Visual Literacy 1

Graphs

Graphs are being used increasingly to convey data because graphs are visual and offer a quick impression, especially for conveying trends (increases, decreases, or stability of results). A table of numbers takes much longer to analyze. In surveys of businesses, many people report that graphs are being used in decision making with increasing frequency.

If you know what to look for–and look out for–graphs can be very useful to you. If you are not careful, though, they can be very manipulative.

Graphs must be viewed carefully for the following reasons:

• Graphs have more impact than tables or “raw” data because graphs are visual. They can bypass the analytic part of your brain unless you take care to look closely.
• Graphs represent interpretations of data. When the data has been specially selected, then the graph represents an interpretation of selected data.
• A given set of data can be plotted to create almost any impression. Because of the techniques available for constructing a graph, a graph may show a dramatic trend upward or downward when in fact the data do not support such apparent change.
• The person who draws the graphs influences the impact of the information and thus influences the decision to be made from the graphs.

Characteristics of an Ideal Graph

An ideal graph has several characteristics that help to insure its fairness. These characteristics are as follows:

1. The number axis begins at zero. The axis that measures something in the graph (whether dollars, bottle caps, tons, or units) should begin with zero so that the measurement scale is shown whole rather than in part. An axis that begins with a number larger than zero is usually a sign that the graph creator is distorting the graph–emphasizing an increase, for example. As a simple exercise, find a bar graph whose number axis does not begin at zero and, according to the scale of the graph, draw the bar lines down to zero. What difference in impression do you get from the two graphs?

2. Both axes are labeled. If one of the axes is not labeled, you don’t know what is being displayed. All you have is a vague visual representation. Unlabeled axes produce a graph that is the visual equivalent of emotive language–all sizzle and no steak.

3. Numbers are identified (dollars, percent, births, etc.). Labeling an axis with 10, 20, 30, 40 doesn’t mean much if you don’t know that those numbers represent.

4. X to Y axis scaling is fair. A favorite technique of deception is to stretch one axis out either to exaggerate or understate a trend. The X axis is the horizontal axis. The Y axis is the vertical axis. For line graphs, if the X axis is stretched out relative to the Y axis, the line will take on a less steep look, reducing the impression of any upward or downward trend. If the X axis is scrunched or reduced relative to the Y axis, the line will look steeper than it otherwise would.

5. The graph as a whole is not deceptive. Taken in its entirety, does the graph present a fair representation of the data?

Reading Tables

The key to reading tables is to understand first, what is being represented and second, what that data means.

In a table, look for the following characteristics:

1. What is it? What is being represented—births, deaths, birth rates, yearly data, trends, etc.

2. Where is it? Is this information about the world, the US, California, a city?

3. When is it? Is this for a particular year, a series of years, a particular month, an average month?

4. Which direction is the chronological progress? That is, if the table is annual totals for something, do the years increase from right to left or left to right? Many sources use left to right, which is good for graphing and makes chronological sense. Other sources, however, present most recent data first, and allow you to compare earlier data as you move to the right. (In the latter case, “most recent” is on the left end rather than on the right end.)

5. What is the unit key or legend? Are the numbers in units, percent, millions, or what? Look for a legend. This is a critical piece of information to determine, because it will affect profoundly the meaning of the table. For example, look at this table:

What does this table tell us? Unless we know what the 121 stands for, we really do not know how many sodas were consumed in 1996. Do those numbers represent

• sodas per capita in the US
• thousands of gallons of soda
• number of 52-gallon barrels
• millions of cans
• millions of 8-ounce servings

or some other unit value? Unless we learn what a number represents, we cannot understand the table. Note the improved table here:

A key to working with information in tables is to realize that the unit listed in the table may not be the unit you need to work with. The table may present 40-gallon barrels and you may need to calculate millions of gallons. Or the table may show a rate per thousand people and you will need to know per hundred or per ten thousand. So be prepared to convert one unit value to another.