The following definition is found in Cabell's Directory of Publishing Opportunities: Psychology and Psychiatry, David W.E. Cabell, Editor, 2007-2008 on page XX.
With some exceptions a refereed article is one that is blind reviewed and has two external reviewers. The blind review requirement and the use of external reviewers are consistent with the research criteria of objectivity and of knowledge.
The use of a blind review process means that the author of the manuscript is not made known to the reviewers. With the large number of reviewers and journals, it is also likely that the name of the reviewers for a particular manuscript is not made known to the author. Thus, creating a double blind review process. Since the author and reviewers are frequently unknown, the manuscript is judged on its merits rather than on the reputation of the author and/or the author's influence on the reviewers.
The use of two (2) reviewers permits specialists familiar with research similar to that presented in the paper to judge whether the paper makes a contribution to the advancement of knowledge. When two reviewers are used it provides a broader perspective for evaluating the research. This perspective is further widened by the discussion between the editor and reviewers in seeking to reconcile these perspectives.
In contrast to these criteria, some journals that have attained a reputation for quality do not use either a blind review process or external reviewers. The most notable is Harvard Business Review that uses an editorial review process. Its reputation for quality results from its readership whose continual subscription attests to its quality.
In addition to these criteria, some researchers include the journal's acceptance rate in their definition of a refereed journal. However, the method of calculating acceptance rates varies among journals. Some journals use all manuscripts received as a base for computing this rate. Other journals allow the editor to choose which papers are sent to reviewers and calculate the acceptance rate on those that are reviewed that is less than the total manuscripts received. Also, many editors do not maintain accurate records on this data and provide only a rough estimate.
Furthermore, the number of people associated with a particular area of specialization influences the acceptance rate. If only a few people can write papers in an area, it tends to increase the journal's acceptance rate.
Although the type of review process and use of external reviewers is one possible definition of a refereed article, it is not the only criteria. Judging the usefulness of a journal to the advancement of knowledge requires the reader to be familiar with many journals in their specialization and make their own evaluation.