| To Register SMDM Homepage | |
|
|
|
||||
Purpose
Research on survey methodology has demonstrated that seemingly innocuous aspects of a survey’s design, such as the order of questions, can produce biased results. The current investigation extends this work by testing whether standard survey introductions alter the observed associations between variables. We fielded a brief survey of health and well-being to Parkinson’s patients, and varied the survey introduction to test whether informing participants that the study focuses on Parkinson’s disease (PD) would make health status more salient, and thus increase the correlation between health and overall life satisfaction.
Method
153 PD patients agreed to a short phone interview on health and well-being. After initially agreeing to participate, participants were randomly assigned to receive either standard survey instructions, indicating that we were contacting people with PD, or a version that indicated only that we were contacting people living in their area. All participants were fully informed of the purpose of the survey at the conclusion. The survey first assessed life satisfaction, and subsequently health satisfaction.
Results
Health satisfaction was a much larger predictor of life satisfaction for the group who received the “Parkinson’s” introduction (r2 = .48) than for the group who received the introduction with no mention of the illness (r2 = .16; regression interaction p < .01). An examination of means indicated that the PD survey introduction significantly decreased reported life satisfaction for participants below the median in health, and slightly increased life satisfaction for participants above the median in health (interaction p < .05).
Conclusions
When participants were informed prior to the survey that its purpose was to examine well-being in PD, health satisfaction was a much more important component of life satisfaction, accounting for 3 times as much variation. We hypothesize that the survey introduction primed participants’ health status, resulting in an artificially large correlation with life satisfaction (a previously published meta-analysis on the link between subjective health and life satisfaction agrees with the smaller estimate). An implication is that fully informed consent can bias survey results. In innocuous surveys, it may well be preferable to provide full information about the purpose at the conclusion of the survey.
See more of Poster Session - Clinical Strategies; Judgment and Decison Making
See more of The 26th Annual Meeting of the Society for Medical Decision Making (October 17-20, 2004)