{
  "id": "1995-may-when-an-apple-isnt-an-orange",
  "slug": "1995-may-when-an-apple-isnt-an-orange",
  "url": "https://dwillis.github.io/uplink/article/1995-may-when-an-apple-isnt-an-orange.html",
  "year": 1995,
  "month": "May",
  "monthNum": 5,
  "headline": "When an apple isn't an orange",
  "kicker": "",
  "author_name": "Andrew Lehren",
  "author_title": "NICAR staff",
  "affiliation": "NICAR",
  "summary": "Philip Meyer, a Pulitzer Prize-winning pioneer of computer-assisted reporting, shares five statistical techniques journalists can use to strengthen their stories, including adjusting for inflation, controlling for lurking variables, using scatterplots, and applying the chi-square test.",
  "keywords": [
    "statistics",
    "journalism",
    "data"
  ],
  "topics": [
    "Investigative Methods",
    "Data & Databases",
    "Journalism & Newsroom"
  ],
  "technologies": [],
  "full_text": "Philip Meyer, a founder of computer-assisted reporting and author of The New Precision Journalism, believes there is much more reporters can do with numbers to improve their stories. In a recent Uplink interview, the Pulitzer Prize winner listed several tips:\n\nAvoid apples-and-oranges comparisons. Among the most common, he said, \"is any story where there is a dollar comparison over time.\" In other words, a newspaper may have cost only a few cents at the turn of the century. But before you get misty eyed about the good old days, don't forget to figure for inflation. Reporters can do that getting the Consumer Price Index issued by the U.S. Bureau of Labor Statistics.\n\nWatch out for \"lurking variables.\" Journalists often write about how one thing correlates with another. Meyer believes they should always check for other variables, and see what happens when those other variables are held constant.\n\nFor example, he cited stories that rank states based on Standard Achievement Test scores. The problem is that, in some places, a smaller percentage of students take the SAT scores the college-bound who, overall, post better scores. Elsewhere, a larger percentage of students take the test. By using a simple regression, reporters could adjust for differences.\n\nShow a scatterplot. No math here. This kind of graphic is good for showing relationships. Say you are reporting about how a natural disaster reaped heavier damage in neighborhoods built during a certain period. Peg the vertical (y) axis to the year constructed; the horizontal (x) axis to percent destroyed. Then plot each block on the graph. The scatterplot marks each bit of data effectively. Newspapers don't use scatter blocks enough, Meyer said.\n\nThe chi-square test. What are the odds that something happened by coincidence? The chi-square test is a quick way to find out. A variety of books describe the chi-square test, including Meyer's The New Precision Journalism.",
  "provenance": {
    "extracted_by": "pre-v2",
    "extracted_at": "",
    "verified": true
  },
  "similar": [
    {
      "slug": "1996-may-winners-use-car",
      "score": 0.207
    },
    {
      "slug": "2006-september-october-time-for-meyer-awards",
      "score": 0.193
    },
    {
      "slug": "1996-april-uplink-update",
      "score": 0.186
    },
    {
      "slug": "2005-septemberoctober-philip-meyer-journalism-award-contest-honors-stats-pioneer",
      "score": 0.185
    },
    {
      "slug": "2005-novemberdecember-entries-in-philip-meyer-journalism-awards",
      "score": 0.167
    }
  ],
  "html_url": "https://dwillis.github.io/uplink/article/1995-may-when-an-apple-isnt-an-orange.html",
  "json_url": "https://dwillis.github.io/uplink/article/1995-may-when-an-apple-isnt-an-orange.json"
}