Looking at the rigorous evidence on this is the remit of a new review paper by Lucia Diaz-Martin, Akshara Gopalan, Eleonora Guarnieri, and Seema Jayachandran. It gives us a broad tour of the evidence and helps set up a range of interesting areas for further research.
To start off, Diaz-Martin and co. do a nice job of framing how we might think of the impacts of groups. They focus on thinking of why one would want to use groups instead of targeting individuals and lay out two reasons: (1) groups can be used as a platform for delivery and (2) a program can leverage the interaction among group members.
My naïve prior was that groups as a platform offer a nice opportunity for economies of scale in delivering an intervention. But Diaz-Martin and co. point out that the costing of this (relative to individual delivery) hasn’t really been well done, especially in cases where groups need to be formed from scratch. They also point out that we need to take into account the potential additional costs of participation that are placed on the intended beneficiaries.
So, what makes a group? Here Diaz-Martin and co. follow Gugherty et. al. and include “(i) voluntary membership, (ii) member contribution of time, labor, money or other assets; and (iii) regular face-to-face interaction among members.” And they are going to look at groups in three areas of interventions: livelihoods and financial, health, and adolescents.
Let’s start with livelihood and financial groups. These are a wide range of economic interventions from self help groups to village savings and loans and microcredit groups. They also often (but definitely not always) include some sort of training.
In terms of economic outcomes, the bottom line here seems to be positive but not transformative outcomes. This in line with the existing, multiple, reviews of microfinance and VSLAs (and Diaz-Martin have a nice overview of these). Looking at labor supply, Diaz-Martin and co. find that there are two studies (one VSLA and one SHG) who show increased labor supply. In terms of overall household welfare, the evidence is mixed, with a couple of positive results, some mixed ones and a bunch of null results. When looking at women’s decision-making power, the results are pretty much split with 7 of 15 showing a positive impact. So, this gives us an area for further research: why? As Diaz-Martin and co. note there is a lot of programmatic overlap between the positive and null results so it doesn’t seem to be a clear case of program design.
A bunch of these programs also include significant add-on trainings. These include things like health, but a range of gender relations interventions including family coaching and “gender transformative learning.” Evaluations of these programs tend to measure dimensions of gender norms, acceptability of intimate partner violence as well as things like aspirations and self-efficacy. Here Diaz-Martin and co. look at the evidence and conclude that you need these add-on interventions to be on the more intense side to get positive impact on this family of outcomes. This gives us another area for research: given the expense of these (and questions of scalability): how much is enough? It would be instructive to see a trial of a deeper and lighter option within the same context. The simplest version of this would be to vary the number of sessions, but further testing could look at different components. One other interesting result in this domain (which suggests again that some degree of depth is critical) is that impacts on IPV seem to only manifest when accompanied by improvements in other measures such as social networks or decision-making.
In terms of leveraging the social interactions, this logic is key to microfinance groups and VSLAs as they build the mutual accountability of groups and, indeed, a number of studies show this. Interestingly, two studies look at the provision of credit through groups versus to individuals somewhat directly. One study finds some positive impacts on economic outcomes, while the other shows no repayment differences between the group and individual arms. Taken together, these suggest some further work on how to both enhance the group functioning, as well as how to get it to contribute to other program outcomes would be useful.
Does the group dynamism within the program spill out into other dimensions? Again, evidence is mixed (at best) with positive impacts in one case (on community and political participation in India) but null results in four other countries (all in sub-Saharan Africa). One interesting question that comes up here is whether mixed gender groups will have more of an impact on this and other outcomes or not? The one study that covers this suggests the answer might be no, but further work, perhaps with some focused facilitation (maybe around gender relations), could prove interesting.
On to health groups. Here the results seem to line up with a simpler story. Participants (mostly) show increased awareness and knowledge of the material being conveyed. There is mixed evidence that this leads to increased utilization of formal health care. The evidence on improved practices (e.g. exclusive breastfeeding, attended births) at home, however, is more positive. And these, for the most part, lead to improved health outcomes (e.g. neonatal mortality, maternal mortality), with some of these effects being quite large.
Can we leverage these groups for more collective action? While there is some evidence that this doesn’t work, there are three studies showing that community monitoring health groups can boost women’s engagement in the community. Diaz-Martin and co. point out that “researchers did not investigate to what extent social networks and moral support were crucial mechanisms that made health groups effective in achieving positive outcomes.” As we move forward this is important to understand (and hard to do).
Finally, Diaz-Martin and co. tackle adolescent groups. These tend to be the provision of a safe space for adolescent girls to meet and usually some training in some set of life skills as well as vocational training. Here, the evidence for economic outcomes is generally positive (3 studies). We also have two studies showing positive impacts on education outcomes. A number of studies also show positive impacts on girls’ empowerment. On this, Diaz-Martin and co. raise the question on what happens to these outcomes in the long-run. Indeed, it’s worth looking at how these programs do (or do not) change the overall life trajectory of these young women.
Five studies find evidence that these groups can strengthen some dimensions of girls’ social networks. Finally, in terms of unwanted sex and IPV as well as early marriage and contraception, Diaz-Martin and co. find more mixed evidence.
One big issue with many of the evaluations in this space is that group interventions are rarely just one intervention. The adolescent groups are a great example of this, but they definitely aren’t the only case where this happens (something about a group of people getting together just makes people want to add more interventions). Moving forward, it would be good to do more to unbundle these interventions to figure out where the action is coming from (and yes, it might just be the combination). Of course, at some point, you are just left with the group – and as indicated above, that’s a point that runs through the paper: how and what does the group itself add? On this we definitely need more.