Scientific papers are often filled with data that are difficult to interpret. Tables can make complicated data clear and accessible to your reader, but they need to be used properly to be effective. In this blog post, I explain how to make sure that your tables communicate your results clearly to your reader.
Text or tables?
The first thing to consider is whether the table is necessary. A table that does not communicate a clear message will confuse your reader and clutter your paper. Small amounts of data can be described concisely in the text, for example:
The mean (standard deviation) BMI was 26.4 (3.7) in orthotopic transplant recipients, 27.4 (3.7) in living donor transplant recipients, and 26.9 (3.6) in split-type transplant recipients.
A table is not necessary here. But if you have large amounts of data to present, then a table will help your reader make comparisons and find specific values.
Components of a table
Tables contain multiple components. These may include some or all of the following:
- Table number and title
- Column headings (identifies the data entered in a specific column)
- Column spanners (additional column headings that further identify the contents of two or more columns, e.g. Sex over two columns headed Male and Female)
- Table spanners (these are additional headings that cover the width of the table and further divide the data, e.g. Baseline and Six-month follow-up to separate patient data collected at two different time points)
- Stub (the left-hand column that contains the row headings)
- Data field (the main body of the table composed of individual cells)
- Table footnote.
An informative title
The title should be a brief description of what the table shows. It should not contain too much background information and should not interpret the results. Take a look at the following examples:
Survival of cancer patients (too vague; this does not specify what the table shows).
Survival of cervical, prostate, testicular, bladder, and breast cancer patients in the United States, Europe, and Australia (too much information; the different variables can be presented in the table headings).
Survival rates are higher among cervical cancer patients (not suitable; interprets the results and does not say what the table shows).
Survival rates for common cancers in different countries (a good title; it tells the reader what the table shows without going into unnecessary detail).
Different categories of information should have separate columns in the table. Column headings should be brief but informative. Undescriptive headings like:
do not help your reader. Choose something more specific:
You can also use spanner headings to avoid unnecessary repetition in the column headings:
Also make sure that your column and row headings apply to all items in the respective column or row. For example, do not put n values into a column headed %. And capitalize your table headings consistently, according to the journal’s style guide.
The data field
The data field is composed of individual cells. Each cell contains a value, some text, a symbol, or a combination of these. Do not leave a cell blank unless it is clear that data do not apply here or an explanation as to why the cell is blank has been given in the table footnote. Use 0 to indicate the value is zero, and an ellipsis (…) or NA (not applicable or not analysed) to indicate that no data were collected.
Make sure that the data in your table are accurate – are the totals and percentages correct? Do the percentages add up to 100 (if appropriate)? Do the values in the table correspond to those in the text? Explain any discrepancies (e.g. due to rounding) in the footnote.
Values in the tables should correspond to those used in the text. For example, if you present the number and percentage of patients with postoperative infection in a table, do not refer to mean and standard deviation values in the text. And give comparable values to the same number of decimal places if possible.
Table footnotes provide additional relevant information that would otherwise overcrowd the table, such as definitions for abbreviations and source acknowledgements. They may also include notes that refer to a specific row, column, or cell; these are usually indicated by superscript letters:
aThis patient died during follow-up.
Symbols used to denote p values may also be explained in the footnotes (e.g. * = p < 0.05). Make sure that the same p value is designated to the same number of asterisks throughout the manuscript.
There are no set rules on how to organize information in tables. But the data must be presented logically and clearly so that the reader can quickly understand what is being shown.
Be concise. Limit the content of your table to essential information so that your reader is not overwhelmed and can easily find specific values. This also applies to supplementary tables. Although these typically contain more detailed information, this information must be relevant.
Remove any redundant information. For example, do not repeat the same data in different tables. And if all the elements in a column are identical, remove the column and put the information in the table footnote. For example, if all patients received laparoscopic surgery, then a column entitled Type of surgery is redundant. A note to this effect (All patients received laparoscopic surgery) would be better.
Present your data in a way that best conveys your message. You can arrange information in the data field to help the reader either evaluate the data or find a specific value. For example, in a table presenting mortality rates of patients with different tumours, the mortality rates can be presented from lowest to highest so the reader can easily see which tumours have the highest mortality rates. Or the tumour types can be listed alphabetically, so the reader can quickly find the mortality rate for patients with a specific tumour type.
Crowded tables that are not well spaced are difficult to read. Consider adding extra white lines in your tables (if space restrictions allow it) to avoid overcrowding and promote clarity. Adding units to every data field is another source of unnecessary clutter. Give the units in the column or row headings, or in the footnotes.
Referring to tables in the text
Every table should be mentioned in the text. If you haven’t referred to a table, then you probably don’t need it. Number your tables in the order they are first mentioned in the text. And refer your reader to a table when the data it presents are first mentioned, even if these data are discussed in more detail later on.
Refer to tables by their number (and capitalize Table when referring to a specific table). Do not refer to the Table above or the Table below because the table may change its position in the final typeset version of your article. You can refer to tables in parentheses:
Most patients had laparoscopic surgery (Table 1)
or make the table the subject of the sentence:
Table 1 shows that most patients had laparoscopic surgery.
No further information needed
Tables should stand alone without reference to the main text. A clear title and table footnote will help with this. Non-standard abbreviations that are defined in the main text must be re-defined in each table. Standard abbreviations and symbols (e.g. no. for number, % for percent, and SD for standard deviation) do not need defining in tables.
A clear overview
Tables are useful tools in a research manuscript because they make complicated information easy for the reader to understand. But they need to be used properly to be effective. Following the tips outlined in this article will help you to create clear, easy-to-follow tables in your next research paper.
Make sure your reader knows what your data are showing
Claire Bacon is a former research scientist with professional qualifications in copyediting and medical editing. She edits scientific research papers and teaches courses on scientific writing.