PILOT: Tracking local campaign cash
Building your own database reveals more than just who gave what to whom.

By David Gulliver
The Virginian-Pilot

Did we know this guy got more than half his contributions from developers?

That’s what I asked reporting partner Bob McCabe. It was a Thursday afternoon, I think. We were looking at an Access query. But that’s getting way ahead of the story – four days ahead. Virginia has elections when no one else does. In November, we elected a governor. In May, we were electing city councils and school boards. We cover races in five cities – 77 candidates.

For two previous council elections, before I joined The Virginian-Pilot, reporters had compiled a campaign contributions database. Candidates turn in paper forms on a Monday, eight days before the election. We pound the information into a database and turned out a quick story.

Those stories were starting to look stale. We’d done the “big donors” story, the donations-by-zip-code thing. I polled the city reporters. Jason Skog wanted to look at donations by interest groups. The team agreed. Our existing database captured donor name and address, employer, occupation, date, amount and type of contribution (in-kind or cash). That’s all you need for basic stories, but to track interest groups we needed more.

Tracking interest groups

First, think about what you want to see. You want to assign donors, based on their jobs, to a few definable groups. Don’t do this from scratch – steal ideas from someone who has done the work already. The Center for Responsive Politics and the Virginia Public Access Project both break down campaign money by interest group.

On the VPAP site, click the pull-down menu for “occupation” and you get 17 categories. Some aren’t relevant for local races – for example, defense and energy companies don’t lobby city hall very often. We brainstormed and settled on 10 groups, including construction/real estate, environmentalists, government employees, school employees, tourism industry, political parties or politicians. Most panned out; some, like civic groups, didn’t.

I wrote a short guide, giving examples of which companies and people belonged in which groups. You can’t anticipate everything, but at least everyone starts on the same page. Then I asked the team how data input went last time. I took their suggestions and tinkered with the database’s front-end form.

I used to scoff at front-ends, but they’re important. Most people aren’t database-proficient, but they’ve all seen Internet forms. So make the database look like a Web page by adding a front end. The Baltimore Sun’s Mike Himowitz has a great tutorial on making a search form at www.himowitz.net/html/front_ends.html. Or contact me and I’ll forward a copy of ours for your reverse-engineering pleasure.

Some design tips

• Consider your source material: We put the data fields in the same order as the paper forms. That way, you type as you read, no fumbling around.

• Anticipate the data. Study the paper reports: You’ll see examples of odd data like two-name donors (“John and Jane Smith”) and two-address donors (both P.O. box and street address) that you need to handle. You’ll also get ideas of recurring themes. In Virginia, companies can donate to candidates. To easily compare business and individual donations, we added a flag field (type “Y” if a donor is a company).

• Minimize typing and mistakes: We used Access’ “combo box” feature to create pull-down lists for candidate and for interest group. That way, bad typists don’t mangle the name data, but good typists can enter the first letters and the rest auto-fills. You also can use standard abbreviations – for example, we used “VB” for Virginia Beach. Run a cleanup routine later.

Finally, news researcher Diana Diehl and I test drove it. Then I went on a long-planned vacation – returning the Monday morning the reports were due. I was there, but the data wasn’t. Most candidates didn’t file until 5 p.m., or the next day. It took until noon Tuesday to round up all 77 reports. When we did, we found a surprise – some amazingly thick filings. They’d been busier than in past years. (In the end, we had 2,270 contributions, 60 percent more than last time.)

We scrambled to borrow some reporters and editorial assistants. Most input into separate copies of the database. That was by necessity – some bureaus are off the network – and by design. If a file got corrupted, we wouldn’t lose everything. (One reporter did accidentally delete records, but we repaired the damage.) I appended the files later.

Typing ran well into Thursday. I ran preliminary results so we could start reporting, then did cleanup and consistency checks into Friday. Leave time for those checks. One drawback of manpower is that people interpret and code data differently. I ran queries to test that – for example, did Bob code John Smith as a attorney while Jason called him a developer? Surprisingly, there was little variation. About 40 donors needed standardizing.

Friday morning, I hacked out a first draft and ran final numbers while Jason and Bob added sections on key races. The data had pointed us to a handful. A day before, Bob and I had broken down the sprawling suburb of Chesapeake, where some residents want open space, others more growth and a bigger tax base.

W.P. “Pete” Burkhimer looked like a shoe-in, the top fund-raiser among nine people running for five seats. The database showed he raised 60 percent of his money from the construction group. Burkhimer, a consultant who represents developers before city boards, told us that made sense. He said he was running because he wanted to put out a welcome mat for business.

Land use is the most local of issues, so you expect developers to be politically active. And they were – they gave more than one-fourth of all the money raised. But they supported Burkhimer far more than other candidates. Developers’ dominance, with Burkhimer as a lead example, was the theme of our Sunday 1A story. Voters went to the polls Tuesday. By that night, he had placed sixth in the five-seat race, missing out by 697 votes.

David Gulliver is database projects reporter for The Virginian-Pilot. He can be reached at DGulliver@pilotonline.com.

Return to Tracker.