In this issue of Health Services Research, we introduce a new, occasional series, the Methods Corner, which will aim to improve the practice of health services research. The article in this issue, entitled “Interaction Terms in Nonlinear Models” (Karaca-Mandic et al. 2011), describes the best approach for estimating and presenting the results of nonlinear regression models that include interaction terms among the explanatory variables. Although this topic has received attention over the last few years in the economics literature, practice remains suboptimal in many of the articles submitted to our Journal and to other journals in our field.
The Methods Corner, which we believe will be exceedingly valuable tzpo health services researchers, will feature articles on challenging problems that researchers frequently encounter in the analysis of data in our field. These articles will have very different goals and structure from the Methods Articles and Methods Briefs that we have always published. Specifically, while our current Methods Articles and Methods Briefs generally describe new methods or new applications of existing methods, the Methods Corner series will feature articles about established methods or applications where actual research practice is often deficient or suboptimal, perhaps because the source articles are not well known in our field or are difficult to access. In other words, Methods Corner will aim to improve current research practice, and the articles will have a decidedly didactic goal. While our criteria for the articles may evolve as the series develops, initially we will seek articles that:
Clearly and succinctly describe the method or application addressed in the article, and comment on why current health services research practice is often suboptimal.
Provide a clear explanation of the problem and the solution or solutions, using at least one example from health services research.
Explain why some approaches are better than others and whether there are any circumstances that make one or another approach more appropriate.
If relevant, provide well-documented statistical software computer code that shows how the solution can be implemented and suggest how the code could be modified to accommodate the researcher's own data. Our preference is to use Stata code, but code for other programs can be used, particularly if other programs are more appropriate for the particular application. In general, this code should be included as an appendix, to be published in the print version, electronically, or both, depending on length.
Many of the methods presented in these articles are likely to have a strong mathematical basis. In these cases, we will favor articles that present the key issues in clear, accessible text, with limited use of mathematics, and that develop a more complete mathematical treatment in an appendix. This appendix would be an electronic-only publication and as such would not be included in the word or figure maximum counts.
Our Journal's usual length limits will apply to the articles published in the Methods Corner: 4,800 words and five tables and/or figures. However, we will consider longer articles if there is clear justification.
We anticipate that most Methods Corner articles will be solicited. Researchers rarely write articles with the characteristics we seek on their own, at least in part because such articles are difficult to publish. We are excited that the new series will provide a publication outlet for this type of article.
Finally, we request the assistance of the health services research community in launching the Methods Corner series and ensuring that it will become a regular and helpful feature in Health Services Research. You can assist us by suggesting topics for Methods Corner articles, identifying experts who would be able to write the type of article we seek, or volunteering to write an article yourself. You may contact us at: