My fishing trip seeking data on child abuse in Alabama did not go as planned
My big reporting project had become another chapter in a story I’ve come to know well in my decade as a reporter in Alabama: Our state doesn’t know things because it doesn’t want to know. Or at the very least, it doesn’t want to spend the money to find out.
I filed a records request with the Alabama Department of Human Resources (DHR) in October 2018 for several years of data on child abuse deaths that the state is federally required to provide under the Child Abuse Prevention and Treatment Act. I hoped to examine the data for patterns and uncover stories hiding inside.
Turns out, that would take a while. Countless phone calls and emails later, I finally received the data in March 2019.
Some information was protected by state and federal law, including Alabama’s notoriously vague open records act. Some information I’d requested could only be found on paper copies of hundreds or thousands of undigitized reports, presumably sitting in filing cabinets somewhere.
The goal for my Center for Health Journalism Data Fellowship project was to examine child abuse deaths and near-deaths in Alabama to better understand how and why children die from abuse here. I hoped the data might show how poverty contributes to the maltreatment of children in communities across our state, or how victims interacted with state agencies before their deaths.
The first step was to build the most complete and accurate database of possible child abuse and neglect fatalities in Alabama from 2011-2015, and to show our readers what I found. I believe I succeeded there.
But when it came to finding truly surprising patterns or stories within the data, I hit wall after wall. Sometimes the data just didn’t show the surprises I’d hoped to find. At other times, promising leads became difficult to report without more data points that I didn’t have. For example, detailed information about prior state involvement with the victims wasn’t part of the data that DHR was able to provide.
In the end I was able to report and publish a story about Alabama’s physical child abuse problem, generating reader engagement and drawing attention to groups in the state that are working to fix it. We showed readers what abuse looks like in Alabama, from the age and race of the victims to the amount of time the abusers spend in jail.
But in other ways, the project lacked the bombshells I’d been hoping to find.
When a data fishing expedition doesn’t answer your questions, how do you make the time and effort worthwhile for readers?
1) Come clean.
Give yourself the best shot at finding a pattern by making sure your data is organized correctly. During the fellowship’s week of training at the 2018 Data Fellowship, I learned how to clean up the byzantine spreadsheets the state agencies often send, including the one I finally received from DHR. I also learned about the kinds of additional data points I might need to help me find potential stories.
A quick example: In the original DHR dataset I received, the cause of death listed for each child wasn’t written in a way that would let me easily parse, for example, how many general blunt-force trauma deaths there were. Instead, one death might be listed as “head trauma,” and another might say “brain trauma” and yet another might say “skull fracture” or “traumatic brain injury.” All were similar situations, but they were impossible to filter for my needs.
So, I added a column where I slotted each death into a larger category: blunt force trauma, shooting, hot car, unsafe sleep, etc. That made it much easier for me to see how children were being killed. And that ultimately led me to focus my story on physical abuse — a problem that’s worse in Alabama than in much of the nation.
2) News reports and court records are your partners.
The data that Alabama DHR made available on each child abuse death wasn’t enough to tell me what I wanted to know. The spreadsheet on child abuse deaths I received had 13 columns. I began adding columns as I did more research, eventually ending up with 36 columns of data or partial data for each death.
That research was a time-consuming necessity. Alabama DHR isn’t responsible for keeping keep track of things like whether abusers are prosecuted, how much jail time they get if convicted, or whether the child died in a rural or urban setting.
As a result, I spent weeks, off and on, reading news reports and sifting through court records to fill in the gaps in what I knew about the circumstances surrounding each death.
News reports helped me identify some inaccuracies in the DHR data and filled in details that were missing. They often provided the narratives I needed to find stories worth telling. Court records told me what happened after DHR’s involvement ceased — after the child’s death — and whether anyone was ever held accountable.
3) Nothing takes the place of good interviews.
Data can get overwhelming. It took interviews with experts to help me understand what I was looking at, and to see where the stories might be.
During one interview, an offhand comment by the director of a child advocacy organization gave me the first inkling of an idea that would eventually become the basis of my story. She mentioned that national data showed Alabama has a rate of physical child abuse that’s three times the national average. Once I started looking at the causes of death for the victims in my dataset, I found that most of the deaths weren’t from things like drug exposure or unsafe sleep or neglect; they were from physical abuse.
From there, I found the story of one particular little girl, beaten by her mother’s boyfriend, who exemplified a pattern I saw repeated over and over across the state: catastrophic physical abuse, a young victim, a male perpetrator.
4) Sometimes a fishing expedition doesn’t yield the fish you want.
I began this fellowship with the idea that a certain dataset — the information Alabama DHR is federally-required to report about deaths resulting from child abuse or neglect — would tell me all kinds of things about how and where children die from abuse.
I soon learned that there were limitations to what I could know or infer from this particular data set, for a variety of small and frustrating reasons. It was maddening, at times. My mentor, Paul Overberg, had to talk me down more than once.
But even though I wasn’t able to find all the connections I’d hoped to find, I was able to show our readers what the data tells us about child abuse victims and their abusers. It helped me craft a story that highlighted an overlooked problem in Alabama, and to tell it in a compelling way.
I have a rich data set that I fully intend to mine for future stories throughout the rest of the year and beyond. Being a 2018 Data Fellow taught me how to fish, and it’s a skillset I’ll use again and again.