Reply to Roseman, Milette, Zhao, and Thombs
- PDF / 84,959 Bytes
- 3 Pages / 595.276 x 790.866 pts Page_size
- 64 Downloads / 157 Views
LETTER TO THE EDITOR
Reply to Roseman, Milette, Zhao, and Thombs Michael F. Scheier, Ph.D. & Joel B. Greenhouse, Ph.D. & Heather N. Rasmussen, Ph.D.
Published online: 1 April 2010 # The Society of Behavioral Medicine 2010
Roseman, Milette, Zhao, and Thombs [1] purport to have found “numerous inconsistencies” in our meta-analysis [2] of the relationship between dispositional optimism and physical well-being that lead them to question the scientific basis for our conclusions. We respectfully disagree with their assessment. We use our reply to respond to the “errors” that they identified, to make a few general points about the practice of meta-analysis, and to answer the question they raise in their provocative title, “Is optimism associated with physical health?” Their Table 1 describes two kinds of coding errors: one involves coding for the nature of the covariates that the study used, and one involves the coding for study design. As Roseman et al. [1] note, information about the coding for the covariates factor was not included in the original paper, but was provided by us at the request of Dr. Thombs. Under extraordinary pressure from Dr. Thombs for a rapid response, a version of our data that was not current and contained some errors was inadvertently sent. It is important to note, however, that the database that was used for the analyses reported in our original paper did not contain these errors. It is unfortunate that Roseman et al. did not contact us directly when the apparent discrepancies
M. F. Scheier (*) Department of Psychology, Carnegie Mellon University, Pittsburgh, PA 15213, USA e-mail: [email protected] J. B. Greenhouse Department of Statistics, Carnegie Mellon University, Pittsburgh, PA 15213, USA H. N. Rasmussen University of Kansas, Lawrence, KS, USA
in the database were discovered, as the source of the inconsistencies would have been easy to identify. Still, we apologize for the incorrect information that was forwarded. Roseman et al. also claim that there are inconsistencies in our coding of study design. According to our explicit decision rules, longitudinal studies are defined as those that assess optimism at one point in time and measure some outcome at a later point in time, but do not adjust for baseline factors. Prospective studies are longitudinal studies that adjust for baseline factors. For some of the entries in their Table 1, e.g., Contrada et al. [3] and De Moor et al. [4], Roseman et al. are correct that we miscoded the studies. We agree that using multiple coders could possibly have caught these errors. For other inconsistencies listed, the resolution is less clear. Take, for example, the study by Giltay et al. [5]. Roseman et al. suggest that overall survival (the outcome used) was only adjusted for age and gender and thus was longitudinal in nature. Our response is twofold. First, age and gender are predictors of survival and as such, according to the coding rules we used, would classify the study as prospective. Second, in the text of the paper, Giltay et al. [5] report an anal
Data Loading...
