Return to Stat Topics

Graphing Guidelines

If the graphs below do not appear, try using Mozilla Firefox as your internet browser.

A large percentage of people find reading graphs difficult, so you should make your graphs as clear as possible for their sake. The advantage of this is that everyone else will find that your graphs easier to understand. The following are some rules of thumb for creating graphs. As you create more and more graphs, you will gain a better understanding of what works and what doesn't, and you can modify these guidelines accordingly.


Present the data clearly.

Start frequency and relative frequency scales at zero. If you have a choice, choose bigger graphs over smaller graphs.

Focus on the data.

With the exception of frequency scales, numeric scales should extend from the minimum data value to the maximum data value but not much farther.

Provide explanations as needed.

Add titles to graphs, label axes, and identify different data groups clearly if there are more than one.

Avoid distracting details

DON'T use 3D bar charts or 3D pie charts. While the three-dimensional look of 3D bar charts and 3D pie charts may intrigue the viewer, 3D bar chars and 3D pie chats are more difficult to interpret than the 2D varieties. Also, do not code different data groups with letters or numbers when descriptive names for each group could be used instead


Examples of Good and Bad Bar Charts


Figure 1: GOOD Bar Chart

This is a good bar chart.  The chart has a title and each bar is labeled below by its region. The frequency scale starts at zero.  Since there are horizontal lines every 20 units, the highest horizontal line is at 140, which is the first multiple of 20 above the tallest bar.

Figure 2: BAD Bar Chart

The frequency scale on this bar chart goes too far up.  As a result, the chart does not focus on the data.  The horizontal lines here are marked at multiples of 50, but since there is no bar above 150, there is no reason to go above 150.  Notice how the bars in Figure 1 are twice as tall as those in Figure 2, even though both graphs are about the same size.


  Figure 3: BAD Bar Chart

This graph does not present the data clearly.  The frequency scale starts at 20, instead of 0 like it should.  South America's bar is barely there giving the impression that almost no one comes from South America even though South America’s frequency is fairly close to that of the “Other” group.


Figure 4: BAD Bar Chart

This is called a 3D bar chart.  Because of the way the bars are drawn, they look like they are three dimensional.  Unfortunately, drawing the bars this way, makes the chart more difficult to interpret.  In this chart, it looks like the frequency for Central America is less than 120, but if you look at Figure 1, you can see the frequency is more than 120.  Similarly, the frequency for Asia looks like it is less than 80 in this chart, but Figure 1 shows that the frequency is definitely above 80.  The three-dimensional look is a distracting detail.  Using 2D bar charts instead of 3D bar charts eliminates the problem.



Graphing Rules for this Class


For the graphs you create in this class, you will be expected to label and apply scales to the axes of all graphs on tests and assignments. The numeric scales must be uniform. This means that if on one axis, the distance from zero to one is one-centimeter, then any other one-centimeter distance on that axis should represent one unit. The other axis may have a different scale. Always use tick marks to show the location of numeric values on numeric scales. All frequency scales or relative frequency scales must start at zero. Finally, use a ruler when drawing graphs. Failure to do any of these things may cost you points.


Return to Stat Topics