Department of Statistical Science
Duke University
presents:
Valerie Williams
National Institute of Statistical Sciences
"Controlling False Discoveries in Educational Research"
Abstract: Results from national studies of educational variables are intended to be useful by informing policymakers and the public about the condition of education. Because the collection of data about educational achievement involves the testing of school children, it is time-consuming and costly; therefore, it is important to make full use of state-of-the-art statistical methodology in the analysis of available data, the reporting of findings, and in the design of future studies.
The problem of multiplicity in the context of large-scale educational assessments will be described, with attention to issues arising in the testing and reporting of differences in student proficiency across comparison variables, years, and jurisdictions. The motivation for the consideration of different kinds of errors will also be discussed, as will be alternative procedures for error control, specifically, techniques for controlling the false discovery rate, or the expected proportion of falsely rejected hypotheses. Examples from the National Assessment of Educational Progress demonstrate the advantages of these techniques over the traditional Bonferroni and sequentially-rejective Bonferroni procedures.
Friday, April 26, 1996
11:45 am - 12:45 pm
130 Sociology/Psychology Building Any questions concerning the seminar may be addressed to Cheryl McGhee @[919] 684-8029, e-mail cheryl@stat.duke.edu, or finger seminar@stat.duke.edu