Margaret A.WinkerMD, Senior EditorIndividualAuthorPhil B.FontanarosaMD, Senior EditorIndividualAuthor
To the Editor.—The article by Dr Astin1 appears to contain a methodological error. Table 1 shows the independent variables used in the study, one of which measured general satisfaction and satisfaction with practitioners. The researchers note the use of a Likert scale for the response set for each question. However, in reading the table, it is clear that 2 of the items used a 4-point scale and 1 used a 5-point scale. Likert scales conventionally use the 5-point scale, which offers a neutral anchor point from which the respondent can select. The use of a 4-point scale infers the absence of such a neutral point and is often referred to as a forced-response Likert scale.Questions with different response sets should not be combined into a single scale. Any statistical analyses such as factor analysis and logistic regression done on questions with different scaling methods would be invalid.
Lawler MK. Why Patients Use Alternative Medicine. JAMA. 1998;280(19):1659-1661. doi:10-1001/pubs.JAMA-ISSN-0098-7484-280-19-jbk1118