Bad Statistics Lead to Misinformation

This article describes the danger of using "bad" statistics leading to the creation of misinformation in the field of highway safety, as reported in a new study on how motor vehicle death rates have been misinterpreted. Death rates per mile traveled, for instance, are widely used to gauge success, but fail to account for externalities such as the fact that where congestion is high, the risk of death is much lower. Too many other factors can influence the outcome, such that it is not a good determinant of the success of countermeasures. Similarly, deaths per registered vehicle or per population can be affected by outside influences, such as the composition of vehicle fleets and demographics. Well-known, competing assertions from the U.S. Centers for Disease Control and Prevention (death rates falling on U.S. roads from the 1970s through the 1990s) and from researcher Leonard Evans (U.S. death rates not falling as rapidly as in other developed countries) don't give full answers because they don't take other factors into account. Similar disputed or incomplete data have been used in the debate of the 55 mph speed limit, the SUNflower report (Sweden, the United Kingdom and the Netherlands), and state-by-state data comparisons. Illustrations include a graph showing the influence of three urbanization factors on death per billion miles traveled 2001-03. Higher population density, higher numbers of urban residents, and higher congestion all contribute to lower fatality rates. Suggestions include fine-tuning evaluations and avoiding broad-brush assertions.

• Record URL:
  http://www.iihs.org/iihs/sr/statusreport/archived/2006
• Availability:
  • Find a library where document is available. Order URL: http://worldcat.org/issn/0018988X
• Publication Date: 2006-4-22

Language

• English

Media Info

• Media Type: Print
• Features: Figures; Illustrations;
• Pagination: pp 1-6
• Serial:
  • Status Report
  • Volume: 41
  • Issue Number: 4
  • Publisher: Insurance Institute for Highway Safety
  • ISSN: 0018-988X
  • Serial URL: https://www.iihs.org/status-report/

Subject/Index Terms

• TRT Terms: Data quality; Fatalities; States; Statistical analysis; Traffic safety; Trend (Statistics); Urbanization; Vehicle design
• Subject Areas: Data and Information Technology; Design; Highways; Safety and Human Factors; I81: Accident Statistics;

Filing Info

• Accession Number: 01024865
• Record Type: Publication
• Source Agency: UC Berkeley Transportation Library
• Files: BTRIS, TRIS
• Created Date: May 31 2006 7:55AM