Text vs. tables: Summarizing research results

Nurse Author & Editor, Fall 2001 by Tulman, Lorraine

Presenting quantitative research data clearly and succinctly is the key to communicating the findings of a study. Summarizing data in tables is the way to paint the clearest picture for the reader. However, tables must be accompanied by sections in the text that tell the reader what to look for in the tables. This experienced researcher and author provides tips on how a manuscript can present research results clearly and succinctly by coordinating text and tables.

Authors, reviewers, and editors work together to write or edit a research manuscript, so that the results are presented clearly in a combination of text and tables. Here are two common situations:

Situation 1. You have completed data collection and analysis and have made an outline or list of the significant and non-significant findings. Now the time has come to write the results in a report or manuscript. You begin to write and very quickly realize that your findings are quite detailed and would be clearer if they were presented in a table. You wonder, "What should go in the text and what should go in a table?"

Situation 2. As a reviewer of a manuscript submitted for publication, you notice that the authors have used several pages of text to describe inferential statistical tests used to analyze the data of their study. As you read, it is hard to keep all of the results straight in your mind. Presenting the data in tables would help the reader to focus on the important findings. You ask, "What should I tell the authors to move from the text to tables?"

This article answers these questions as it gives some practical advice on which findings to put in tables and which ones to summarize in text.

Look Closely at All Numerical Data

Consider putting all detailed numerical data in tables; well, almost all. A helpful guideline I use is: if you have more than two consecutive sentences containing numbers that relate to each other, consider putting them in a table that shows this relationship. Numerical data that relate to the demographic and main variables of your study are prime examples of data that often fit well in a table.

Demographic variables. Descriptive statistics such as means, standard deviations, ranges, and frequency distributions used to describe the demographic variables of the sample may fit in a table. These variables may include age, gender, education, income, marital status, and health status. If you have randomized your subjects to groups, then this table should be broken down by group to illustrate (hopefully!) equivalence of the groups. An example of an effective table of demographic variables is in Table 1.

Study variables. Descriptive and inferential statistics for the main variables of your study can be emphasized well in table format. Group means, standard deviations, and ranges for individual variables as well as statistical ratios and p values are clearly displayed in a table. Table 2 shows how study variables displayed in a table emphasize the research results.

Construct Strong Tables

Tables require considerable thought and organization to make them communicate the data clearly and efficiently. The following guidelines can make table construction easier for the author and editing easier for the reviewer or editor.

Use percentages instead of numbers wherever possible. Percentages are easier for the reader to compare in a mental picture of the distribution than absolute numbers. When using percentages, however, make sure that the numerator and denominator are clear to the reader. For example, if describing the percentage of mothers who are employed full time, clarify if the denominator is the number of all mothers or just all employed mothers.

Keep number of decimal places to two. Our statistical programs give results to three or four decimal places. However, using more than two decimal places usually makes for very cluttered and hard-toread tables. "Numbers should be rounded to reflect the precision of the instrument or measurement" according to the American Medical Association Manual of Style (1998, p. 530), which for many statistical measures is two decimal places.

Report p values precisely, if the journal's style permits. Remember that p values of 0 or

Put recurrent words in column or row headings. The main culprits here are "time" words such as "days," "weeks," and "months" or mathematical and measurement units such as "%," "mg.," or "cm." One reason to put these results in a table rather than text is to reduce these redundant units, which have to be included for each occurrence in text.

Make the table freestanding. Think of each table as a slide that you might present at a conference. A table should have a title that is descriptive of its contents. The number of subjects must be included. Notation should be made as to what the numbers represent-for example, counts, percentages, or statistical ratios.

Don't Duplicate the Data in the Text

Once the table is developed, check the content of the text section. You want the text to enhance, but not repeat, the text. The following are tips writers and reviewers can use to write or suggest revisions for the text part of the manuscript.