Why context is as important as the data itself

May 21, 2010  |  Design, Statistics

John Allen Paulos, a math professor at Temple University, explains, in the New York Times, the importance of the before and after of when you get that data blobby thing in your hands.

The problem isn’t with statistical tests themselves but with what we do before and after we run them. First, we count if we can, but counting depends a great deal on previous assumptions about categorization. Consider, for example, the number of homeless people in Philadelphia, or the number of battered women in Atlanta, or the number of suicides in Denver. Is someone homeless if he’s unemployed and living with his brother’s family temporarily? Do we require that a women self-identify as battered to count her as such? If a person starts drinking day in and day out after a cancer diagnosis and dies from acute cirrhosis, did he kill himself?

In a nutshell, statistics is a game of estimation. More often than not, the numbers in front of you aren't an exact count. They could easily change if you shift the criteria of what was counted. As a result, there's always some amount of uncertainty attached to your data, and it's the statistician, analyst, and data scientist's job to minimize that uncertainty.

So the next time you see a list of rankings like "fattest city" or "dumbest town," don't take it for absolute truth. Instead, think of it as an educated guess. Similarly, when you analyze and visualize, remember the context of your data.

Catch Paulos' full article here.

5 Comments

  • Agreed! Context can make a number mean something, or something else…. a take on web operations and context:
    http://www.kitchensoap.com/2009/05/10/context-and-operational-metrics/

  • Yep – context is essential for meaningful comparison. Statrs me thinking about ontologies, linked data and the semantic web.

  • That’s one of the main issues I have with pop-infographics: they rarely provide an estimate of uncertainty. It’s like reporting regression coefficients without standard errors.

  • True words indeed … I always find it important to realize that all our (statistical) tests are associated with a large number of assumptions that are excluded from the test itself. Indeed, these assumptions include the data collection and coding, assumption regarding the substantive theory being used, and of course the statistical procedure itself.

  • We agree data needs a context in order to be valuable. Numbers only tell a part of the story; we need meaningful ways to define, categorize and interpret the data. This is especially important when it translates into real-world decisions and actions. We wrote a feature on how we can give meaning to data through beautiful, communicative designs: http://www.healthymagination.com/stories/decoding-data/. Check it out if you have the chance.

Unless otherwise noted, graphics and words by me are licensed under Creative Commons BY-NC. Contact original authors for everything else.