the graduate student who has had at least one course in research methods and completed both Research Design (POEC 5312) and a graduate-level course in data analysis. S/he should be familiar with multiple regression techniques such as Ordinary Least Squares (OLS). This course is intendedto be of value both to those who will go on to conduct their own research in their chosen fields and to those who intend to use/evaluate the findings of research conducted by others. COURSE READING RESOURCESCourse readings will be drawn from a variety of sources. In addition to the books listed below, a variety ofarticles and related instructional materials have been posted on the WebCT website. You may access WebCT from the UT-D Galaxy site