Federal data highlight state hit-and-run trend

Investigative Reporters and Editors, Inc. The IRE Journal, Jan/Feb 2004 by McCormick, Erin, Cabanatuan, Michael

When police found the body of a San Francisco college student lying crumpled in the street - run down, then abandoned by a hit-and-run driver - it seemed like a recurring nightmare.

In the months preceding, hit-and-run deaths had peppered the pages of the San Francisco Chronicle: a 5-year-old killed by an unlicensed illegal immigrant, a teenage girl facing charges in the death of a first-grader, and numerous other tragedies.

The Chronicle decided to find out whether the drumbeat of deaths was more of a trend than a coincidence.

Using an online query system available from the U.S. Department of Transportation and a slice of data provided by the IRE staff, the Chronicle was able to quickly examine hit-and-run deaths to discover that California was truly facing an epidemic.

The resulting stories revealed that hit-and-run drivers were causing a higher percentage of highway deaths in California than in any other state - and about half of those runaway drivers who were eventually identified did not have valid driver's licenses.

The Chronicle's coverage sparked a debate on the role of illegal immigrants in California's hitand-run crisis - at a time when driver's licenses for illegal immigrants was one of the state's most hotly contended issues.

The stories arc an example of how, by tapping into the some of the data-crunching shortcuts that are increasingly available to journalists, papers can turn routine breaking news into illuminating short-term enterprise reporting.

In this case, the data work was the easy part. Getting public safety officials to respond to a hit-and-run crisis they had never even considered - and flushing out details of years-old traffic fatalities - turned out to be the more time-consuming part of the story.

When editors returned from Memorial Day weekend to hear that 22-year-old college student Srijaya Dalton had been run down while walking home from a late-night graduation party, Chronicle Managing Editor Robert Rosenthal asked: "How often is this happening?"

The paper wanted answers quickly. We went to the California Highway Patrol's Web site and found that the state tracks only the barest statistics on hit-and-runs, an annual count of the total number of accidents. Calls to state officials didn't help much either. One CHP official explained that the agency was so busy dealing with drunken driving and seat belt laws, it hadn't even considered hitand-runs.

So, clearly, if we wanted answers we were going to have to get them ourselves. Fortunately, the federal Transportation Department now makes its entire Fatality Analysis Reporting System available through a Web-based query system. The query system allowed us to quickly answer the first question on our minds: Is this a story worth pursuing?

Each state is required to report the details of all deadly accidents into the federal FARS database, including such variables as exactly where the accident happened, the type of vehicles involved, the age of the drivers and whether drunken driving was involved. Ironically, the federal government publishes California's data months before the state gets around to releasing its accident statistics and the federal information is far more detailed.

Like many of the database query systems that are popping up online, the PARS data available at www.fars.nhtsa.dot.gov/queryReport.cfm?stateid=0 &year=2002, is not exactly user-friendly. The long list of variables and references to things like "univariate tabulations" can be a bit intimidating. But virtually anyone with the patience to spend a few hours plodding through the accompanying help menus and terminology lists can learn it. No database expertise is required, although it helps to know a little about Excel.

We found one of the keys to using this system was to have a succinct question in mind. The query system, with its complex labyrinth of variables and options, is a lousy place for a fishing expedition. It would be difficult, if not impossible, to use it to find the deadliest roadway in your region. But it is great for pinpointing answers to specific statistical questions that come up in news coverage. Our question was simple: How does California compare to other states in the number of hit-and-runs?

For our first query, we choose only one variable from the list of 100 or so available: hit-and-runs. It was no surprise to find that California had the highest number of hit-and-run accidents; it has the highest population and by far the most cars of any state. So, we did another query to get the total number of fatal accidents in each state. We downloaded the two queries into Excel to calculate the percentage of accidents that involved hit-and-run drivers. California was the top state in this measure as well, with only the District of Columbia having a higher percentage of fatalities caused by hit-and-run drivers.

After a half-day's work, we had the nut graph for our story - and an excuse to go on to the next question, which was: Why?

For this we wanted to massage the actual PARS data and see all the possibilities the dozens of variables offered. So, we called IRE and took advantage of one of journalism's best-kept secrets. If you call the IRE and NICAR Database Eibrary with a breaking story and a tinge of adrenalin in your voice, you are likely to get more enthusiastic assistance than you dreamed possible. Within hours, the Database Library staff had cut us a portion of the PARS database that included only hit-and-run accidents in California over an eight-year period. This meant we could skip having to cull through tens of thousands of accidents of no interest to us and, instead, focus immediately on the 2,000 or so records that addressed the question at hand.