How can Positive Deviance and Big Data be a set up for a Date?

Experiment 4 Data Powered Positive Deviance
P3 Prototype

June 22, 2020
Monique Sternin, Cole Zanetti, Lars Thuesen

This blog post was written by Monique Sternin, co-Founder of the Positive Deviance Approach and Adjunct Assistant Professor, Tufts University, Cole Zanetti, DO, MPH, Positive Deviance Facilitator, Director of Digital Health at Rocky Vista University College of Osteopathic Medicine and Lars Thuesen, Positive Deviance Facilitator, Founder and Change Leader, WIN, the Welfare Improvement Network.

The article was first published on the initiative's Medium Blog.

Doing things different

The partnership of big data and Positive Deviance (PD) seems like a set up for an unusual first date. On one hand you have big data, looked at as an ambient collection of the digital behaviors of people. On the other hand, you have Positive Deviance, which is traditionally a people intensive, community-driven cultural exploration of behaviors that have overcome the odds despite great obstacles. So how do we ensure there are no fights over dinner between big data and Positive Deviance? Like a typical matchmaker you are looking for what they have in common, in this case, it is finding what others commonly fail to see.

Whose problem and how to pick the right problem to work on?

How to successfully combine Positive Deviance and big data really starts with helping define the problem. Often the way challenges are perceived by management and experts seem somehow irrelevant and abstract to communities even though they are supported by big data or some kind of statistical data. How is it that the perceptions of problems differ? Well, management and experts usually see things from a generic and holistic perspective as they should be doing because it is part of their job descriptions. Communities, front line staff and citizens on-the-ground on the other hand are motivated to solve problems that are right in front of their eyes. Specific problems that are near and dear to them. Though the generic, holistic perspectives and the community perspectives are aligned they also differ in their focus.

Let us look at a few examples. In our recent work with Roma communities the problem was defined by experts and management as reducing school drop-out. The Roma communities found school drop-out abstract and more relevant to focus, school attendance, getting homework done, illiteracy and child marriages. Focusing on these issues really got them started and committed. They found kids and teenagers who were champions and students who had graduated successfully. And the problems and solutions were found closer and nearer to the kids and families, maybe because drop- out/ graduation was something in the distant future. Of course school drop-out and the problems that were defined by the Roma community are related, e.g. better attendance and doing homework are likely to increase graduation rates, but it is extremely important to focus on what keeps the community awake at night — not data and statistics that come from the outside — from above.

Another example. When we started working with Positive Deviance in the Danish prisons, the problem seen from the top management point of view was the extremely high numbers of staff absenteeism. When we engaged with guards and other front-line staff in the prisons, they did not find it relevant. It was a managerial problem and sometimes even as an asset/ advantage because staff absenteeism made it possible for other guards to take more shifts and thereby earn extra money. What really interested them was avoiding threats and violence, lack of communication and relationship building, stress etc. So, they began working on these issues and tried to find solutions. Guards and social workers discovered champions that experienced fewer incidences of threats and violence, were excellent in communication and relationship building and came to work almost every day without stress. So, again we worked with their local problems, found positive deviants who both had successful strategies that solved the local problem and the top management problem (staff absenteeism) at the same time.

These examples illustrate why the process of picking the right problems that the communities find relevant is so important in Positive Deviance and thus, Data Powered Positive Deviance. Here (big) data from the outside can help getting started, focus and qualify discussions, but cannot stand alone. (Big) data can help focus on challenges, but the analysis needs to be complemented by local problem discovery processes, on-the-ground expertise and thick data.

A reasonable approach for this to occur successfully is applying the participatory action research frameworks which invites the community to participate as equal partners with researchers and directs them throughout the project. A way to combine the advantages of big data with the “traditional” PD approach is to consider the community from which the data are “harvested” as partners or even co-researchers in this endeavor. Participatory Action Research (PAR) format offers such an opportunity. PAR differs from traditional research in that the research participants act as co-researchers who are involved in the entire decision-making process, from initial conception to data collection and analysis and the development of action oriented feasible recommendations.

To do that the researchers/academics/experts need to access a group of local informants who will help them develop the tools for the PD inquiry and their on-the-ground-expertise and knowledge of communities. They are indispensable /mandatory to achieve the inquiry successfully, i.e.; finding uncommon but successful strategies and behaviors that can be shared and acted upon by the population of interest. It requires high flexibility from the donors and academic researchers to face the inevitability of reframing the problem: to adapt big data to the local context and the socio-cultural background in which the quantitative data have been “harvested” through the filter of local partners’ perception of the problem.

In this optic, 3 fundamental questions need to be addressed when combining Positive Deviance with big data:

  • Who should be involved in accessing, developing and analyzing the data? Who owns the data? How to use the data?
  • How to involve the local partners in gathering the data, identifying potential PD communities, districts, etc. and designing the PD inquiry successfully?
  • What is the role of the researchers in PAR context?

One research project by Ashley Lackovich-Van Gorp illuminates how the PAR can be used in a research framework:

“The Sternins and Pascale note that “experts are ‘answers’ looking for problems to solve. PD practitioners, on the other hand, are community mobilizers. PD allowed me to follow the direction of the participants rather than compel them to follow my direction.”

Another PD research project in Ecuador on the Chaga disease highlight the potential role of the communities in a research project. Although the researchers did not use the PAR, limiting themselves to using participatory tools for their investigations, they stressed the importance of further involvement of the communities for sustainable result in health promotion against the Chaga disease:

“Exploring the potential of participatory health promotion processes to break the cycle of neglect that defines NTD (non-Transmittable Diseases)s remains an important endeavor that could facilitate productive dialogue between the people affected by diseases, decision-makers, donors, policy designers and researchers (Ventura-Garcia et al. 2013). By recreating the stories shared through the dialogical interactions that occurred during this PD research, we aimed to approach the research participants and communities beyond labels such as neglected, poor, at risk, sick or isolated, to be able to share complex ideas about the local knowledge, expertise, culture and expectations, all of them associated with the idea of a healthy living environment. This PD research provided valuable data to frame health promotion strategies associated with the introduction of the HHHL project with context-specific information derived from knowledge and practices currently applied by community members and social dynamics structurally associated with Chagas disease as a disease of poverty. Most importantly, this PD research illustrated an environment of positive collaboration that can potentiate community engagement in the construction of healthy communities”.

What kind of data?

It is also critical to make sure that there is a maintained partnership between the community and researches when defining what to measure.

Qualitative data like stories about PD behaviors (the what and the how) are equally important because they enable peers to feel and understand why (the rationale) behind what and how the PDs are doing what they do. Ensuring an adequate balance between hard quantitative and soft, behavioral, qualitative data is crucial. A PD initiative in partnership with the UN Palestine Gender Lab worked with men and boys that were engaged in the project. Through this collaboration a community scorecard that ensured a community-based monitoring process initiated by themselves with quantitative metrics, e.g. reduction in early and relative marriages, reduction in approval stamps for early marriages from the community leaders (mukhtars), number of imams speaking of avoiding early marriages. And they also defined qualitative metrics relating directly to positive change in community behaviors. E.g. how people gradually got inspiration to change habits, men learning to share household work by visiting families where household work was shared and improving women heritage rights.

In order for PD and big data to work successfully together, it is crucial to strike a balance between quantitative (big) data and qualitative thick data and to let the communities who both own the problems and the PD solutions decide on what success looks like and how to measure and monitor progress. If this is the case, we can make use of the strengths of both and set them up for a lasting and happy partnership.

Positive Deviance, big data and Covid-19

At this time the Coronavirus pandemic has created uncertainty amidst a lack of clear evidence. This is a wicked problem that PD was made to solve. At the same time there is massive data collection going on. The opportunity to combine a big data approach to Positive Deviance is here. The last blog post written by Richard Heeks and Basma Albanna from University of Manchester gives valuable insights on how DPPD might be applied in Covid-19 responses. These big data are extremely relevant and can obviously qualify and focus the discussions and potentially also help discover positively deviant behaviors. Yet, to enable local, community-driven processes to (re-)frame problems so they are meaningful and relevant at the local level, and to determine positive deviant behaviors, it is crucial for people to engage and commit — the communities need to create ownership in the game and opt in.

For reflection you might think about:

  • Whose problem is it? And how to pick the right problem suitable for Positive Deviance?
  • What kind of data are needed and who should be owning and monitoring them?
  • How can big data and Positive Deviance be combined in a useful design to help solve some of the important problems relating to the current Covid-19 crisis and to development challenges?