{
  "id": "1998-november-out-of-date-data",
  "slug": "1998-november-out-of-date-data",
  "url": "https://dwillis.github.io/uplink/article/1998-november-out-of-date-data.html",
  "year": 1998,
  "month": "November",
  "monthNum": 11,
  "headline": "Out-of-date data",
  "kicker": "COMMERCIAL CENSUS UPDATES",
  "author_name": "Paul D'Ambrosio",
  "author_title": "",
  "affiliation": "Asbury Park (NJ) Press",
  "summary": "The author argues for annual census counts instead of the current decade-long gap, highlighting how outdated census data makes local demographic reporting inaccurate and forces journalists to rely on commercial vendor estimates that use varying methodologies.",
  "keywords": [
    "census",
    "demographics",
    "estimates"
  ],
  "topics": [
    "Demographics & Census",
    "Data & Databases",
    "Investigative Methods"
  ],
  "technologies": [
    "Claritas"
  ],
  "full_text": "Note to the next Constitutional Convention: Please make the census an annual event. The Founding Fathers sure didn't have journalists in mind when they came up with the count-everyone-once-a-decade concept, confounding our obsessive need for total accuracy.\n\nTry answering these questions on a municipal (or lower) level: How many immigrants live in town? How many toddlers are there (whose parents may not even have met before the last census)? What is the richest neighborhood in the county?\n\nThe 1990 census is almost a decade out of date, making such microdata as age woefully inadequate – if not downright paradoxical – to use: If there were 200 toddlers in town in 1989, does that mean there are none now, but there are lots of 10- to 15-year-olds? There are only two solutions. Ignore the 1990 census data by taking a sabbatical until 2000 or buy census updates from commercial vendors.\n\nEducated guesses\n\nCommercial updates have been the lifeblood of census information at the Press for the last five years. But keep in mind that the methodology for calculating estimates differs from one company to another. Most use algorithms based on death and birth lists, change-of-address data from the post office, and even mini-surveys.\n\nIn short, what you get with an estimate is a big ol' guess. Maybe a pretty good educated guess, but a guess nonetheless. Given the choice between using the 1990 census or an update, I would choose the update every time, with a note to readers that the information is based on estimates.\n\nI use Claritas, which packages data mainly for businesses and marketers. Claritas comes with a mapping program and is",
  "provenance": {
    "extracted_by": "anthropic/claude-sonnet-4-6",
    "extracted_at": "2026-07-05",
    "verified": true
  },
  "similar": [
    {
      "slug": "1998-november-census-resources",
      "score": 0.288
    },
    {
      "slug": "1998-november-county-population-estimates",
      "score": 0.287
    },
    {
      "slug": "1998-november-quick-hits-and-the-long-term",
      "score": 0.249
    },
    {
      "slug": "2002-mayjune-using-census-data-for-non-census-stories",
      "score": 0.247
    },
    {
      "slug": "1997-april-census-profiles-communities",
      "score": 0.243
    }
  ],
  "html_url": "https://dwillis.github.io/uplink/article/1998-november-out-of-date-data.html",
  "json_url": "https://dwillis.github.io/uplink/article/1998-november-out-of-date-data.json"
}