Technical Editing & Production



Graphics and Visuals

This section covers using graphics (figures and tables) in text. In general, everything that isn't a table is a figure.

Graphs and data presentation

In most instances, you have main 4 different way of presenting a set of numbers.

  Trial 1 Trial 2
Frank 304 234
Joan 302 286
Phillip 493 239


For data that must have exact values. People have a hard time getting trend information out of tables

Bar graph.

For discrete data.

The way to determine the choice between bar graphs and line is not which you think look best. Use bars when there is no meaning to the values between the points. Use a line graph for contineous data. Consider an example that shows salary by department. The line graph makes no sense because there is not value for a point between any of the department points.

Line graph.

For continuous data. It makes sense to ask what the value is at any point on the line. For example, iif the data plotted shows the relationship between voltage and resistance, any voltage value could be evaluated, not just the points at which values were recorded.

Pie graph.

For showing percentage that is parts of a whole. The parts must add up to 100% and it must make sense to think about it that way. For example, how a budget was spent.



Graph design

All graphs must have a title at the top. The title given in the figure caption is not sufficient. The title should reflect the content of the graph. Also, be sure to label the X and Y axis. A reader must be able to understand the graph without reading the text.

Excel has a habit of putting a Series box on each graph, which it defaults to a name of "Series 1", etc.. If there are more than one set of data being graphed, the series box is good, but must be labeled properly.. If there is only one, the remove it. This example shows both the series box and the title. In this case the series box should have been removed.

When you create graphics, make sure you control the graphing program; it must not control you, nor must it control how the data is presented. Graphs with elements like the "Series 1" on this graph show the writer didn't understand the program and brings into question the quality of the data.

Graphs should be drawn at a size that makes sense for their content. Many graph packages default to full page size. Yet, if the graph only had 6 bars, like the example above, then making it fill a page is a waste of paper. It doesn't need to be much larger than it appears here.

Writing about graphics information

When you write in the text about the information in a graph, summarize the information and tell the reader what they should get out of the graph. DO NOT simply repeat the information in text form. The reader can read that. You placed the graph into the document for a specific reason; by telling the reader what to look at and how to interpret the data, you ensure they get out of the graph what you want.

In this graph, notice how the following 2 paragraphs are very different in their write up of the information. The first one just tells what the graph says, the second one interpets the information and uses it as part of the arguement being made. After reading the first paragraph, the reader now knows that page count increased (so what). In the second, the reader understands what the page increase really means.

Paragraph 1 (don't explain graphs like this)

During the past four years the number of pages produced monthly by each writer has increased. In 1991, we produced 40 pages per month, in 1992 the production was 44 pages, in 1993 it was 55 pages, in 1994 it jumped to 58 pages, and in 1995 production was 60 pages per month. Figure 1 shows this change.

Paragraph 2 (explain graphs like this)

Figure 1 reveals a steady increase in page production. During the past four years the number of pages produced monthly by each writer has continued to increase at a rate of about 2 pages per year. Growing from 40 pages per month in 1991 to 60 pages per month in 1995. The jump between 1992 and 1993 occurred when we switched to Frame.

3D graphs

While Excel and most graphic packages make the 3D graphs look neat, they are very hard to read. Avoid the use of 3D charts unless they are strictly for glitz and the user had no need for the real data. For example, in a corporate annual report many of the graphs are 3D. The user wants to see up trends, but is not interested in the real numbers.


Three-dimensional graphics are often difficult to interpret. Understanding the relationships of the three lines in first figure is difficult because of the 3-D grid, which can create an optical illusion. A simple two-dimensional line graph of the same information would be less artistic but easier to comprehend. Similarly, because the axes slope upward in the three-dimensional bar graphs, readers have difficulty visually comparing the relative heights of the bars. The bars located in the back of the graph appear higher than the ones in the front.



3D graphics also can cause distortion because of the perspective. In the following pie chart, notice that there are several slices that are either 2% or 4%, but each slice looks to be a different size. A reader looks at a graph and mentally compares the size of slices, they do not read the labels before reaching initial conclusions about the information.


Sort of the same (equivalent) information with different presentations

showing carbon emissions


Lay readers and scientific graph scales

Notice how the scales increase in this graph. The y-axis has marks of .1, 1, 10, 100, 1000, 10,000. The x-axis uses a similar scale. These are both log scales, which are very useful in presenting information with a wide information spread. It also will take an exponential curve and produce a straight line. Because of this, they are often used in scientific papers. On the other hand, I'm willing to be that most of you don't have a clue what I wrote in the previous sentences. Scientists may understand graphs with log scales, but the general public does not.

If this graph didn't use a log scale, everything from the 100 line (the middle of the graph) would be smashed into the bottom (with equal spacing, how much is 100 when the scale tops out at 10,000?).

If you are doing the translation from science article to general public information (not unlike the health brochure assignment), you must consider how the various graphs will be interpreted and if the public has the knowledge to interpret them. If not, you may have to figure out a different way of presenting the information.

More on true scales and comparing information

These were two graphs I pulled off a news web site from a story about banks and their current loan balances. Each graph is ok alone, but putting them one over the other means they will be compared. But look at how problematical they are.

Each of these problems can cause people to ms-interpret the graphs. There is also the issue that most readers have no clue whether these are good numbers, poor numbers, etc. Perhaps the values are very low and the up curve on the bottom graph means the values are retuning to normal...or maybe up curves on this graph is bad (but defining the meaning of the graphs is for ENGL 6715.)

Notice in this graphic that since the y-axis starts at 610, it looks like Yahoo users have a credit score of less than half of either Gmail or Bell South users. Putting a graph like this into advertizing could get your company a phone call from the Yahoo or AOL lawyers about deceptive advertising.

Or this one. The bear market shows a decline of 56%, but it sure looks like a lot more than half. And the rebound rally would be 70% of the bottom, but it looks more like three times.

Areas versus heights of graphs

Notice how the youngest LuAnn looks much bigger than the middle LuAnn, even though it is less than twice as tall. The bars show the same relative heights as the comic, but they look much closer. The problem is that the area increases as a square of the dimensional increase. The eye responds to the area increase and thinks the largest item is much bigger than the others.

On a related issue, people are really poor at estimating area increases. If you use various sizes of rectangles to show two different values or as a means of showing relative size, people have a difficult time determining when an area is twice as big. They tend to underestimate it.




Design by Michael J. Albers Copyright 2012. All rights reserved.
Send me an email.