Remember, the reason for using tables is to help readers compare data, both within and between tables. We've already noted that, because items can be more easily compared down a column than across a row, it is best to place independent variables in rows and dependent variables in columns.
To facilitate comparison between tables, be consistent in table presentation. When the reader scans the column headings from left to right, and the stub headings from top to bottom, some sort of order should be apparent in both. Pre-treatment measurements, for example, might precede post-treatment ones; disease symptoms could be arranged from mildest to most severe. If there is no compelling reason for some other type of order, even listing material in rows or columns by the size of the numbers will help readers make mental comparisons.
The poorest sort of arrangement appears in analyses where data are arranged by arbitrary numbering of experimental subjects. Unless the numbers are truly relevant to understanding the results and some sort of explanatory code is included, experimental designations (A-307, D-10, and PC-2069) mean nothing to anyone but the investigator. In most cases, one can number the subjects by a conventional numeric system if they must be noted individually in the text. If they do not need individual mention in the text, omit subject numbers entirely in the accompanying table.
Research on illustration effectiveness (Macdonald-Ross, 1977a, b) suggests giving row and column averages as reference points. These averages can provide a visual focus that allows readers to inspect the data easily. However, do not clutter up a table with columns of numbers that could easily be derived from other columns by simple arithmetic.
Was this article helpful?