In each of the four cases above, the process of assembling a results chapter led the students to reflect on their results and to some extent re-interpret them. Perhaps, you might argue, this reflection should have happened earlier. My sense is that it can’t happen earlier—it is the discipline of writing it all down in a coherent form that al- lows identification of issues.
On that reasoning, there are two lessons. One is that the writing of this chapter should not be left too late. Indeed, the writing of all of the chapters shouldn’t be neglected, but this one is particularly important, because no student wants to be in the position of changing what their thesis is about at the last moment.
The other lesson is that doing research is a continual cycle of activity, learn- ing, reflection, and refinement. It doesn’t stop when the method is ‘proved’ or the results or gathered or even, often, when the thesis is submitted. You should expect to continue to have new insights about your work—and, perhaps unfortunately, to find new problems in it as well—for a long time after the bulk of your investigation is complete. You should always be prepared to look at your work with a critical eye and, if there is a problem to address or a refinement to make, then do so.
108 8 Outcomes and Results
Illustrations
A typical results chapter consists of argument and narrative supported by illustra- tions, that is, graphs, diagrams, pictures, and tables. But why are you using them?
The immediate answer to this question is typically: ‘I include a figure when it ex- presses the point I wish to make more clearly than does the written word’. On this principle, illustrations are likely to play a role in many parts of your thesis. I discuss them here because the results chapter is one place where they are not just helpful, but essential.
However, the use of illustrations to make a point is, I believe, only part of the reason they are of value. If you wish to get the best out of your illustrations, you need to put yourself in the position of the reader. Do they read the written text un- til they get to the sentence, ‘Fig. 6.2 shows that increasing the population density decreases the per capita consumption of petrol’, and then dutifully find Fig. 6.2 to check that this is indeed so? Probably not. My experience is that, long before read- ers begin to work through your results chapter, they will have opened the thesis and skimmed through it, ‘reading’ the diagrams and looking at the graphs and tables. It is this, rather than the careful reading that comes later on, that seeds the examiner’s understanding of your results.
After other preliminaries, such as looking for the aim of the research and reading the conclusions, the real reading begins. The written text develops ideas in the way that the writer intended, and readers will no doubt follow this development. But at the same time they will be generating their own interpretations and impressions.
They will compare material in one chapter with diagrams or text in another in ways that the writer had not anticipated. They might refer to and puzzle over Fig. 6.2 long before they read the text that discusses it. They might return to it again when some- thing written in Chap. 8 triggers another train of thought. Such exploratory reading is an essential part of thorough understanding of complex work.
Readers use several complementary channels of communication simultaneously, some using words and some using visual images. They do not use one at a time, switching from one to the other; rather, they use all of them at the same time, per- haps giving one more attention than others at any given moment. Nor is learning always linear. Think of lectures you have been to where the lecturer has used slides to complement the spoken word. You are busy looking at one of the slides and thinking about it, while still listening to what is being said, when suddenly, much to your annoyance, the slide disappears. The lecturer, already busy with the next point, didn’t think it was of any more interest to you, although you were busy integrating it with the rest of what was going on in the lecture.
This leads to some rules about visual material:
• The reader should not have to read the text that refers to the illustration to un- derstand what the illustration is meant to demonstrate. Although an illustration should always be ‘called up’ by the written text, it should make sense by itself.
In the caption, you should explain the context and how the illustration should be
interpreted, and draw attention to features you wish the reader to note, even if you have discussed these in some detail in the text.
• Don’t cram in too much detail. When I ask students their view on the functions of tables, for example, they often reply that it is to record data such as experimental readings in a systematic way. This being so, a table might have to contain large amounts of data, perhaps extending over several pages. In my view such data should not go in the main text, but rather in an appendix. A table in the main text must be a complementary channel of communication, and illustrative rather than exhaustive—that’s why they’re called illustrations.
• Reserve the use of illustrations for things that are important. The reader will fo- cus on them and assume that they are the most significant part of your work; use of illustrations for minor outcomes can skew the reader’s understanding of your argument.
• Put in a table only when the patterning obtained by arranging things in rows and columns tells the reader something better than or different from a written description. If the data in your table seems to you to demonstrate some trend or correlation, you should consider displaying the trend by means of a graph, and banishing the figures to a table in an appendix.
• A diagram should be a net aid to understanding. If the work of explaining a dia- gram that illustrates, say, risk factors in diet is more work than simply explaining the risk factors, then the diagram is an unhelpful burden and should be discarded.
(It can be helpful, though, to develop such diagrams for your own benefit, as they can clarify your understanding and help you focus in on what you are try- ing to say.) That said, never overlook the possibility that a pertinent diagram can greatly improve your presentation.
There are three kinds of figures: diagrams, graphs, and images such as photographs.
This book is not the place to give detailed advice on preparation of such materials, but it is important that you be aware of some general principles.
Some authors like to describe aspects of their work as line diagrams made up of boxes, circles, arrows, labels, and so on. Such diagrams can be a powerful way of explaining relationships, but they are inevitably simplifications of complex situa- tions, and may degenerate into caricatures. A diagram may be a useful way of il- lustrating the biology of food digestion, for example, but that doesn’t mean that a diagram is the right way to show the connection between the sociopolitical factors that control foodstuff quality.
Graphs are used to demonstrate trends or correlations; so you need to think care- fully about what you are trying to demonstrate. Usually you will be either confirm- ing an established model or developing a new one, and you should have this in mind when plotting your graph. A common error is the introduction of an extraneous variable. In a striking example of this, the authors were trying to demonstrate that reducing the lead emissions into urban air from the combustion of petrol would reduce the lead concentration in the blood of children. The data available to them were figures taken in the US over the five-year period 1976–1980. The writers plot- ted both the quantity of lead in petrol sold in major cities and the lead content of
110 8 Outcomes and Results
the blood of children in those cities against time in years. The two curves followed each other almost exactly. The obvious (but wrong) conclusion that the unwary reader would draw is that lead in blood is proportional to lead from petrol, with the corollary that all that one had to worry about in a program to control lead in blood was reducing the lead emissions from burning petrol. If the authors had plotted one against the other, without worrying about the distraction of the years in which the various values were generated, they would have found that the correlation was much less pronounced, showing that, although lead from petrol was an important contributor to lead in blood, it was not the only one.
With the ready availability of rich tools for creating graphs, such as three-dimen- sional and coloured plots, the challenge for the author is to repeatedly ask the ques- tions: Is this element necessary? And it is tasteful? No-one wants to read a thesis where every page is a blaze of iridescent colour like some sort of firework display.
The more detail, the less likely that it will be correctly understood. My advice is that you should test your graphs (and other diagrams too of course) on someone, and listen to what they have to say. If they don’t understand what is going on, simplifica- tion may be required.
Despite this availability of plotting tools, some people, mysteriously, still take their data and plot it by hand using software that is intended for creating pictures and diagrams. Do not do this; it will look ridiculous (or fraudulent) to every expe- rienced reader.
Some theses in lab disciplines used to include photographs of equipment such as assemblies used for preparation of chemicals; maybe they still do. I am not per- suaded that such photos are more useful than a diagram, and, in general, while I’ve seen a good number of cases where the photograph was included but seemed un- necessary, I can’t recall a case where I thought more photos were required. Some- times it is clear that photos are necessary (images of healthy and unhealthy cells, for example, or captured images of computer screens that demonstrate a software interface) but, for example, is it really appropriate to include photos of the teenagers interviewed about their eating habits? Maybe, but probably not; we all know that unhealthy eating can have consequences for appearance, and such photos can pass judgment on the individuals and suggest that the researcher has biases that may undermine the objectiveness of the research.
A picture speaks a thousand words, we’re told. What words does a poor picture speak? One thing that really bugs me, and forgive me if I have said this already in some form, is the fact that some students appear to be comfortable with including truly appalling illustrations in their work. I am always astonished by the students who labour for days or longer over a fragment of text but are comfortable with a jumbled, clumsy picture that doesn’t really illustrate anything. Unfamiliarity with tools is certainly part of this problem, as is the ‘but I am not an artist’ excuse. Per- haps they say to themselves that artwork is out of their expertise, and use this as a reason to quickly sketch something without even seeking advice.
The elementary tools for drawing figures and graphs with which most students are familiar, when used in an elementary way, are designed to be used by people whose use of computers is, well, elementary. They are used by children even before
they can read and write. That is not to say that these tools can’t be used better—but their default settings are certainly not intended for pictures that are to be included in research publications, and maybe it is a good point to remember that every paper ever published in a major journal over the last 350 years or so can still be found in a library somewhere. The vast majority of them are online at a permanent URL; your bad artwork can survive for a long time.
There are many specific things that I find jarring. This list is far from exhaustive:
• Lack of principles. Does a box represent an individual, or a collection, or an ac- tion? What is the difference between a black arrow and a coloured fat arrow? Do the colours have any significance? Why so many fonts and font sizes? Why are they so different from each other, and so much bigger or smaller than the regular text?
• Congestion. Lines that cross each other unnecessarily; arrows that end in space, or just inside the thing being pointed at; lines that might be pathways, but might be part of a boundary. Things under other things.
• Clip art. Are comical sketches (drawn by someone else, and often not very good) really what you want as the most visible feature in your thesis?
• Badly rendered photographs.
• Graphs with grid lines and unnecessary boundaries, poorly captioned legends, and too many lines. Missing data, out-of-proportion marks and line widths, poor use of space, and inappropriate sizing can all make a graph impossible to read, or worse, impossible to take seriously.
There are many good software tools for presentation, some of them free, in addition to those that are included in the common word-processing packages. Take the time to ask around and evaluate the options; you may be surprised at how much differ- ence there is between them.
Many of the principles noted above for figures are just as applicable to tables.
They should be self-contained, with detailed captions; they should not be amateur- ish. Choices need to be made about what grid lines to include or omit, how to align data in columns, what is important, and so on. Complex tables can present particular difficulties, when, for example, the data is multi-factorial but needs to be repre- sented on a two-dimensional piece of paper.
I’ve found that a great way to locate examples of good tables is to leaf through a first-rate journal. These journals use professional typesetters to arrange tables, and have deep experience in which table elements are clarifying and which are unhelpful. This is again a case where the defaults in word-processing software are poor—indeed, I would say they are outright dreadful. Find good examples from your discipline, and imitate them.
A final thought on this topic is that many of the problems I see in figures and tables are, I suspect, due to the fact that authors don’t anticipate how much work they are going to be. Given that the illustrations will be a conspicuous element in your thesis, expect them to require real effort to get right.
112 8 Outcomes and Results