This preview shows pages 1–7. 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 Document
Unformatted text preview: Introduction to Econometrics Chapter 9: Assessing Studies Based on Multiple Regression Geo rey Williams gwilliams@econ.rutgers.edu April 23, 2010 Geo rey Williams gwilliams@econ.rutgers.edu Introduction to Econometrics Chapter 9: Assessing Studies Population studied, population of interest We begin with a distinction  when we do a regression, there are two populations we have on our minds: The population studied (in the case of the schools data, the California school districts in 1998) The population of interest (in the case of the schools data, any school system outside of California, or more than a few years before or after 1998) Geo rey Williams gwilliams@econ.rutgers.edu Introduction to Econometrics Chapter 9: Assessing Studies Internal and External Validity We are then considered with two types of validity: Internal validity: the statistical inferences about causal e ects are valid for the population being studied. External validity: the statistical inferences can be generalized from the population being studied and setting studied to other populations and settings of interest, where the setting" refers to the legal, policy, and physical environment and related salient features. Geo rey Williams gwilliams@econ.rutgers.edu Introduction to Econometrics Chapter 9: Assessing Studies Assessing External Validity of a Regression model Here we ask the question of whether the results we get from our model are general or are only applicable to the data that we studied. That is, external validity asks whether we can extend our results from the population being studied to another related population. For example, in the class size example from the text, we studied the e ect of class sizes on student's test performance using data from California can we take the lessons learned from this study and use it to make policy in New Jersey? can we continue to understand the California school districts of 2010 with 1998 data? Geo rey Williams gwilliams@econ.rutgers.edu Introduction to Econometrics Chapter 9: Assessing Studies External Validity The answer to this depends on many things. Some of which are What are the di erences between the population studied and the populations of interest? A study of California, a large diverse state, would be more helpful in understanding the school systems in New Jersey than of Montana or Vermont, say. What are the di erences in settings? A study in 1998, before the passage of No Child Left Behind, would have a very di erent policy setting from a population of interest in 2010. There is no easy rule of thumb, we should be wary of using studies from elsewhere to make predictions about what would happen locally. Geo rey Williams gwilliams@econ.rutgers.edu Introduction to Econometrics Chapter 9: Assessing Studies Assessing the Internal Validity of a Regression Model Internal Validity is de ned as: the statistical inferences about causal e ects are valid for the population being studied....
View Full
Document
 Fall '10
 MacroWilliams,MicroYoshi

Click to edit the document details