Monday, March 23, 2015

Bad Visualizations


An example of bad visualization. The visualization is trying to show the distribution of Spanish speakers in the US.

Source of the data: http://www.pewhispanic.org
Source of infographic is: http://www.daytranslations.com/blog/2015/03/infographic-the-surge-of-spanish-speakers-in-the-us-5955

It is extremely confusing and the message doesn't get conveyed for the following reasons:


  1. The percentage of Spanish speakers as a population is marked as a colored fill in the area of a state. Since different states have different sizes the percentages are not normalized to one quantity visually. For instance Florida and Colorado have nearly the same percentage of Spanish speakers but the visualization does not convey that. 
  2. The colors used to fill the state boundaries have no meaning and are distracting.
  3. There is no theme or pattern that emerges from the visualization.
  4. There is an overlaid graph of the growth of Spanish speaking population in the US. It adds nothing to the message of the visualization (how has the growth affected the distribution of Spanish speakers?) and makes it more busy and confusing.
  5. The growth chart has a shadow effect inserted that serves no purpose.
  6. The y-axis for the growth chart is in percentages whereas the data values appear to be absolute population numbers.
  7. There is a 17% line running through the growth chart that seems to indicate the current (2013)percentage of the Spanish speaking population. However it is misleading for all other years as the base population for other years was not 316,128,839 and the percentage in 1970 may well have been 17 as well, however the chart gives the message that it was much lower.

A much better visualization of the same data is done via a heat map of the percentage population shown below:

Distribution of Spanish Language Speakers in the United States

  1. The heat map clearly indicates the percentages of Spanish speakers in each state using one normalized signal, color. The shades are also distinct enough so that color blind people can also visually follow it.
  2. Hovering over individual states gave the exact percentages for each state, which reduces clutter in the figure and provides information as needed.
  3. A clear pattern emerges from the visualization: Southwestern states have a high concentration of Spanish speakers and there are pockets elsewhere (e.g. Florida, Illinois and New York) with a sizable number of Spanish speakers.
  4. The population growth chart has been eliminated.

A simple population growth chart can be depicted in the following way:

  1. The units are consistent. Absolute population is plotted vs year.
  2. The misleading percentage figures have been eliminated.
  3. From the chart it is evident that the percentage of Spanish speakers in the US has not changed significantly unlike what the original infographic implies.