You may be wondering why I’m devoting a post to such an apparently trivial aspect of technical writing, but the truth is, it’s really important and most people do not do it particularly well. There are three key aspects to consider when it comes to using figures and tables effectively: relevance, citing and formatting. We’ll discuss all three.
Before you include any figure or table in your thesis or report – ask yourself a few questions. Why am I including this figure (or table)? How is my reader going to use it? How will it help me to get my message across to the reader? In terms of figures, the old saying that “a picture is worth a thousand words” is useful to keep in mind when deciding on relevance. If including the figure will save you a lengthy and convoluted explanation and make it easier for your reader to understand your message, then it’s definitely worth including. Similarly, if you find yourself writing a paragraph listing various numbers (for example, a bunch of parameter values), it might be a little less tedious for your reader if you provide this information in a table instead.
What about data? Should you include figures and/or tables showing every piece of data you collected? Again, I would suggest that you consider the relevance questions above. If you have 10, 20 or more figures that look similar, perhaps it would be best to show just one or two example figures for each type of data in the main body of your thesis or report and put the rest in an appendix. As for tables, I suggest you limit those to a few columns and about 5 to 10 rows maximum. That’s about all the human brain can absorb effectively. Most people simply pass right by large tables without looking at them at all. Whenever I see a huge table of numbers, I immediately think, “why didn’t they plot this in a graph?”
Assuming you accept the premise that figures and tables are there to help your reader, then you must also accept the fact that you’ll need to explain them to your reader. Don’t simply add a reference to them as an afterthought and don’t expect your reader to know how to interpret them without some guidance from you. For example, consider the following text and figure.
“The proposed model is more suitable than Hibler’s model (Figure 1). “
We’re definitely going to need a bit of help to interpret such a complex figure. In particular, we’d be justified in asking why the proposed model is better. The figure turns out to be extremely useful for making this point, but only because the actual paper (She et al., 2009) provided the reader with information on how to interpret it. Along the same lines, don’t just cite a bunch of figures as a lump and leave the reader to struggle through them. If you can’t come up with a couple of sentences to describe each and every figure – then you really have to reassess whether they’re actually needed. In this way, relevance and citing are closely related – often it is the explanation, as much as the figure itself, that determines a figure’s (or a table’s) relevance.
Finally, don’t describe the substance of a figure and then mention the figure afterwards. There is nothing more annoying than struggling through a tedious, intricate and lengthy description of something only to read in the last sentence, “Figure 2.3 show this…”, especially when the figure is essential to understanding the explanation. Now your reader has to go back and re-read this whole paragraph again, an extremely frustrating and irritating experience. In fact, if they’re a member of your committee, they will probably start a list of nasty questions to ask you at the defense right around this time. Think about it – you need a topic sentence for this paragraph anyway and isn’t it always a pain trying to come up with these? Here’s a great one: “Figure 1 shows …”, then the rest of the paragraph explains the figure.
Proper table and figure formatting is extremely important – yet most people put very little thought or effort into it. First of all – be sure to number them in the order they are cited. (This may seem obvious, but if I had a nickel for every time it’s come up…) When formatting tables, make sure the column headings are meaningful. Also, be sure to use a consistent precision or a consistent number of significant figures, whichever is more applicable, for each parameter or variable. (Many spreadsheet programs leave off trailing zeros, which is totally inappropriate in engineering and science. It’s essential to format these cells to present the numbers correctly.) For figures, there’s a lot more to think about. I am not sure who came up with the default graph formats in the most common spreadsheet programs, but they just don’t work for the engineering/science crowd. Judge for yourself for the example below – which of these two figures do you think would be more useful to your reader?
What’s wrong with the top one? In the first place, it’s in colour and it needs to be; looks what happens when it’s printed off in black and white.
It’s starting to get hard to figure out what’s what. Imagine what happens when the figure gets more complicated? Of course, it would become totally incomprehensible unless printed in colour. If you have 100 figures in your thesis and you’re paying the cost of duplicating these for the entire defense committee, would you really want to have this many colour copies to make? Probably not – it’s way too expensive. In fact, most students prepare figures just like this (in colour) but print them in black and white for the defense committee, making the legends completely useless! It’s extraordinarily frustrating. I’ve asked many students about this at defenses and I always get the same answer, “Oh well, I was just going to make them in colour for the final copy.” I’ve got news for you – probably the only people likely to read your thesis, aside from your family, are the members of your defense committee! The rest of the world will wait for the journal papers. Just remember – anytime you torture the defense committee unnecessarily, you run the risk of inspiring them to come up with really difficult questions for the defense – it’s just human nature to want revenge when you’re being tortured. 😉
So should you abandon colour altogether? Not necessarily. You can use colour in your figures, but use different symbols and line patterns, too, so that the figure can be interpreted whether it gets reproduced in black and white or in colour. The same applies for journal papers where colour figures can cost ~$500 to $1000 each. If you create figures that are equally interpretable in black and white or in colour, you can opt to have them reproduced in black and white in the print version of the journal AND have them appear in colour for the on-line version. This strategy also works well in business practice. Your company may like to go with glitzy colour figures, but always keep in mind that most technical reports end up getting photocopied and redistributed in black and white. Whatever your technical writing project, you can’t go wrong by making your figures interpretable in both black and white and colour.
It is also critically important to format your figures as you go – don’t leave yourself a big mess to deal with at the end, unless you want to spend a month doing nothing but formatting figures (definitely a very boring and tedious prospect). In addition, if you draft your figures up to be journal ready right from the start, you only need one set of figures for reports/theses and publications. Many spreadsheet and other graphing programs allow you to define custom graph formats; invest a little time to learn how to do this and you’ll save tons of time overall. Here below are the formatting specifications I ask my students to use in their theses – they just come from the requirements I find typical of many of the journals we publish in.
- Use a sans serif font throughout (e.g. Arial or Helvetica).
- Use these minimum font sizes (keeping in mind that many journal figures get reduced in size dramatically):
- Axis titles 12 pt bold
- Axis labels 10 pt
- Legend 10 pt
- Box the graph (axes) but do not put a border around the graph.
- Do not use the automatic title at the top.
- Box the legend, but do not use a shadow. Also, usually, the legend goes inside the graph.
- Use symbols for discrete data (unless so dense that it would be illegible).
- Use lines, not symbols for model or equation results.
- Do not use thin line weights – they do not photocopy well, nor do they come out well in PDF files.
- Turn off line smoothing; it can distort the graph to the point that it is meaningless (as seen in my example above).
- Increase the line weight of both axes, and the tick marks – the default weight is too light to print well when copied or reproduced in a journal. Don’t make them too heavy though. (I find that a 1 pt thickness for these is a good compromise.)
- Use “cross” for major tick points, “inside” for minor tick points.
- If the graph includes negative numbers, move the origin to the minimum, don’tleave it at zero. Draw a thin line across at zero – using a fake data series (omit it from the legend though).
- If the x-axis is in days, set the major increment to 7 , 14, or 28 days (i.e. multiples of weeks) and the minor increment to 1 or 7 days, whichever is more suitable. If parts of a day are more suitable for the minor increment use 1, 6, or 12 hours.
- Use superscripts, subscripts and actual symbols in labels and legends. (e.g. m3/s not cms, 10°C not 10C, etc.).
- Be sure to label the axes! Also, units should follow the axis label. e.g. Water Elevation, m or Water Elevation (m). Whichever method you use, be consistent throughout.
- Do not use variables in axis labels, unless the definition follows as well. Variables may be used in the legend, if defined in the figure caption.
- Use error bars on data whenever appropriate.
- In a spreadsheet, place the graph on a worksheet page not a graph page. This gives you much more control on graph properties, especially size. Also, make sure the graph, and the data for the graph, are in the same workbook.
- Put the figure number in the worksheet tab (e.g. ‘Fig 10’) and label the corresponding data worksheet as ‘Fig 10 (data)’. Create an index of all your files.
Keep in mind that the journals you publish in may have more specific or different requirements, as may your thesis supervisor (or company, if you are writing technical reports). The important thing to remember is to be consistent and always think of your reader!
Thanks for reading this blog – comments are welcome! In the upcoming posts I’ll be providing strategies for doing your literature review and I’ll provide some advice on outlining and thesis content, as well.