How a regional paper investigated the state lottery and smaller organizations can follow suit
In September, The McClatchy-owned Charlotte Observer, the paper of record in North and South Carolina, published “Against All Odds,” a report on potential fraud in the state lottery. The investigation resulted in three articles, an animated short and two video interviews. The Observer’s Gavin Off, who shared a byline with Adam Bell on the articles, explained how they reported and presented the story. Off has been the data reporter in Charlotte since 2011.
The genesis was in a time peg, and other newspapers’ stories
The Observer was planning a deep dive on the North Carolina Education Lottery for its 10-year anniversary. After originally planning to examine the lottery’s growth throughout the state, Off said, the reporters took an idea from investigations by the Boston Globe and Palm Beach Post, which found some lottery winners who won so often they defied odds. Some clerks have been found to
steal their customers’ winning tickets, and some people volunteer to cash winners’ tickets for a cut of the money, so the real winner can avoid tax deductions for child support.
“Sure enough, we came across the same findings that they did,” he said.
In early 2015, Off requested from the state lottery all claims of more than $600, which the IRS requires states to disclose, dating back to 2006. He said the agency fulfilled his request within about a month, more helpful than other state departments.
The response came in a text file, with about 189,000 entries, small enough that they could handle it with Microsoft Access, rather than needing SQL Server. But, he said, there was a catch. The dataset didn’t include whether the lottery winner was also a retailer. They needed to know that to identify potential store clerks who were keeping customers’ winning tickets for themselves.
“So we went back and said, ‘Hey, we’re missing a field,’” he said.
After another month, they had the data they wanted.
After the data arrives, the real work starts
Determining whether entries were repeat winners required clean name and address data. Their dataset included misspellings they had to find and correct.
“Charlotte might be spelled with three ‘t’s, compared to two. A retailer address might have a ‘North’ written out, instead of an ‘N,’” Off said.
Issues like inconsistencies in winners using their middle names meant they had to compare the locations of the shops they won at to reasonably conclude they were the same person.
After they were done cleaning, Off said, they pulled disciplinary records and investigative reports and interview experts. They went to stores, met with lottery officials in Raleigh and sat on conference calls with statisticians. The project includes hidden camera video of an undercover agent giving a store clerk a fake winning ticket, which the clerk says is actually a losing ticket but then keeps for himself.
For the story’s online presentation, Off used Tableau to quickly build graphics that were easy to plug into the page. He said with tight resources, he expects to keep using the program for easy visualizations.
Having time helps, but smaller papers could scale down:
Off is on the Observer’s investigations team, and says he spends about 50 percent of his time on long-term projects like this one. Though he was juggling a couple projects at once, his position allowed him a lot of freedom to wait for documents and data. He said Bell was given about three days a week to focus specifically on the lottery project.
He gave two tips to smaller news organizations who want to investigate their local lottery:
“One, you can always limit the time frame,” he said. The Observer looked at 10 years of data, but he was confident red flags would pop up in five years’ worth.
Second, he said, reporters can limit requests to lottery claims in their county.
Off said busy newsrooms can “hit it when [they] can,” doing investigations in small bits over time when they’re not focusing on daily coverage. “Run some queries now, come back into your reporting later.”