I don’t know about you, but my mental health suffered during COVID related lockdowns. An exciting recent paper by M Vlassopoulos, A Siddique, T Rahman, D Pakrashi, A Islam, and F Ahmed shows us a remedy for this with evidence from rural Bangladesh.
Vlassopoulos and co. are looking at the impact of four mental health counseling sessions (one about every two weeks) provided to women across 357 villages in Bangladesh. The intervention is a combination of counseling and information provision, focusing on four areas: covid awareness, coping with stress, self and child care, and communication (and the authors go the extra mile to put the modules online, in English, here.) Each session lasts for about 25 minutes and is administered by trained female para-counselors with no prior counselling experience.
Vlassopoulos and co. randomize this treatment amongst households who have an adult woman and a mobile phone (added bonus: they show us these folks look similar to those who have a phone in a representative national household survey). They survey folks with a baseline and at 1 month and 10 months after the intervention.
The effects are pretty remarkable – I haven’t seen so many stars since my youngest beaned me with a wooden ball. Mental health improves -- there is a 22 percentage point drop in the incidence of being stressed and a 21 percentage point drop in being depressed after one month with pretty similar results at 10 months. And this goes through to economic outcomes as well: an index of food insecurity drops by 0.28 standard deviations after one month (when they were emerging from lockdown) and by 0.52 standard deviations by 10 months (when a new lockdown was in place). The treated mothers are spending more time with their kids – working on school work but not playing – in both of the follow-up surveys.
The treated women are also more likely to take COVID precautionary measures at the first follow-up; this index goes up by about 1.2 standard deviations. And, at the second follow-up they were 5.9 percentage points more likely to be vaccinated; measured against a control mean of 21.6 percent this is a sizable effect. As Vlassopoulos and co. note “this impact is quite remarkable given that the counseling sessions did not include any discussion of the benefits of vaccination.”
Subjective well-being increases at the first follow-up: happiness is up 0.23 SD, life satisfaction is up by 0.24 SD and future aspirations is up by 0.39 SD. At the same time there are also improvements in a gender empowerment index (0.13 SD), attitudes towards gender norms (0.17 SD) and attitudes towards intimate partner violence (0.25 SD). This is a paper where the appendices are interesting. Digging into the empowerment index, the improvements were driven by greater decision making power over spouse income and savings, food spending and finances, as well as mobility. Attitudes towards violence shows some intriguing heterogeneity when it gets unpacked: respondents are actually more likely to agree violence is justified if the woman doesn’t take care of the kids, but less likely to agree its justified in case of an argument and because of the cooking.
Vlassopoulos and co. bring some machine learning to tease out the heterogeneity. An interesting pattern seems to emerge: women who were more stressed, older, and lower income at baseline show the higher impacts after one month. However, after 10 months it is the less-stressed, younger, higher income women who are showing the larger effects, maybe because they are more likely to still be practicing techniques taught in the sessions.
Vlassopoulos and co. do some work to try and shed some light on the potential channels. They find that treated women are more likely to likely to follow the advice of the counsellors. Of course, women also got some useful information on COVID and did have someone to talk for those 100 minutes. It gets more interesting when Vlassopoulos and co. dig into economic activities. Treated women were 9.8 percentage points to borrow money from friends and family, and they were 13.7 percentage points (against a control mean of 26.1 percent) to have a new income generating activity. So that’s likely playing a role too.
Overall, this isn’t super-expensive. Cost per participant was around $14 – the bulk of which was salaries for the para-counselors and airtime top-ups (for both participants and counselors). This compares favorably (in the mental health outcome space) to psychosocial interventions. And, as yet another added bonus for those who may wish to adopt, Vlassopoulos and co. lay out the preparation schedule in an appendix.
It’s work like this – both that people were doing it and the results – that help us break out of the COVID doldrums.