1. McGraw-Hill/Irwin Copyright © 2008 by The McGraw-Hill Companies, Inc. All rights reserved. CHAPTER 2 Tools of Positive Analysis

2. 2-2 The Role of Theory  Economic models virtue of simplicity judging a model limitations of models  Empirical analysis

3. 2-3 Causation vs. Correlation  Statistical analysis Correlation Control group Treatment group  Conditions required for government action X to cause societal effect Y X must precede Y X and Y must be correlated Other explanations for any observed correlation must be eliminated

4. 2-4 Experimental Studies  Biased estimates  Counterfactual  Experimental (or randomized) study

5. 2-5 Conducting an Experimental Study  Random assignment to control and treatment groups

6. 2-6 Pitfalls of Experimental Studies  Ethical issues  Technical problems  Response bias  Impact of limited duration of experiment  Generalization of results to other populations, settings, and related treatments  Black box aspect of experiments

7. 2-7 Observational Studies  Observational study – empirical study relying on observed data not obtained from experimental study  Sources of observational data Surveys Administrative records Governmental data  Econometrics Regression analysis

8. 2-8 Conducting an Observational Study  L = α 0 + α 1 w n + α 2 X 1 + … + α n X n + ε Dependent variable Independent variables Parameters Stochastic error term  Regression analysis Regression line Standard error wnwn L α0α0 Intercept is α 0 Slope is α 1

9. 2-9 Types of Data  Cross-sectional data  Time-series data  Panel data

10. 2-10 Pitfalls of Observational Studies  Data collected in non-experimental setting  Specification issues

11. 2-11 Quasi-Experimental Studies  Quasi-experimental study (= natural experiment) – observational study relying on circumstances outside researcher’s control to mimic random assignment

12. 2-12 Conducting a Quasi-Experimental Study  Difference-in-difference quasi-experiments  Instrumental Variables quasi-experiments  Regression-Discontinuity Quasi-Experiments

13. 2-13 Pitfalls of Quasi-Experimental Studies  Assignment to control and treatment groups may not be random  Not applicable to all research questions  Generalization of results to other settings and treatments