Published on Development Impact

Having a good excuse: Incentives’ visibility increases the sharing of stigmatized mental health information among refugees. Guest blog by Emma Smith.

This page in:

This is the 10th in this year’s series of posts by PhD students on the job market.

Stigma does not just hold people back from using mental health care – it can even prevent them from learning about it. In low-income countries where institutional capacity is often low, people rely on “social learning” from friends and community members to find out about services and new technologies. However, this can make it hard to learn about stigmatized services. People might hold back information about mental health care, HIV treatment, or food and housing assistance because they worry about “outing” themselves as users of those services. Alternatively, they might withhold information because they worry about insulting the message recipients by insinuating they need these stigmatized services. Indeed, through two experiments I conduct in Jordan, I find evidence that these concerns prevent Syrian refugees from learning about and using mental health services. 

In my job market paper I use a field experiment to investigate how to overcome stigma barriers when promoting mental health services to Syrian refugees in Jordan. Other work by co-authors and I shows that in this population, which lives mainly outside of camps, almost half of adults have symptoms suggesting depression. Separate representative surveys I conduct with collaborators find only 11% of Syrian refugees in Jordan who likely have depression or anxiety are using mental health services, representing one of the most severe treatment gaps across low-income countries. Baseline data from my field experiment shows discriminatory attitudes are common, with 50% of participants worrying others would consider them unreliable if they used mental health services, and 40% saying they would not marry someone who ever used these services. With over 100 million people forcibly displaced globally and mental health problems especially common among refugees, addressing stigma barriers may be a key component to refugees accessing the help they need.

In the main experiment I recruit 847 “senders” who are nominated by a representative sample of Syrian refugees for being well-regarded, community-minded, or good at spreading information. I first ask each “sender” about his or her close social network. I then randomize whether the senders are asked to share mental health content over WhatsApp with the friends they mentioned, totaling 2,665 potential recipients between treatment and control. The WhatsApp content was designed by the International Rescue Committee in collaboration with the refugee community and advertises the Jordan River Foundation’s free phone counseling.

To understand if people hold back mental health information because of social image concerns, I randomly vary the introduction that I ask senders to share. In the first introduction, I test whether senders share more when they have "social cover", by encouraging the senders to disclose to their friends that they are paid to promote the information. While all senders are paid, disclosing this offers a justification for sharing. Suppose that, before, someone who talked about mental health services was thought to very likely be a mental health care user. Now, an observer might think, "Maybe she used the services, or maybe she is just sharing this to get the payment." If the sender is worried about being stigmatized, this can help her feel more comfortable sharing the information. In the second introduction, I test whether senders share less when sending is more revealing that the recipient is in need. I do this by randomizing whether the sender is encouraged to suggest to recipients that they may benefit from the services. If senders do not want to draw attention to their friends’ need, then they might withhold information when the messages seem "targeted".

The main finding is that although people withhold mental health information from their friends, providing “social cover,” by suggesting that people disclose they are paid to promote the information, increases sharing by 37%, or by 6.2 percentage points relative to 16.5 percent (Figure 1). Overall, only 22% of intended recipients receive the mental health information, measured using recipient self-reports, screenshots from senders, and trackable links. There is no significant difference in senders’ sharing of “targeted” versus “non-targeted” message framings. But encouraging senders to disclose that they are being paid to share information, thereby giving them social cover, increases sharing by 37% (6.2 percentage points) and clicks on shared links by 52% (6.5 percentage point increase relative to 12.3 percent).

Figure 1. A good excuse: Identifying openly as a paid promoter offers “social cover” and raises information sharing by 37%.

Figure 1. A good excuse: Identifying openly as a paid promoter offers ?social cover? and raises information sharing by 37%.

Disclosing the excuse of compensation is especially valuable for senders who have used mental health services before. Consistent with the visible compensation providing “social cover” for senders, I find prior users of mental health services react the most. They increase their sharing rates by more than 100% when suggested to disclose that they are compensated.

Who, me? When recipients feel targeted there may be backlash

Recipients could feel singled out by framings that imply that they need mental health services. I use a follow-up experiment to investigate how message framings affect recipients. In the second experiment an enumerator, rather than the sender, contacts recipients and introduces them to the helpline while referencing the sender’s name. This shuts down potential selection in which recipients are contacted, which could be a concern in the main experiment. I find that, when recipients know their friend is a paid promoter, saying their friend thinks they are in need leads to less interest in the phone counseling service. However, recipients react positively to hearing their friend thinks they are in need when recipients do not know the sender is paid. This results in a 9.1 percentage point (13%) gap in helpline interest for the two “targeted” framings, and the divergence is largest for more in-need recipients (43 percentage point, or 94%, gap). Together these results indicate that suggesting recipients are in need is a risky strategy that can interact with other information to generate positive or negative effects. 

Why community-based “senders”?

I also test whether senders have useful local knowledge that they could use to target information. I find that senders in this setting do have private knowledge of who among their friends is in need. Friends who senders identify as “most in need” are 25% more likely than other friends to have depression at baseline. This cannot be explained by observable characteristics, which for example external organizations might use to target, and highlights the possible value added from involving community members. Though senders in the experiment do not always target the most in-need recipients, they use their knowledge to target more carefully when they face higher social image concerns. When senders are not nudged to disclose that they are paid to promote, they share less overall. However, they then target their fewer messages to their more depressed or anxious friends.

Figure 2. Community members can use their knowledge to target the most in-need recipients

Figure 2. Community members can use their knowledge to target the most in-need recipients

Policy Implications

Providing “social cover” may help people circumvent stigma in discussing sensitive topics and enable others to learn about programs that could help them. Nudging promoters to reveal they are paid is one way to do this, and involving community members can also capitalize on local knowledge. The results suggest people who have used these stigmatized programs can be particularly enthusiastic about promoting them but need social cover to do so.


Emma C. Smith is a PhD candidate at Harvard University. Her main research interest is at the intersection of development and behavioral economics, with a focus on refugees and migrants. For more about her research visit . [Twitter handle: @emma_c_smith2]

Join the Conversation

The content of this field is kept private and will not be shown publicly
Remaining characters: 1000