|
Activity Number:
|
269
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 4, 2009 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Risk Analysis
|
| Abstract - #303461 |
|
Title:
|
A Critical Statistical Review of Some Transportation Safety Studies
|
|
Author(s):
|
Norma F. Hubele*+
|
|
Companies:
|
Arizona State University
|
|
Address:
|
3702 E. Dakota Drive, Phoenix, AZ, 85044,
|
|
Keywords:
|
National Automotive Sampling System ; sampling errors
|
|
Abstract:
|
The National Automotive Sampling System is used by transportation researchers to assess crash frequencies, risk and policy impacts at the national level. The three stage probability sample survey developed in the late 1970s by the National Highway Traffic Safety Administration collects over 500 data elements on about 5,000 crashes annually. Originally designed to assess big policy issues, some researchers are now drilling into the databases to levels that result in extremely small sample sizes with very large sampling errors. Their studies are being published without critical statistical review. For example, in a child safety study all of the results are incorrect; 60% of the tabulated results are based on subsamples with less than 10 observations. We use a wide assortment of examples from archived literature to demonstrate the severity of these poor analyses and mistaken conclusions.
|
- The address information is for the authors that have a + after their name.
- Authors who are presenting talks have a * after their name.
Back to the full JSM 2009 program |
