How I wrote it: “In Buffalo’s ‘digital deserts,’ more than half of households lack Internet”

A story I wrote for last Sunday’s paper relied on a pretty shocking statistics: In a neighborhood only minutes from the Buffalo News, fewer than two in five households have home Internet service.

Poynter was kind enough to include the story in its morning media newsletter Tuesday, where Tom Jones called it “the type of important story that news organizations in every major U.S. market could and should do” (yay!). Since then, I’ve gotten several emails from local reporters looking to unearth this data for their city.

Fortunately, it’s not at all hard to do. You just have to know where to look. So below, find my abbreviated, three-step guide to reporting on the digital divide near you.

Step 1: Find neighborhood-level Internet data

This story pulled from the Census Bureau’s 2013-2017 American Community Survey, the first to report Internet access rates down to the tract level. A Census tract is the bureau’s second-smallest geographical unit, consisting of 1,200 to 8,000 people.

There are 297 Census tracts in Erie and Niagara counties. I downloaded data for all of them using You can do the same by advance-searching your local city/county, selecting “Census tracts” as the geography filter, and choosing “B28011 — Internet Subscriptions in Household — 2017 ACS 5-year estimates” as your document title.

Step 2: Match Internet data and demographics

Access rates are half the story, though. I also wanted to know why some neighborhoods were better-connected than others. Studies have pointed to a range of reasons some people don’t have Internet, from age and disability to poverty and issues with rural access.

So to help narrow that field, I also added to my growing Excel sheet tract-level ACS data on demographic factors including age, poverty levels, English fluency and disability status. I also coded each tract as “urban” or “rural” based on Census categorizations.

When Excel’s CORREL function revealed a strong correlation between access, poverty and educational attainment, I knew I had enough data to start pursuing an angle. I contacted public schools and community groups in the worst-connected urban neighborhoods to find out if access among low-income folks was on their radar.

Step 3: … But wait, there’s another spreadsheet!

As I began interviewing, I heard two complaints over and over again: that a single Internet provider held a “monopoly” in Buffalo, and that the best and fastest Internet speeds weren’t available to city residents.

To verify this, I turned to the Federal Communications Commission’s broadband deployment records, also called Form 477. Internet companies are required to file twice-yearly reports with the FCC declaring where they offer service and at what speeds.

By sampling this data for representative Census tracts in Buffalo and its suburbs, I was able to confirm that (a) city residents essentially have only one terrestrial broadband option and (b) that option runs more slowly in Buffalo than it does in many suburbs. (Fwiw, I think this is a prime area for follow-up once I figure out a way to stop the full file from crashing my computer.) You can find a guide to Form 477 here.

That’s all the heavy lifting, though! After that, it was all phone calls and interviews and frantic consultations with a News developer who helpfully explained to me how the actual mechanics of the Internet work. And two weeks later, I had a story that — as one source told me post-publication — may “really move the conversation forward.”

If you have further questions about this data, I’m happy to answer them. And if you end up doing a story in your market on the digital divide, please holler! I’d love to read it.

(As an aside, I plan to do regular blog posts of this sort. I have NO IDEA if anyone will read them, but I’m on a mission to show more of my work.)