Authoring a PhD

because it is distorted by the top two scores. The table’s notes give the range (the variation between the highest and lowest scores) and the midspread (the variation between the quartiles).
They also confirm that the top two observations are upper out- liers (that is, they lie more than 1.5 times the midspread above the upper quartile), and hence are highly unusual.
Layout.
Table 7.1 is made hard to read by being overly spaced out across the page. Many students (and some journal and book publishers) still seem to believe that every table should use the full width of the page, no matter how few columns it has. The effect is always that the row numbers are put further away from their labels than is necessary, an impact intensified hereby not using boxing or shading inside the table and by having the row numbers in column 2 spaced unevenly because some of the row labels use two lines. In addition the table uses a smaller font than surrounding text even though there is plenty of space on the page. It includes small superscript note numbers inside the table cells, which cloud the second column. And a succession of note numbers also clutters up the bottom of the table with unnecessary details. There is no clear finish to the table at the bottom and no source is given.
By contrast Table 7.2 uses minimum-width columns without overspacing, bringing row numbers and labels into closer proximity. Within the available space, always use the largest possible font size for tables, up to a maximum set by the main text font, as here. With large tables use a whole page in landscape layout to keep table fonts readable. And as here, you should box the rows and columns (which usually helps readers. The median and quartiles are highlighted in Table with light background grayscale shading (you could also use yellow or very pale shades of other colours with a colour printer. The design is uncluttered by note numbers within the table or other distractions. If some form of reference detailing needs to be given that is not essential to understanding the table, it is best handled by using an endnote in the main text accompanying the table. Table 7.2 uses a line under the notes and sources to achieve a clear finish to the table.
All these differences in Table 7.2 from Table 7.1 are generally applicable to every table you have to design. Just to recap, the 7 AUTHORING AP H D

most important principles are:
◆
Always have completely informative headings and labelling,
including full details of units or measurement and what the cell contents show.
◆
Use the need to know criterion to pick an appropriate level of detail for numbers. Choose the minimum number of decimal points needed. If you make use of index numbers or ratios, choose levels which give the most easily understandable numbers for readers. Consider how many
‘effective digits are needed, and use rounding or number simplification appropriately.
◆
Design all tables to show a numerical progression (except for tables showing overtime trends or categorical variables with a fixed order).
The final issue to consider about tables is whether you really need them at all. Would it be better to use a chart or graph instead of a table Inmost cases charts will be better because they are clearer and more visual. Tables should principally be retained for the following circumstances There are only a small amount of data to present, so that a simplifying chart is unnecessary Readers need to know numerical values more precisely than would be shown in a chart, for instance if there are fairly small variations in results The data to be displayed have very strong variation between the lowest and highest numbers so that it would be difficult to display the range of the data effectively in a chart. For instance, isolated high numbers for one or a few years in an overtime chart might necessitate a scale which would mean that readers could not detect any visible differences in other years figures, whereas in a table they could still be seen You want to compare data scaled in very different kinds of units or indices, and they could not easily be accommodated on one chart. Alternatively you might have the numbers indifferent columns which are of the same kind, but of such different sizes that they would be hard to scale together on a graph. Here tables can save space, since otherwise you
H AND LING ATTENTION POINTS 1

would have to provide a series of different charts for each column of numbers being covered You want to both present some primary data numbers, and then show calculations of how index numbers or ratios or compound statistics are derived from them Tables are being used to put reference material onto the record, for instance in Annexes or Appendices.
Designing charts and graphs
We live in a graphical age. In general if it is possible to display data in chart form rather than in tables it is desirable to do so,
subject only to the exceptions enumerated just above. Charts and graphs automatically screen out too much data being thrown at readers. They are easier for you to analyse correctly as an author, and for readers to interpret. Charts are especially important in showing the relative importance of different components or phenomena giving trends overtime and rates of growth and illustrating more complex patterns in data than just linear relationships, such as ‘curvey’ relationships. There are now many different types of chart for displaying simple data available on spreadsheet packages (like Excel or Lotus) and widely used data-analysis programmes (like SPSS or Stata). Both
PhD students and established academics often make mistakes about choosing the right kind of graphic to go with their data.
Figure 7.1 shows eight of the most commonly used charts and for each of them points out a few uses for which they are well or poorly adapted.
As with tables it can be useful to briefly compare a poorly designed and a well-designed chart version of the same data tables discussed in the previous section. Figure 7.2 (on pis a vertical bar chart version of the table in Table 7.1; and
Figure 7.3 (on pis a horizontal bar chart version of the table in Table 7.2. The differences in the accessibility of the two bar charts are every bit as noticeable as in the readability of the two tables, and again it is worth briefly itemizing why.
Labelling.
Figure 7.2 has a very poor heading and axis labels compared with Figure 7.3. The choice of a vertical bar design for Figure 7.2 means that there is no space for the health board 7 AUTHORING AP H D

HANDLING ATTENTION POINTS Index of potato production (1997
=
100)
100 110 120 121 135 140 122 1998 Year 2001 2002 2003 0

Download 2.39 Mb.

Share with your friends: