STATISTC 697ME - ST-Measurement Error

Institution:

University of Massachusetts Amherst

Subject:

Description:

Measurement error (errors-in-variables) which is ubiquitous, occurs when one or more of the variables of interest in a model cannot be observed exactly. In ecology this is referred to as observation error, while in economics it is called the problem of unobservable variables/use of proxy variables. This is typically due to sampling error, instrument error or a combination of the two and includes misclassification of categorical variables as a special case. Examples abound across epidemiology, ecology, economics, the physical sciences as well as most other disciplines. This course examines the impact of measurement errors on standard statistical analyses which ignore them (so-called "naive" analyses) and describes methods of correcting for measurement error using additional information or data about t he measurement error process (usually arising from replication or validation data). We examine these questions for i) misclassification in estimating one or more proportions and in two-way contingency tables; ii) measurement error in predictors and/or the response in simple and multiple linear regression as well as error in the response in estimating and comparing one or many means; iii) measurement error in nonlinear regression, including binary regression (e.g., logistic or probit) and Poisson type models. The focus is on understanding models and methods and applying them to examples from a variety of disciplines. Computing will use STATA and SAS, the two software packages for which measurement error programs have been developed. Prior experience with SAS or STATA is not required. ST797ME will have more of a theory component. Prerequisites: A background in probability and statistics at the level of STATISTC 515-516 or equivalent, some familiarity with regression analysis including experience with regression models in matrix form (e.g., ST505 or equivalent). Prior exposure to nonlinear and logistic regression is not essential, we will review the usual methods there when we get to these topics. ST797ME requires ST607-608 and ST705 (or current enrollment) or equivalent.

Credits:

3.00

Credit Hours:

Prerequisites:

Corequisites:

Exclusions:

Level:

Instructional Type:

Lecture

Notes:

Additional Information:

Historical Version(s):

Institution Website:

www.umass.edu

Phone Number:

(413) 545-0111

Regional Accreditation:

New England Association of Schools and Colleges

Calendar System:

Semester

The Course Profile information is provided and updated by third parties including the respective institutions. While the institutions are able to update their information at any time, the information is not independently validated, and no party associated with this website can accept responsibility for its accuracy.