Page 91 - How to Write and Publish a Scientific Paper, 8th Edition 8th Edition

Page 70

Consistency in Representing Data

The measurement intervals you choose should stay the same throughout the graph. For example, if a time-line graph
represents yearly increments, make each time interval on the x-axis equal to all others, both in measurement and in the
period of time represented. If you are using lines, ensure that each plot line in the graph has the same visual weight as
the others. Vary the visual weight only if you wish to make one particular variable plot in the data set more highly
featured when compared to the others. An alternative is to create lines of equal weight and color, in which each line
uses a different symbol at the node point. Define each data set clearly, either with a legend or by using called-out text
next to each line.

Preparing Graphs for Publication

There are many advantages to creating graphs electronically for publication. One is the ability to work in the final
width specified by the publication for reproduction. Small graphs are usually one column in width; large graphs can be
two columns wide, if necessary. Make sure that all the text is large enough to be read easily but not so large that it
dominates the page and confuses the viewer. Most publications prefer a sans serif typeface such as Helvetica for
labels and captions. Numbers are easier to read and reproduce in a sans serif type. Final type size (if the graph will be
reduced) should be no smaller than 8 points. Boldface type can be used for important labels, such as axis names.
When creating bar graphs, avoid using patterns to differentiate between data sets. Black, white, and two or three levels
of gray usually suffice. When using shades of gray to differentiate the data, use the following percentages: 25, 50, or
75% of black. When your data set

Page 82
includes a larger number of variables, labels can be used to name each variable.

When using lines as the data variables, keep them at least 0.5 point thick, but no larger than 1 point if your lines are
100% black. Lines can be differentiated by symbols or by using a different dashed line pattern for each data variable.
When using symbols, keep them all the same size; 10 or 11 points works well. When working with dashed lines, keep
all the data lines the same thickness but make sure that each line pattern is sufficiently different from the others for
clarity. Avoid using gridlines within the graph unless they are needed for clarity. Tick marks alone can frequently
provide the information needed. Information describing the data variables can often be made clearer by creating a
legend showing their use. When writing the legend for a graph, describe each plot in the data set so that it is
unambiguous and easy to follow.

The graph shown in Fig. 5 is the graph displayed earlier in Fig. 3 redone in DeltaGraph Pro. The key at the upper right
of Fig. 5 describes each data variable visually. The legend below the graph describes each variable in detailed text.

file:///C|/...208%20Books%20(part%201%20of%203)/How%20to%20write%20&%20publish%20scientific%20paper/14.htm[4/27/2009 12:58:18 PM]