Tuesday, 23 April 2013

Data Presentation Is Always Subjective

One of the endearing qualities of data is that it is factual.  Facts are a piece of truth and are not simply a matter of one's opinion.  Data can be misrepresented, incomplete, or erroneous, but accurate data is always objective.

At least that's what I believed in my early days of working with data.  The reality is that any presentation of data is subjective, and therefore the objectiveness of the underlying raw data becomes less significant.  One can avoid subjectivity by never presenting the data, but that also assures the data remains useless.

Jake Porway has an interesting blog article at Harvard Business Review entitled The Chart Wars Have Begun.  He refers to an interactive chart in The New York Times that shows how job data and the unemployment rate can be presented three ways:  the "normal" way, through a Democrat lens and through a Republican lens.  None of the views are dishonest or misleading, they are just choices that must be made for any presentation of data.  It's a simple but excellent demonstration of the power of data presentation.

Screen capture of the NY Times chart mentioned above.  Click here for the interactive version.

Every data chart or table requires the creator to make decisions that include:

  • what minimum and maximum values should be shown on the y-axis.
  • how much historical data should be displayed.
  • what scale should be used.  (i.e. linear or logarithmic)
  • whether multiple data sets should be displayed together.
  • the type of chart (e.g. pie, bar, line, scatter) or the organization of the table (e.g. which fields are displayed in which order).
  • whether certain information should be highlighted, such as an arrow showing a trend or an arbitrary baseline.

Each of these decisions will lead the reader to realize or overlook certain conclusions from the data.  The data itself may be truth, but the presentation will always be dependent on the subjective choices of the presenter.