ABSTRACT
Conducting and reporting reliability assessments of the content analysis protocol and process, which includes one or more reliability coefficients, are a necessity. A full report of the reliability process should be made in a research article, which also includes study protocol definitions and procedures. (The full protocol should be made available by study authors on request.) Furthermore, information on the training of coders, the number of content items tested, and how they were selected should be included in the article. At a minimum, the specific reliability coefficients along with confidence intervals should be included for each variable in the published research. In calculating reliability coefficients, researchers should randomly select a sufficient number of units for the tests, and then decide whether each variable reaches acceptable reliability levels based on coefficients that take chance agreements into consideration. Failure to systematize and report the procedures used, as well as to assess and report reliability, virtually invalidates whatever usefulness a content study may have for building a coherent body of research. This chapter also discusses the nature and appropriateness of often-used reliability coefficients.
