All the same, in this post I would like to convince you that, by following a few basic principles, we can greatly improve the quality of a table, thereby making it easier for readers to properly interpret the results of our data analyses.
Specify the message / information
Each table placed in a report should present a specific conclusion drawn from the data analysis. The analyst should design a table so as to ensure that its message is clear and legible for the recipient "at first glance". Message formulation is of critical importance and helps guide the series of decisions required regarding the structure, content and look of your table.
Choose the best form of the message
The analyst must decide which data are to be in table rows, and which in table columns. Is the table structure one- or two-directional? What will be the sequence of columns? How will the rows be sorted? What statistics will be presented? All these decisions determine which data will be easy to compare, and which comparisons will become more difficult. Let's see the example below. Letters A, B and C mark three versions of the same table, which presents the number of concluded transactions broken down by years and regions. Two of them (A and B) have a one-directional structure – all the numerical values are presented in a column and can be compared with each other only in this way. The table marked with letter C has a two-directional structure – value comparisons can be made both in rows as well as in columns.
As you can see, when using one-directional tables it is important which variable will be assigned as the superior variable, and which as a nested variable. Table A enables easy data comparisons for regions within one year, but a comparison of the results for one region in different years would be more difficult. In table B it is the exactl opposite. Table C is most universal – it allows to easily compare both regions as well as years, and in addition takes least space, which gives it a clear advantage.
Design the visualization so as to show data
Data in the table should include not only the presented numeric values, but also row and column headings, titles and symbols that will help the reader to properly interpret the table content. The second consideration are structural table components such as grid lines, filling colors, typefaces, font styles and sizes, and even empty places that separate particular columns or rows from each other. These additional elements provide support in reading the table, however, their incorrect use may have the opposite effect by distracting attention from the data and diffusing the main message of the table. Thus, moderation in their application is recommended. When designing tables, the overriding principle is the pursuit of simplicity. Anything that is not necessary should be removed from the table. This applies both to data as well as additional elements. Too high a level of detail in the data may make it difficult for the recipient to understand the table's main message. Here are a few common mistakes:
- presentation of results with the accuracy down to 4 decimal places while it would be sufficient to round the values;
- presentation of all variable categories, even if only some of them have real importance (in this case, the solution could be to merge not very numerous categories);
- placement of statistics which could be omitted without damage to the message.
An illustration of the latter can be seen in tables D and E. They present results of a questionnaire survey regarding shopping tendencies. One of the analyst's tasks was to present dependencies between the respondent's own financial situation and the declared frequency of visiting shopping malls for pleasure. In Table D the analyst presented the results both as a percentage and as a raw number. Table E, however, presents the same results, but with the numbers omitted. They have been included only in the row "Total", allowing the reader to know the percentage calculation basis.
With a reduced number of the presented values, the table has become clearer and conclusions can be drawn from it more readily. In this case, it appears that the better their financial situation, the higher the possibility that the respondent visits malls. The worse the respondent's financial situation, the smaller the possibility that he or she spends time in this way. Additional table elements should also be subject to reflection. If we notice any unnecessary elements, they should be removed and the remaining ones, whenever possible, "muted" to prevent them from distracting attention from the data. For example, in some tables grid lines may turn out to be unnecessary (perhaps it is enough to separate rows and columns with an empty space?). In case we want to leave them, try to mute them by changing the color from black to light-grey and/or narrowing the line (compare table F and G).
Once we have disposed of unnecessary table elements such as grid lines, we gain extra space to optionally underline data that is important for our message. We may also select a bold font, add a frame or color to draw the recipient's attention to a specific part of the table.
By following the above principles tables will be more useful for our recipients, and the conclusions from our analyses will have a greater chance of being noticed and taken into account of by decision makers.