Uncovering bad bridges at risk in mudslides

A journalist uses data analysis to investigate poorly maintained bridges in Ventura County, California after mudslides kill ten people, discovering that 99 of 485 county bridges are deficient or obsolete and four of those damaged in the disaster had poor condition ratings.

FIRST VENTURE

By Dani Dodge, Ventura County Star

I had just jumped onto the hotel treadmill in Columbia, Mo., for a quick morning workout before computer-assisted reporting boot camp started when a picture of my hometown fire chief flashed on the television news.

Record rains had triggered massive mudslides back home in Ventura County, Calif. Ten people were dead. And I was missing the big story. My editors wanted me to finish the week-long IRE and NICAR Boot Camp, so I did what any journalist would have done and figured out an angle that I could begin to cover from more than 1,500 miles away.

My story: Money for bridge maintenance was going down in the state and county as the bridges were aging beyond easy repair. The consequence: An extra surge of water could wash them away.

It had taken nearly a year to persuade my editors to allow me to attend the January 2005 Boot Camp. We were into the third day of training when heavy rains flooded Ventura County, causing mudslides and mayhem. So while other boot campers researched federal Small Business Administration loan records or calculated AIDS rates for women, I used Microsoft Access to delve into the data for bridges and dams. If hillsides were coming down in the rain, bridges might be also. I bought the National Bridge Inventory database from the IRE and NICAR Database Library before I returned to Ventura County.

After I got back, I was quickly thrown into the daily coverage to relieve reporters who had been working 18-hour days. But I also continued to probe the data. The flooding had destroyed two bridges in Ventura County and damaged others beyond repair. Had those bridges been maintained well? What about the other bridges in the county? I looked in the data for answers. I ran queries for the state, neighboring counties and Ventura County. In the queries, I asked about how many bridges were deficient and obsolete. I compared the rates of disrepair. I looked at the condition ratings of bridges.

The data showed that 99 of Ventura County's 485 bridges were deficient or obsolete. Four of the bad bridges were the ones that had been washed out or severely damaged. I discovered that within Ventura County, city- and state-maintained bridges had a much lower percentage of bad bridges than the county-maintained bridges. When I called county officials, though, they said their bridges were completely safe despite the poor condition ratings and the rains: "Just a little damage." So I went after the paper inspection records. At first, the county public works officials balked: Someone would have to pull them for me, and they were too stretched, they said.

I threatened to file a public records request, but at the same interview I also asked to see the area where the records were stored. Once I had seen how easy they were to access, the county official realized he had no excuse for not letting me go through them myself. From there, the fun began.

In their reports, state inspectors wrote that several county bridges were "unstable" and needed immediate attention. The date on the inspections: 2001. But, according to the reports and interviews, no work had yet been done. County officials could no longer continue to deny problems existed. Looking at some of those bridges and several others in the county that inspectors had rated as structurally deficient, I found one bridge with rotted timbers that I could poke a stick through. I also watched heavy trucks rumble over another bridge with a visibly undermined foundation - water had washed away the river bottom under the concrete pier that was supposed to hold it up.

I spent two and a half weeks on the story. Only a small portion of that was spent working with the data, but I went back to it often. Several times, I found something in the paper records, and then found a corresponding trend in the data. I realized only after looking at the reports the importance of words like scour, which means erosion.

"Scour critical" became an important benchmark in my story. I brushed up on the language of engineers by reading Web sites such as the American Association of State Highway and Transportation Officials at www.transportation.org/aashto/home.nsf/FrontPage.

Then I hit a snag: a discrepancy in my numbers. I was down to my final cross-checking when I decided to match the numbers I had from my analysis with the numbers on the Federal Highway Administration's Web site. Mine showed nearly one-third more bridges in California than the Web site. I felt that same cold rush of panic I had experienced seeing the Ventura County fire chief on the news.

First, I called a Federal Highway Administration public relations officer. He said the numbers on the Web site were correct. He didn't know how I could have come up with more bridges. I called Jeff Porter, the Database Library director for IRE and NICAR. He said it hadn't been a problem before, and suggested looking at several things, including whether I had double-counted bridges. He said that some bridges are counted twice because the administration considers the bridge underneath, or underbridge, as a separate bridge. I didn't think that was the problem because each of the bridges I counted had a unique ID number. It didn't make sense to me that any of them would be underbridges.

So, I searched further. I e-mailed a computer analyst at the state transportation department. He didn't respond. I called the number on the fact sheet for the Federal Highway Administration. A woman told me she couldn't talk to reporters.

The world around me started to get black. So I begged. I got teary. I explained I wouldn't quote her, but please don't let me put bad information into the newspaper. She quickly told me several items to check. One was underbridges just as Porter had said. When I explained I had already tried to check that by making sure I included only bridges with unique numbers in my count, she explained that the underbridges also had their own ID numbers. Five minutes later, the problem was solved. (If I had researched Porter's suggestion further, I would have saved myself an hour of panic.)

The story, "County's aging bridges at the breaking point," ran a month after I graduated from boot camp. Sidebars, graphics and a Web video presentation I had shot accompanied it. It can be seen at www.VenturaCountyStar.com.

The county official who hadn't wanted to give me the records was impressed with the report. He asked me if I expected to win a Pulitzer Prize with the story. Yeah, right after I hit the lottery. Nevertheless, I did learn some lessons:

1. Practice is key. To do computer-assisted reporting, I found I have to rewire my brain. The key to that is practice. I try to do a little bit each week. 2. Read first. Analyze later. On the bridge data from the Database Library, there is a readme file and a guide to the fields in the data. Though I thought I knew what I was looking for, I found there was much more to know and that the data had much more to give. I realize I would have saved a lot of time if I had read all the materials that came with the data before diving into the numbers. In my hurry, I did it the other way around and wasted time. 3. Log queries. Keep records. When I started queries on the data, it seemed so simple to do and then so tedious to write. I realized later I should have heeded the advice of the boot camp instructors and written down everything. Then, when I needed to re-create the queries with different modifications, I could have done it easily. Logging queries also seems to give editors a much better sense of security. In addition, I learned it can be very difficult to re-create queries two weeks later under deadline pressure with a stressed-out editor looking over your shoulder. From now on, I'm keeping a notebook by the computer to log the queries. 4. Check your results against other sources. If the numbers are good, they will stand up. If I hadn't checked the California Department of Transportation Web site, and then the Federal Highway Administration Web site, I never would have known how far off I was. I would have risked a massive correction. I would have had fewer chances to do computer-assisted reporting in the future. 5. Give yourself time. The exercises learned in boot camp can be much easier to do than grabbing data in real time. I found it took me much longer to do CAR on my own. And getting the numbers is just the starting point.

Contact Dani Dodge by e-mail at Dani@DaniDodge.com.