Visualization of Data 2: Why Visualize?

Why do we need images? Why can't we just create a data table and write a bunch of words to explain our conclusions? We've heard it said, "A picture is worth a thousand words," but what does that really mean?

It's virtually impossible to get through a formal education without encountering hundreds of graphs, tables, diagrams, and photos intended to convey information. For that matter, we encounter visual displays of information everywhere - in our prescription medication literature, in political discourse, and even in our games. Yet both in my careers as a scientist and as a teacher, I've found that many of us are unable to read a graph correctly, and even less able to create one.


As a reader, you probably appreciate when an author makes something so clear that you cannot doubt their methods or conclusions. As a writer, your goal is to give your reader that experience.

Consider the data table at the right. As a means of recording and processing data, tables are great. From this table, we can observe what's being measured, the precision of the measurements, and what order they were measured in. For independent and dependent variable data, tables are usually sorted by the independent variable (x), but maybe there's something special the author wanted to convey about the order in which the measurements occurred. From looking at these data, you may be able to see that there's some kind of an increasing trend - the dependent variable is generally trending upward as the independent variable increases. However, to see outliers, or to determine a mathematical trend, a table is virtually useless. Check out this x,y-scatterplot of the same data:


With a graph, you can see that there are two obvious outliers. Although you'll need to run statistical tests to determine whether these points fall within your acceptable margin of error, at least now you know which points to test. The data trend is also obvious, and the calculated equation and R-squared values give you something you can use to describe the data's trend.


There are a couple of other things to notice about this graph. The axes are labeled, the title is in Y vs. X format, which is the appropriate format for the title of a scatterplot, and the axis labels have the same number of decimal places as the data points. Without the words on the graph, it could be a graph of anything; the words give the image meaning. As a teacher, I used to tell the students that their graph was obscene if it had no labels; NO NAKED GRAPHS!