Memo to Hyperlocal Editors: You Too Can Be a Nate Silver

The name of election prognosticator Nate Silver is very big on Google Search. It’s twice as popular as Benghazi and 25 times more popular than Twinkies. Not bad for a guy whose FiveThirtyEight blog in the New York Times occasionally refers to the “sum-of-squares formula” or “regression analysis” when making a point about polling reliability.

Silver has turned polling and election data into journalistic gold. For weeks before the Nov. 6 vote, he maintained that overall polling numbers showed President Barack Obama would be re-elected, even when much of the media was fixated on GOP opponent Mitt Romney’s “momentum” and supporters’ “enthusiasm.” On Election Day, when he forecast that Obama had a 90.9% chance of winning, Silver captured 27% of the Times’ 30 million total page views. As Greg Sargent wrote in his Plum Line column in the Washington Post, millions of average election watchers trying “to navigate through conflicting numbers and speculation” found Silver’s common-sense directionals a better compass than the “gut” feelings of media pundits. Imagine, a blog consisting of a staff of two – Silver and his assistant – drawing more than a quarter of the traffic that’s produced by an editorial army of more than a thousand. As for engagement, each of Silver’s posts draws hundreds of comments — plus he’s been a guest on Jon Stewart’s “Daily Show” and “The Colbert Report,” and has been woven into Jay Leno’s, David Letterman’s, Jimmy Fallon’s, and Jimmy Kimmel’s monologues. And then there’s Silver’s 416,611 followers on Twitter.

As I watched Silver achieve such success analyzing numbers, I asked myself: Can hyperlocal news sites learn anything from Silver and FiveThirtyEight? The question is pertinent because numbers of all kind — far beyond those from polling and elections — are gushing from local governments, but not many of them are showing up in meaningful ways on community sites. I’m talking about numbers on school performance, health and wellness, public safety, and all the other categories that define what makes a community tick.

The best way to answer the question, I decided, was to try to adapt Silver’s methodology to a specific hyperlocal issue — analyzing the numbers myself, though I’m certainly no math expert. I’m the editorial director of the Local America team that’s developing a website to rate communities across a broad spectrum of livability factors. We’re building a demo to take to potential licensees. Featured in the demo is the community of Historic Anacostia in Washington, D.C., a 97% black community that was predominantly white prior to the 1960s and which is coming back from decades of decline, crime and civic inattention.

One issue in Historic Anacostia that stands out is the poor performance of its Anacostia High School.  It’s easy to write a story about the school’s low reading and math test scores and low graduation rates. But Local America doesn’t want to rehash this old story. What we want to know (and report to the community) is: After four years of K-12 reform in Washington’s city-run school system, is anything new and promising happening at Anacostia HS? To find out, I borrowed some of the methodology Silver uses, and explains so well, in FiveThirtyEight. Looking at eight variables and trend lines, I charted performance at Anacostia High and two other high schools – one also serving a mostly poor and black neighborhood and the other a less poor, more racially mixed neighborhood (below):

The chart is an eye-opener. Anacostia High is clearly underperforming in reading and math proficiency, but the trend line for both is up, and also for advanced-placement courses. But there are three danger signals: 45% of students are regularly truant, only 30% complete 9th grade the first time and only 42% graduate — numbers far below Wilson.

What looks clear from the numbers and their trends is that Anacostia High and Ballou High have two student bodies – the visible one that actually goes to school and the mostly invisible one that is regularly truant.  So why aren’t the two student bodies rated separately?

There are more implications: Why isn’t there an all-out program to reduce truancy at Anacostia and Ballou? Shouldn’t the community be mobilizing volunteers who can help make sure truant students go to school — especially those who don’t go because they have to take care of other family members, don’t have proper clothing and are fearful of encountering violence?

So, this one Silver-style analysis of numbers has the potential to produces a clutch of news stories that could stay alive for weeks, and which promise to draw the community into an action-oriented conversation.

The Anacostia HS truancy story I am putting together is, I discovered this week, what a new NPR study calls a “news explainer” — one of the nine types of local stories that draw the most engagement on Facebook, the study found. If these stories work for NPR, surely they can work for hyperlocal news sites, which have been getting increasing access to tons of data that provide clues to how well public services — like K-12 education — work.

Publishing these kinds of news explainers and the other engagement-producing stories won’t get any hyperlocal honcho on Jon Stewart or Jimmy Kimmel. But engagement is what increasingly selective advertisers say they’re looking for in the local digital space. And that’s the prize that makes hyperlocal journalism a business.

Tom Grubisich authors The New News column for Street Fight. He is editorial director of LocalAmerica, which is partnering with InstantAtlas to develop sites built around how communities rate in livability. Local America is featured on Reynolds Journalism Institute’s Pivot Point site.

  1. November 30, 2012

    Nate Silver is a star because he found the way to answer the question everyone was asking: Who will win? Most of the questions asked in communities are NOT the type that lend themselves to data-driven answers.

    1. TomGrubisich
      November 30, 2012

      I admit I was a bit unclear on Nate Silver’s lesson for hyperlocal news sites. It’s true that Silver has proved to be a great election prognosticator. But he also operates in the present — by assessing candidates’ strengths right now, not just what they will do in the future vote. For his current-time assessments, he weights variables in polls according to their strengths (from very weak to very strong), thus minimizing the impact of outliers when he averaged the all the poll numbers. That’s what I picked up from Silver — the importance of looking at variables and giving them a relative weight in assessing numbers. When I looked at Anacostia High School’s academic performance, I was not trying to predict how many of its students would go to college and have successful careers, in a Silver-type prediction. What I was trying to do was figure out why performance of some students got better amid overall miserable performance. In other words, I learned from Silver’s methodology, not his prognosticating. Looking at the high truancy rate — a very big variable — I concluded the school actually had two student bodies — one consisting of kids who regularly went to class and the other of kids who regularly didn’t go. That discovery raised the big question — why aren’t these two student bodies rated separately? It was by looking at data and its variables that I was led to asking this question. Without the inspiration of Silver’s methodology, I might still being wracking my brain trying to figure out what was wrong at Anacostia High.

Leave a Reply

Your email address will not be published. Required fields are marked *

Can Businesses Use Hyperlocal to Leverage Customers’ Politics?