Appraising praise: experimental evidence on positive framing and demand for health services

ABSTRACT

Standard economic models of human behaviour take the view that non-informative elements of communication (e.g. tone) do not affect behaviour. Casual observation of consumer and producer behaviour, as well as descriptive evidence, suggests this may be an overly strong assumption in many contexts, including healthcare. For example, 59% of female respondents in the 2011 Ethiopia Demographic and Health Survey reported that ‘rude attitude of health provider’ was a major problem that prevented them from seeking medical advice and treatment. Yet there is only a small body of evidence on the causal effects of non-informative elements of communication from real-world settings. We conducted a field experiment with over 800 HIV+ female sex workers (FSWs) in Ethiopia testing the effects of providing of Praise Message phone calls on retention in antiretroviral (ART) care and adherence to ART medication. We find mixed evidence on the effects of Praise Messages, suggesting further investigation into the effects of praise or other non-informative communication on health behaviour.

KEYWORDS:

• Adherence
• female sex workers
• framing
• HIV/AIDS
• retention

JEL CLASSIFICATION:

• D91
• I12
• I15

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ At least eight other DHS ask respondents about provider rudeness: Albania (2008), Burundi (2010), Gabon (2012), Lesotho (2009), Nepal (2006), Nigeria (2008), Nigeria (2013) and Zambia (2014), with 63%, 25%, 36%, 41%, 48%, 11%, 16% and 33% of female respondents aged 15–49, respectively, reporting provider rudeness is a ‘big problem’ in seeking healthcare.

² An early article on this topic is McNeil et al. (1982), which examines the effects of framing treatment benefits in terms of the probability of living or the probability of dying.

³ There is a somewhat related body of literature on media effects and the effects of positive and negative emotional appeals on other economic outcomes (e.g. DellaVigna and Kaplan 2007; Martin and Yurukoglu 2017).

⁴ Chaiyachati et al. (2014) reviewed adherence interventions and found that 26 studies (out of 124 total studies) examined the effects of a‘treatment supporter’, with roughly 2/3rds finding apositive result for at least one outcome measure. The PM is amuch shorter (and lower cost) intervention than ‘treatment supporter’ and focuses entirely on providing praise in adiscrete event. “In its place, please substitute: ”Chaiyachati et al. (2014) reviewed adherence interventions and found that 26 studies (out of 124 total studies) examined the effects of a ‘treatment supporter’, with roughly 2/3rds finding a positive result for at least one outcome measure. Among the 17 randomized controlled trials with low risk of bias reviewed in Nieuwlaat et al. (2014), several evaluated complex bundles of interventions that often included treatment supporters. The PM is a much shorter (and lower cost) intervention than ‘treatment supporter’ and focuses entirely on providing praise in a discrete event.

⁵ In a discrete choice experiment in Zambia, Hanson et al. (2005) found that provider rudeness was associated with reduced willingness-to-pay.

⁶ Mauer and Harris examine the effect of trust in vaccines on influenza vaccine use.

⁷ Lim, Lee, and Hwang (2011) and Hollard and Sene (2016) examine the effect of social capital, as proxied by trust in ‘people’ and in ‘neighbors’, on use of healthcare, doctor absenteeism, waiting times, and bribes.

⁸ Kovacs, Lagarde, and Cairns (2019) provide evidence on the association between self-reported trust in the provider and an experimental measure of trust.

⁹ Non-financial incentives have been shown to be effective at increasing sales of preventive health inputs (Ashraf, Bandiera, and Jack 2014), further reinforcing the evidence base for the hypothesis that the PM – which links the desired behaviour (i.e. appointment adherence) with a non-financial incentive/non-economic reward (i.e. the praise) – may increase individual investment in health inputs.

¹⁰ MULU is a PEPFAR/USAID-funded HIV prevention programme that provides services to female sex workers in 169 towns across Ethiopia. The 25 participating DICs are among the highest volume DICs.

¹¹ Kebede et al. (2015) demonstrated that ART patients in Ethiopia were willing to use their cell phones to receive medication reminders, suggesting that patients were willing to receive PM calls. Consistent with this claim, we found that virtually all of the participants found eligible to join the study agreed to do so.

¹² There is a large public health literature indicating that ex ante text message reminders can be effective at increasing ART adherence and use of HIV/AIDS services (e.g. Lester et al. 2010; Pop-Eleches et al. 2011; Bigna et al. 2014; Finitsis, Pellowski, and Johnson 2014; Mills et al. 2014; Garofalo et al. 2016; Mbuagbaw et al. 2015).

Additional information

Funding

This work was supported by the United States Agency for International Development [N/A].