This lecture considers four increasingly complicated situations involving estimation of models in the presence of mismeasured variables. The focus here is on practical strategies for identification and estimation. We begin by examining the classical Errors-in-Variables model. The second section considers non-classical mismeasurement while still in a linear regression setting: Binary regressors and other departures from the additive measurement error. The third section turns to measurement error in the dependent variable and considers both linear models and binary choice models such a probit and logit. Finally, we will close consider a few cases of non-parametric bounds. The focus will be upon empirical examples, although some econometric theory and identification will be presented (primarily in the first and third section).
The lecture is *free of charge* and will be held in the Large Seminar Room (2N2.4.16) in ISER. There is no registration fee and refreshments and lunch will be provided.
h3. Programme
|09:30|Coffee and Registration|
|10:00|Section I: The Classical Measurement Error Model|
|11:00|Coffee Break|
|11:15|Section II: Leaving the Classical Model: Error in Regressors|
|12:15|Lunch|
|13:15|Section III: Measurement Error in the Dependent Variable|
|14:15|Break|
|14:30|Section IV: Non-Parametric Bounds|
|15:30|End|
Presented by:
Christopher Bollinger
Date & time:
May 20, 2011 8:30 am - May 20, 2011 2:30 pm