i often talk about data journalism as a creative problem solving tool. i want to talk about a story i did seven years ago that i have never really delved into that i think is a good example of how to get data for a story in a way that i think is unique.

as part of a series on the district's tax lien sales process, we kept hearing that folks who were losing their homes after not paying a property tax bill were not receiving notices that their bills were unpaid or the actual bill for the taxes in the first place.

we could have reported these examples and just been done with it. they were egregious and could have easily carried a story by themselves. but the whole series to that point had been grounded in data and we didn't want to depart from that.

so what did we do? we teamed up with the wonderful folks on our polling desk to send postcards to everyone who had their liens sold the year we published (2013). the postcards asked about the process and such and asked them if they'd like to become sources. but that wasn't all.

the real purpose of the postcards was to determine if the dc tax office had good addresses for the folks whose liens were being sold. to compile this data, we tracked bounce-backs for the postcards. so we now knew which addresses had not received notices of the tax sale.

the result was that we had found that more than one fifth of the liens offered at auction in 2013 had notices sent to addresses that did not exist. here's that story. https://www.washingtonpost.com/sf/investigative/2013/10/11/left-in-the-dark/

all of this is to say that we need to think outside the box when it comes to data for stories. if you want to do a story about bad addresses in a government database, why not test that theory by sending your own mail to those addresses?

that's what i've got. if you haven't, please go back and read the whole series. it might be the best thing i've ever worked on as a journalist. and i don't say that lightly. https://www.washingtonpost.com/sf/investigative/collection/homes-for-the-taking/