{
  "id": "1999-januaryfebruary-crash-course-816",
  "slug": "1999-januaryfebruary-crash-course-816",
  "url": "https://dwillis.github.io/uplink/article/1999-januaryfebruary-crash-course-816.html",
  "year": 1999,
  "month": "January/February",
  "monthNum": 1,
  "headline": "Crash course (#816)",
  "kicker": "",
  "author_name": "Justin Mayo",
  "author_title": "Reporter",
  "affiliation": "The Seattle Times",
  "summary": "A guide for computer-assisted reporters on using accident and transportation safety data from federal and local sources, emphasizing the importance of data normalization and caution with online databases.",
  "keywords": [
    "accidents",
    "databases",
    "transportation"
  ],
  "topics": [
    "Investigative Methods",
    "Public Safety",
    "Data & Databases"
  ],
  "technologies": [],
  "full_text": "Crash course (#816)\nBy Justin Mayo\nThe Seattle Times\n\nWhen things collide, one of the results is data. Generally, the destruction of property or human injury generates huge amounts of paperwork, which eventually ends up in a database. This grim reality is good news for CAR reporters because of the abundance of accidents - and therefore accident data. Additionally, transportation safety is an issue that will always be important to the audience and one that can clearly be enhanced through computer-assisted reporting.\n\nThree things to keep in mind:\n\nFirst, although many databases are constructed or disseminated at the federal level, remember that you can also tap local agencies - especially for automobile accidents.\n\nMost of the time the state will be more current and complete. However, local agencies might be reluctant to give out the entire database without redacting some information. So be prepared to know what you really need to get out of the database and be ready to fight for it. A good strategy is to first request paper records and the record layout of the database. That way you can see what's collected at the scene of the accident and what information is available in the database.\n\nIf you're going to write about individual accidents, always ask for the paper records.\n\nSecond, when trying to make comparisons, you'll need to normalize the data. Usually, raw numbers will not be enough. Normalizing data will depend on the nature of your story. For example, if you're looking at fatality rates for the elderly, you'll have to get statistics based on the number of miles driven by age of the driver. On the other hand, if you want to look at the type of vehicles involved in rollover accidents, the number of cars on the road by type would be necessary.\n\nThird, beware of online data sources. Many of the federal accident databases are available online, particularly aviation data. These sites can be a great resource for deadline searches, but be warned that some are incomplete. It's best to get the entire dataset directly from the agency (or NICAR) if you plan to use a particular database on a regular basis.\n\nNOTE: In the remainder of his handout, Mayo discusses available databases organized under the topic headings of air, roads, rail and agencies, the Web and/or NICAR.\n\njustin Mayo can be reached by e-mail at jmayo@seattletimes.com",
  "provenance": {
    "extracted_by": "anthropic/claude-sonnet-4-6",
    "extracted_at": "2026-07-05",
    "verified": true
  },
  "similar": [
    {
      "slug": "1999-januaryfebruary-planes-trains",
      "score": 0.457
    },
    {
      "slug": "1999-januaryfebruary-transportation-sites",
      "score": 0.364
    },
    {
      "slug": "1999-januaryfebruary-must-have-resource-677",
      "score": 0.317
    },
    {
      "slug": "2002-november-december-bits-and-bytes",
      "score": 0.298
    },
    {
      "slug": "2006-september-october-more-tips-for-reporting-on-trucking-accidents",
      "score": 0.26
    }
  ],
  "html_url": "https://dwillis.github.io/uplink/article/1999-januaryfebruary-crash-course-816.html",
  "json_url": "https://dwillis.github.io/uplink/article/1999-januaryfebruary-crash-course-816.json"
}