|
Activity Number:
|
593
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Thursday, August 6, 2009 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
| Abstract - #305794 |
|
Title:
|
What Scientists Really Want to Know
|
|
Author(s):
|
Eric Vance*+ and Scotland Leman and Leanna House
|
|
Companies:
|
Virginia Polytechnic Institute and State University and Virginia Polytechnic Institute and State University and Virginia Polytechnic Institute and State University
|
|
Address:
|
212 Hutcheson Hall, Blacksburg, VA, 24061,
|
|
Keywords:
|
bayesian methods ; hypothesis testing ; scientific collaboration
|
|
Abstract:
|
Scientists perform experiments and collect data in order to understand some aspect of the natural world. Scientists want to know what is the state of nature, and how does this relate to the conventional wisdom of their field? However, standard Classical hypothesis testing procedures can tell a scientist whether they should reject a null hypothesis as false, but what if the hypothesis and modeling assumptions, though false, describe reality closely? We claim that estimating relationships between the data and sets of parameters is vital for testing hypotheses. We make the judgment that only Bayesian methods lend themselves to answering the questions that scientists really want to know.
|
