《财政学罗森第九版英文课件：Chap002 Tools of Positive Analysis》由会员分享，可在线阅读，更多相关《财政学罗森第九版英文课件：Chap002 Tools of Positive Analysis（13页珍藏版）》请在金锄头文库上搜索。
1、1-1CHAPTER 2Tools of Positive AnalysisCopyright 2010 by the McGraw-Hill Companies, Inc. All rights reserved.McGraw-Hill/Irwin2-2The Role of TheoryEconomic modelsVirtue of simplicityJudging a modelLimitations of modelsEmpirical analysis2-3Causation vs. CorrelationStatistical analysisCorrelationContro
2、l groupTreatment groupConditions required for government action X to cause societal effect YX must precede YX and Y must be correlatedOther explanations for any observed correlation must be eliminated2-4Experimental StudiesBiased estimatesCounterfactualExperimental (or randomized) study2-5Conducting
3、 an Experimental StudyRandom assignment to control and treatment groups2-6Pitfalls of Experimental StudiesEthical issuesTechnical problemsResponse biasImpact of limited duration of experimentGeneralization of results to other populations, settings, and related treatmentsBlack box aspect of experimen
4、ts2-7Observational StudiesObservational study empirical study relying on observed data not obtained from experimental studySources of observational dataSurveysAdministrative recordsGovernmental dataEconometricsRegression analysis2-8Conducting an Observational StudyL = 0 + 1wn + 2X1 + + nXn + Depende
5、nt variableIndependent variablesParametersStochastic error termRegression analysisRegression lineStandard errorwnL0Interceptis 0Slopeis 12-9Types of DataCross-sectional dataTime-series dataPanel data2-10Pitfalls of Observational StudiesData collected in non-experimental settingSpecification issues2-
6、11Quasi-Experimental StudiesQuasi-experimental study (= natural experiment) observational study relying on circumstances outside researchers control to mimic random assignment2-12Conducting a Quasi-Experimental StudyDifference-in-Difference quasi-experimentsInstrumental Variables quasi-experimentsRegression-Discontinuity quasi-experiments2-13Pitfalls of Quasi-Experimental StudiesAssignment to control and treatment groups may not be randomNot applicable to all research questionsGeneralization of results to other settings and treatments
