A shortcut to rejection: how not to write the results section of a paper
Streiner, D.L.
Canadian Journal of Psychiatry. Revue Canadienne de Psychiatrie 52(6): 385-389
2007
ISSN/ISBN: 0706-7437 PMID: 17696025 Document Number: 612351
This article discusses common errors in writing up the results of papers. It covers the following: (1) giving details about irrelevant topics, such as what program was used to enter the data, while ignoring important ones (for example, which options were chosen for various statistical tests); (2) reporting P levels of 0.0000 and negative values for t tests; (3) giving the P levels but not the actual values of the statistical tests; (4) not including confidence intervals and measures of the magnitude of an effect; (5) testing irrelevant hypotheses, such as whether reliability or validity coefficients are significantly different from zero; (6) attributing reliability and validity to tests rather than to the circumstances under which they are given; and (7) reporting levels of accuracy that cannot be supported by the data. Suggestions are made regarding the proper way to report findings.