Data Visualization: Utilizing Data in Charts and Mapping

Many media outlets use data visualization to present information in an easily understood format, so that they can add supporting material to their articles. Through effective use of data, you can save words for more developed arguments and present a more visually pleasing result.

Bar Charts

One of the most common of these formats is a bar chart. I made one, shown below, using the tool Datawrapper after a number of steps to collect and organize the data.

First, I collected a series of data from the U.S. Census Bureau. I selected the topic of median income of prostitutes of different races in 2013. Originally, the material showed the statistics for only one state — Hawaii — and included the income of male prostitutes, in addition to female prostitutes.

In order to make a more detailed but focused comparison, I stripped the male data from the chart so as to streamline it. I also removed (i) the margin of error, to leave an exact statistic, (ii) the category of women who didn’t work full time, and (iii) the combined total of women working full-time and part-time.

This left me purely with data showing the estimated median income of women of different races who worked full-time for 12 months as a prostitute in Alaska in 2013.

However, this allowed for only a limited comparison within a particular state. Therefore, I repeated these steps with the equivalent data from other states, namely Alaska, Florida, Texas and New York. I was then able to compile these stripped statistics into a single table, in order to contrast the economic success of particular races across different US states.

This final data was then compiled into a bar chart:

Conclusions from the chart:

• White and Asian prostitutes are consistently the most economically successful across all of the states
• The New York prostitution industry makes the most money in total than in other states
• Hispanic prostitutes nearly always earn the least income, on average
• Hawaii is the only state — of those listed — where Native Americans or Native Alaskans are in the top two earning races
• In Texas, there is a disproportionate spike of income for Asian prostitutes

All of these conclusions can be easily drawn from the above chart, without needing to consult the various pages of original data. In this way, it can be seen how a bar chart can convey information more quickly and effectively, thus reinforcing an argument made in a piece of journalism.

Maps

Another way of showing data, particularly data that compares information across states, is to use a map:

Median Income ($) of Households by State using a 3 Year Average, 2011–2013

These maps use Google Fusion Tables to create a graphic, such as the ones above. Through merging a data file and a shape file, that maps the state lines, it is possible to create a map that shows information across each state.

I gathered the data from the U.S. Census Bureau once again, this time utilizing the information for Income of Households by State Using 3-Year-Average Medians. The 3 years in question here were 2011–2013. Once again, the data needed to be cleaned up.

Raw Data for the Median Income ($) of Households by State, using a 3-Year Average

I stripped it down by removing the columns listing the standard of area and 90% confidence interval, similarly to how I deleted the columns listing margin of error for the previous chart. I also removed the columns that discussed how many states shared that median income/were above or below that median income. This is because I knew that this would be visible once the final map was created, so it wasn’t needed in addition to the pure data.

First rows of simplified data

Once I had this simplified data, I saved it as a Google spreadsheet. I then used Google Fusion Tables to turn it into a graphic:

Dot Map for Median Income ($) of Households by State using a 3 Year Average, 2011–2013

Following this, I merged the graphic with an existing shape file, so as to change the visual from dot data to state data. This would allow for immediate visual interpretation and comparison, without having to click individually on data points:

Median Income ($) of Households by State using a 3 Year Average, 2011–2013 (zoom in)

The original zoomed-out map is shown above and available on Google Fusion Tables here.

Conclusions from the map:

• The highest median household incomes are on the Northeast coast
• The lowest median household incomes are in the Southeast
• The coasts tend to have a higher median income than the central states, however there are exceptions to this, e.g. Colorado
• The smallest states (clustered in the Northeast) have the highest median incomes
• However, the largest states don’t necessarily have the lowest, e.g. California

This is all quite easily concluded from this map, without needing to analyze a chart with a list of 50 different data points. In this way it is clear how a map can enhance a piece of journalism.

In summary, these two examples of data visualization provide clear demonstrations of how such graphics can be used to enhance the content of journalism.