Results Section

The purpose of the results section of a research paper is to present the key results of a research study without interpreting their meaning. The results should be presented (a) in an orderly sequence so that their sequencing tells a story and (b) where appropriate, with the use of summary techniques and visual aids, such as tables and figures. The results section does not need to include every result obtained or observed in a study. It presents only those results that are relevant to the objectives and the research questions of the study.

Usually the results section is not combined with the discussion section unless specified by the journal to which the research paper will be submitted. The results section in general should not refer to the discussion section, which has not yet been written (but the discussion section should frequently refer to the results section). The results section, however, should closely reflect the methods section. For every result there must be a method in the methods section.

Below is a checklist of the items to be included in a results section:

Text First, the results must be organized in an orderly sequence, such as chronological order, logical order, or in order of most to least important. One recommended order in the results section is as follows: (a) Summary of what the data set “looks like,” with descriptive statistics (e.g., mean and standard deviation) for the relevant variables as found in preliminary analyses. (b) Examination of the hypotheses through further analyses. (c) Verification of which findings were or were not in the predicted direction, based on confirmatory analyses.

Second, whether the results are best presented in the form of text, tables, figures, or graphs is determined. Tables are good for including large amounts of information, especially of a repeated or comparative nature, for completeness. Scatterplots or histograms can provide a real feel for the data (including the noise). With complex data sets, bar graphs are useful so that the reader is not [Page 1281]overwhelmed with too many data points at once. When results are presented in tables or figures, the text should describe only the highlights of the findings and point the reader to the relevant tables and figures. The text should provide background information for the tables and figures, as well as observations not presented in the tables and figures. In other words, the text should complement the tables and figures, not repeat the same information.

Third, the text—not the statistics—should describe the patterns in the data. Statistics should not be used as though they were parts of speech. For example, “the correlation between variables A and B was r = −.26, p = .01” should be translated into words, with the statistics provided as evidence for the reported findings; for example, “a negative correlation was found between variables A and B, indicating that an increase in A was associated with a decrease in B (r = −.26, p = .01).” The word prove should be used with caution. Because statistical tests are based on probability and can be in error, they do not really prove anything. The meaning of the alpha level or the null hypothesis need not be elaborated if the intended audience of a research paper is the scientific community; in this case, the researcher can assume the reader will have a working knowledge of statistics. The word significant should also be used with care. A statistically significant difference is one in which the probability that the groups of data are identical with respect to the characteristic under study is so low (e.g., less than .05) that the investigator can confidently conclude that the difference exists. The opposite is “statistically nonsignificant.” A clinically significant difference is one that is considered important. The opposite is “clinically insignificant.” The sentence “It is significant that we found no significant differences among the groups studied” is a valid, though perhaps confusing, statement.

...