This preview shows pages 1–8. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Unformatted text preview: 91Assessing Studies Based on Multiple Regression (Stock & Watson Chapter 9) Is there a systematic way to assess regression studies?Multiple regression has some key virtues: It provides an estimate of the effect on Yof arbitrary changes in X, X. Each of the jcoefficients we estimated gave us the partial effect of that X variable holding the other ones fixed. E.g., the coefficient on Education was the additional wage earned for each additional year of schooling, EducWage. 92There scenarios in which OLS might yield a biasedestimator of the true causaleffect, so that inferences are not valid. That's what this chapter focuses on. 93sal A Framework for Assessing Statistical Studies:Internal and External Validity (Section 1) Internal validity: the statistical inferences about caueffects are valid for the population being studied. I.e., if the OLS assumptions hold, then we are confident that our estimates based on a sample can be used to make inferences about the population the sample came from. The primary concern is something will cause the first OLS assumption E(uX's)=0 to fail. 94be A Framework for Assessing Statistical Studies:Internal and External Validity External validity: the statistical inferences can generalized from the population and setting studied to other populations and settings.I.e., if the institutional setting of the population that our sample was based on is similar to the institutional setting of another population, we can use our estimates to make inferences about the other population. 95Threats to External Validity of Multiple Regression Studies There are two issues to keep in mind for external validity Differences in populations oI.e., is the population that the sample is from different than the population you want to make inferences about? Differences in settings oSuppose there very little difference in population, now we must ask ourselves if the institutional setting (laws, infrastructure, etc.) of the population the sample is from is different than the population you want to make inferences about. 96Threats to External Validity of Multiple Regression Studies How far can we generalize our wage/education regression using U.S. data from 2007 on headofhouseholds between ages 1865? Differences in populations oCan we use our analysis to make inferences about retirees (people over age 65)? (no) oCan we use our analysis to make inferences about Germany? (okay, except for "settings" below) Differences in settings oSuppose Germany's population looks about the same as the U.S., what can we say about the institutional settings of the U.S. versus Germany? (probably different) 97Threats to Internal Validity of Multiple Regression Analysis (Section 2) Internal validity: the statistical inferences about causal effects are valid for the population being studied. I.e., we need our regression estimates to be unbiased and consistent....
View
Full
Document
This note was uploaded on 09/19/2009 for the course ECON 656820 taught by Professor Megerdichian during the Summer '09 term at UCSD.
 Summer '09
 Megerdichian
 Econometrics

Click to edit the document details