Preparation pays off

By Robert Gebeloff The Star-Ledger

Let's be frank - the 2000 Census is a four-alarm headache.

There's a mountain of data to climb, and at the summit sits some of the most nondescript field names in the history of data processing.

Quick - how many P0010005s did you have in your state? And what did they call P001005 in 1990 and 1980? Then there's the matter of mapping. Learning how to use the canned maps that come with ArcView is no easy feat. But learning how to convert Tiger maps and tie them to the Census data - that's an entirely new skill.

And then there's the none-too-small matter of journalism - You've conquered the data, now what's the story? Can you transform this mountain of information into news, news that would qualify as left(news,3), rather than a rehash of trends you already reported to death?

Be prepared

All of this requires preparation - the Census is just too big to do well on deadline. At The Star-Ledger, we took all the standard measures. We sent people to IRE Census seminars, we acquired and processed historical data, we did battle with the Tiger Maps, and we devised and revised our list of stories, graphics and maps throughout the weeks leading up to C-Day. But we also made a major investment in what I'll call logistical preparation. We cover the state and compete in eight local markets as well. On Census release day, we would need to get data that was cleaned and analyzed to our bureaus in Washington and Trenton, not to mention Woodbridge, West Caldwell, and Morristown - on deadline. What we came up with was a somewhat Rube Goldberg-esque system of Access tables, SQL Server Views and intranet Active Server Pages. The idea was that the data would arrive on Census day, a button would be pushed, and poof the 2000 numbers would flow into place, fuel for thousands of preprogrammed calculations.

Smooth sailing

I expected, of course, that the system would cough and wheeze a few times along the way, but to my amazement, it worked without a hitch, and enabled us to execute our coverage plan pretty much to the letter. The New Jersey files arrived on my hard drive at about 3:15 p.m. on a Thursday afternoon. By the time the last edition was hitting the presses, we had managed to get seven stories in the paper, illustrated by four photos, five maps, and 15 charts. The stuff was flowing through production so fast that in one early edition, a map got printed twice, but otherwise, the paper came out looking like it was the result of a month-long project. And then we were well positioned to come back Sunday with more.

All of this requires preparation - the Census is just too big to do well on deadline.

I mention quantity, but having prepared was a matter of quality too. For example, we got the 1990 PL 94-171 data and went to the voting precinct level to generate demographics for the current state legislative districts. Then we coded each district based on the make-up of the three-member delegation. This enabled us to do precision reporting on political shifts.

While the competition noted generally that population growth appeared to be greater in Republican-controlled districts, we reported precisely that 422,000 of the state's 684,000 population increase occurred in districts with all-Republican representation.

Showing growth

Same idea with sprawl. It was obvious by looking at county numbers that the growth in New Jersey was occurring in places that were mostly rural and suburban. But why approximate? We got a list of Urban/Rural/Suburban town classifications as labeled by the state planning department. We simply tied that list to the Census data and voila we were able to say precisely that the population growth...

Census Lessons:

We'd like to hear from those of you who had interesting experiences in working with the PL94 Census Data that was released in March.

Are there lessons you learned that would be beneficial to others?

If this was your first experience with Census data, was there something you wished you had known earlier?

Did your news organization come up with a unique story based on the data?

We'd like you to share your experiences in an upcoming issue of Uplink, preferably by writing a piece yourself. We could also arrange for a NICAR staff member to interview you for a story.

The stories can be anything from a few paragraphs to a few pages - whatever it takes to share what you learned.

Please contact us as soon as possible if you're interested.