Mesfin Awoke Bekalu 1,2 (MPhil) Steven Eggermont 1 (PhD) 1 KU Leuven, Belgium

1. Framing HIV testing messages for urban and rural audiences: evidence from a field experiment in northwest Ethiopia Mesfin Awoke Bekalu1,2(MPhil) Steven Eggermont1 (PhD) 1 KU Leuven, Belgium 2 Bahir Dar University, Ethiopia

2. Framing health messages • One of the various ways of matching messages with recipient characteristics. • Different framing techniques such as temporal (e.g., Bonner & Newell, 2008), personal/relational (e.g., Ko & Kim, 2010) and gain- vs. loss (e.g., Rothman & Salovey, 1997; Schneider, 2006; Rothman et al., 2006

3. Framing … • Relatively better empirical evidence for gain- vs. loss-framed messages – messages that focus on the benefits of performing a recommended behavior and those that focus on the costs of failing to perform the behavior, respectively (Smith & Petty, 1996; Rothman & Salovey, 1997; Schneider, 2006; Rothman et al., 2006; Abhyankar, O’Connor, & Lawton, 2008). • Framing research has identified differentials in effectiveness of gain- vs. loss-framing based on the type of health behavior promoted –prevention or detection (Rothman & Salovey, 1997)

4. Framing … • Prevention behaviors: preventing the onset of a health problem (e.g., condom use, sunscreen use, etc.) – better promoted by gain-framing. • Detection behavior: detecting a health problem (e.g., HIV testing, mammography, etc.) – better promoted by loss-framing.

5. Framing … • The prevention-detection classification, although very important, does not always capture individuals’ construal of a given health-related behavior. • Research focus shifted to specifying the optimum conditions in which gain- and loss-framed messages would be most effective (Devos-Comby & Salovey, 2002; Rothman et al., 2006; Latimer et al., 2007).

6. Framing … • Optimum conditions such as direct/personal experience and issue involvement. • Direct/personal experience – the primary determinant of how people construe a given health behavior (Rothman et al., 2006). • People with personal experience of testing for HIV are more likely to construe the behavior as a means of monitoring their health (whatever is the test result), whereas those without are more likely to perceive it as a means of detecting the presence of the virus

7. Framing … • Issue involvement: • Prevention behaviors – gain-framed messages tend to be more effective among individuals with high issue involvement, while loss-framed messages are more likely to be effective among people with low issue involvement (Millar & Millar, 2000;Jung & Villegas, 2011). • Detection behaviors: loss-framed messages are more likely to be effective among individuals with high issue involvement, while gain-framed messages are more effective among people with low issue involvement (Maheswaran & Meyers-Levy, 1990; Banks et al., 1995; Meyers-Levy & Maheswaran, 2004; Jung & Villegas, 2011).

8. Framing … • The prevention-detection classification becomes particularly difficult when it comes to HIV testing. • While primarily a detection behavior, HIV testing is being promoted as a prevention behavor across prevention contexts for two main reasons:

9. Framing … • Biomedical: HIV testing  ART  viral load suppression (e.g., Wilson et al., 2008; Attia et al., 2009). • Behavioral: knowing status  protect oneself and others (e.g., Valdiserri et al., 1999; Summers et al., 2000;Painter, 2001).

10. Hypotheses: • Anticipating that HIV testing could be construed as a prevention behavior, we hypothesized: H1 = A gain-framed HIV testing message will be more persuasive than a loss-framed message among individuals with high experience with HIV testing. H2 = A gain-framed HIV testing message will be more persuasive than a loss-framed message among individuals with high concern about and information needs on HIV/AIDS.

11. Hypotheses... • Moreover, we anticipated that urban residents will have much more direct experience and issue involvement with HIV/AIDS and HIV testing than rural residents. • This assumption was made on two grounds: epidemiological and socio-ecological.

12. Hypotheses... • Epidemiological – in most sub-Saharan African countries, urban prevalence tends to be higher than rural prevalence, except Senegal (UNAIDS, 2009). Ethiopia, according to UNAIDS (2009), the urban-rural ratio was 8:1 • Socio-ecological – in most sub-Saharan African countries urban and rural contexts differ in social/cultural norms, life style, infrastructure, etc.

13. Hypotheses... • So, if urban residents have higher direct experience and issue involvement, H3 = A gain-framed HIV testing message will be more persuasive than a loss-framed message among urbanites rather than ruralites.

14. Methods • Gain- vs. loss-framed brochures were prepared. The messages in each version were organized around four parallel topics – Gain Version: early actions, longer & healthier life, protecting loved ones from the virus, and peace of mind; Loss Version: delayed actions, shorter & unhealthier life, exposing loved ones to the virus, and worry (format adapted from Van‘t Riet et al., 2010). • Brochures distributed to 394 participants (199 Urban: 46.2% male, 53.8% female; 195 Rural: 79% male, 21% female). Through pretest-posttest measures of intention to test for HIV, the relative persuasiveness of gain- and loss-framed messages was determined.

15. Methods... • Univariate General Linear Model (GLM) was employed to determine the main and interaction effects of the independent variables on the outcome variable. • One covariate (Baseline Intention to Test for HIV), four independent variables (Gain- vs. Loss-framing, Experience with HIV Testing, Concern about and Information Needs on HIV/AIDS, and Urbanity vs. Rurality) and three interaction terms (Gain vs. Loss X Experience with HIV Testing, Gain vs. Loss X Concern about and Information Needs on HIV/AIDS, and Gain vs. Loss X Urbanity vs. Rurality) were entered into the model.

16. Results • Urbanity vs. rurality, F(1, 385) = 9.28, p < 0.01, η2 = .02; • Experience with HIV testing F(1, 385) = 17.20, p < 0.001, η2 = .04; • Concern about and information needs on HIV/AIDS, F(1, 385) = 18.97, p < 0.001, η2 = .05, significantly moderated the effects of gain- vs. loss-framing on Intention to Test for HIV.

17. Results... • While urbanites, participants with more experience with HIV testing and those with higher concern about and information needs on HIV/AIDS were motivated by gain-framing, ruralites and those with lower concern about and information needs on HIV/AIDS were motivated by loss-framing. • Both gain-framing and loss-framing led to similar outcomes among individuals with low levels of experience with HIV testing, with a slight advantage for the loss-framed message.

18. 0.5 0.45 0.4 0.35 0.3 Gain 0.25 Loss 0.2 0.15 0.1 0.05 0 Low HIV testing experience High HIV testing experience Results... • Figure 1: Interaction effect of Gain- vs. Loss-framing with Experience with HIV Testing on Intention to Test for HIV

19. 0.18 0.16 0.14 0.12 0.1 Gain Loss 0.08 Intention to test for HIV 0.06 0.04 0.02 0 Rural Urban Results... • Figure 2:Interaction effect of Gain vs. Loss framing with Urbanity vs. Rurality on Intention to Test for HIV

20. Conclusion • Urbanites and ruralites are motivated by differently framed prevention messages. • It was also noted that to the extent recipients are concerned about HIV/AIDS and are familiar with HIV testing, gain-framing is more advantageous, suggesting a possible construal of HIV testing as more of a prevention than a detection behavior in such situations.

21. Implication for intervention • If the findings of this study can be replicated in other contexts, urban and rural contexts may need differently designed (framed) messages.

22. Limitation • The experiment used brochures and thus only literate participants were eligible.

23. Acknowledgements: • KU Leuven • HIV Research Trust (funding the fieldwork part of this study)

24. Thank you!