• you must demonstrate that you've used the program
• you must make extra runs and use the plots for direct comparisons
• you recognize that your audience is comfortable with that output format.

In most cases, however, the plots the computer program gives you via its default settings aren't going to be easily readable and may force you to make extra explanations. I have seen (for instance) electronic and control systems block diagrams that are difficult even for experts to read.

If you have to use direct program outputs, you can make them more readable by changing the program's plot defaults. That's not always as easy as it is in Excel, but it's nearly always doable.

• Remember that a block diagram is usually used to describe what a system does, more than what each individual block does.
• Remember that you can organize multiple blocks into "macros" (or, subsystems) and describe them in detail elsewhere for those who are really interested.
• Remember that scales and legends are usually adjustable and make use of that to focus in on whatever's most important.

Even small steps like these can enhance your readability quite a bit.

Pie and Bar Charts

Some authors believe that pie charts, bar charts, and their variations are not particularly useful for communicating information. Pie charts, in particular, are one-dimensional and use a lot of ink. But, given that most engineers have a good idea how to create pie charts, I'll write about 'em anyway. :-)

Tufte is one writer who thinks typically-used charts contain a great deal of "chartjunk" and actually obscures information from the audience. But list member Fred Klingener offers another view, in this case of the colors used (for instance) as Excel defaults:

Tufte's otherwise excellent advice presupposes a certain sophistication on the part of the viewer, a condition not always realized. When I prepare graphics for presentation to management, for instance, I use lots of bold splashes of primary colors. This seems to engage them.

This is to say that what's written here is not what Tufte endorses -- it's more like a compromise, submitted to engineers who'll use pie and bar charts no matter what anybody tells them. :-) Given that they'll be used, we must do what we can to make them presentable.

References

Tufte, E., Visual Explanations. Graphics Press, 1997.
Hilligoss, S., Visual Communication: a Writer's Guide. Addison-Wesley, 1999.

When there's a problem, it usually takes one of these forms:

1. You don't think (or know how) to override spreadsheet (Excel) defaults, even when they result in things that are useless, or worse. Consider: a legend with for a single curve, labeled "Series1." If it doesn't help you tell your story, you need to leave it out; and turning off legends is a single click.

2. You don't think (or know how) to get the most mileage out of color. Excel's default background for most charts is gray. Gray is in general not a good color for the contrast your readers need. Switching to white background is not that difficult. When you create bar or pie charts, since Excel's default colors are all over the board, you have to be sure that the colors of two different bars or slices aren't too similar -- otherwise, it's possible for the reader to get them mixed up. This is especially a hazard for those of you who print the charts in black and white. How many different shades of gray can the human eye discriminate? Answer: not many. You can get around this by
   • using patterns instead of colors to fill bars or slices
   • using fewer bars or slices (if possible)
   • labeling bars or slices (with text boxes) instead of using legends

   any of these techniques are usually an improvement over Excel defaults.

   List member Lisa Henn points out, however, that

   ...sometimes you have to really dig around to figure out how to change the Microsoft defaults. And there are enough things to change about the plots each time they are created as to make it tempting to give up and do it Microsoft's way.

3. Sometimes you may forget to tell the readers what it was you were plotting. Your readers actually need plot headings, axis titles, and sometimes pointers to whatever information is really critical.

4. Scale abuse takes several forms, but the one that seems to be most likely to cause trouble is tick marks that don't make sense. Too many significant digits obscure the message of a table, and that effect is even worse on an axis label. Use the simplest axes you can get away with and your readers will bless you. That means
   • ticks should be spaced evenly
   • the spacing should be according to small numbers of significant digits
   • the axes should cover the range of the data UNLESS there is some reference outside the data range (maybe zero, sometimes) that the readers need
   • the ticks don't all have to be labeled; label the ones that cover the data range and are still easy to read.