Persuading Physicians to Test Their Patients' Level of Kidney Functioning: The Effects of Message Frame and Point of View

Abstract

A two-part field experiment was conducted to determine the effects of message frame (gain vs. loss) and point of view (personal vs. impersonal) on physicians' intentions and behavior to test their patients' level of kidney functioning. One hundred and fifty-one physicians returned a survey that accompanied one of four different experimental cover letters or a generic control letter. One hundred and twelve (74%) of these physicians also completed and returned a follow-up survey sent approximately 4 months later. Physicians who received a letter (vs. the generic-letter control group) believed their patients were more susceptible to kidney disease, believed that kidney disease had more severe consequences, and also demonstrated greater intentions and behavior to test their patients' level of kidney functioning. Additionally, there was a significant frame by point of view interaction effect, in that physicians receiving the gain-framed personal letter or the loss-framed impersonal letter demonstrated greater intentions and behavior than physicians receiving other versions of the letter. These results extend the theoretical scope of the EPPM by suggesting that threat to other can motivate behavior change, and also can have significant practical application for the development of messages targeting physicians.

Notes

¹In order to reduce complexity and provide a more focused report, this article focuses on the effects of messages varied in terms of frame and point of view. In a separate study (Roberto & Goodall, 2009), we also test the major tenet of the extended parallel process model.

²As part of another experiment that took place after this study (Roberto, 2009), we sent cover letters to a different group physicians that served as the experimental group for this other study. About 4 weeks later, we sent these same physicians a short post-test survey that included an item designed to assess how many of them recalled receiving that letter (n = 130). Even though we sent a cover letter to every one of these physicians about a month earlier, only about 38% recalled receiving the letter. While it is possible that some physicians received the letter and did not remember it, our informal discussions with several physicians suggest it is more likely that these individuals never received the letter in the first place (i.e., a gatekeeper or screener simply did not pass them along; Harbaugh, 2002). Assuming a similar process happened during the experiment reported in this study, that means only about 38% (n ≈ 461) of physicians received our materials, of which 151 (or approximately 33%) returned the survey.

³Age and years of practice were significantly and positively correlated, r(149) = .88, p < .001. Thus, those who are interested in increasing response rate by asking fewer or less sensitive questions (Dillman, 2006) might consider using just the years-of-practice item in future research.

⁴A series of one-way ANOVAs was also conducted to see whether the four experimental groups varied on any of the perceived threat and perceived efficacy variables under investigation. No differences were observed for susceptibility, F(3, 111) = .16, p = .92, severity, F(3, 111) = 1.50, p = .22, response efficacy, F(3, 111) = .50, p = .69, or self-efficacy, F(3, 111) = .64, p = .59.

⁵Analysis for the behavioral measure on the initial survey was not included in this analysis because the item referenced a time period that occurred before physicians in the experimental groups were exposed to the experimental manipulation. Thus, physicians in the experimental group should not differ from physicians in the control group this measure. To be thorough, one-way ANOVA with contrasts was conducted to test this supposition. As expected, the experimental groups did not differ from the control group with regard to past behavior on the initial survey using a two-tailed t-test, t(142) = 1.79, p > .05.