Research for stronger health systems during and after crisis

Social Network Analysis: Encouraging Old Dogs to Learn New Tricks

Friday, 14 Mar 2014

How ReBUILD has used this technique to enable government in Uganda to understand more about differences in district performance.

Freddie Ssengooba  MBChB, MPH, PhD.

Associate Professor, Health Policy and Systems Management, Makerere University School of Public Health, Kampala Uganda. 

Who says that you cannot teach old dogs new tricks? I now believe that this famous idiom should be revised or extended. Out of necessity, old dogs can learn new tricks! We can teach old dogs new tricks if we create high-powered “necessity” for the new trick!  In the summer of last year, I was part of a group of mature learners being introduced to Social Network Analysis (SNA) at the prestigious Manchester School of Social Sciences (MSSS) in the UK. The necessity for me was to analyze and make sense of our research data that captures the relationships between health agencies in conflict affected districts of northern Uganda.  This necessity was so strong that I enrolled on a 5-day summer course at MSSS. Previously I had tried an online course (Coursera) but poor internet was a major constraint. The 5-day summer course at MSSS was one of the most memorable and impactful short trainings I have had in years.  

Unlike many health researchers that are wedded to particular research methods or traditions, researchers like myself who are interested in health systems and policy research field are constantly shopping for research methods that can effectively address the research questions that we need answers to. Many times we are interested in questions that explain variance in the performance of health systems – a situation that makes traditional methods that handle variance of study objects inadequate.   When the research question is about how the pattern of relationships between different entities correlate with outcomes that require joint efforts, social network analysis provides unique additional insights that variance analysis methods do not address well.  The study of collaboration among actors is vital to many public health goals. For such studies, the interactions among actors are more important. By collecting attributes of each entity separately, most quantitative research methods fail to capture well the relatedness or collaborative relationships. The need to preserve the relational nature of collaborative public health enterprise for common public health goals, created the necessity for my research team to embrace SNA. SNA is not commonly taught in standard courses that prepare public health professionals, yet it has a lot of useful applications.

Our study under the ReBUILD Research Consortium focuses on Aid Architecture. We are studying how aid agencies interact among themselves and with local government and health providers in the post conflict districts of Amuria, Gulu and Kitgum in northern Uganda. The questions we were asking are; “Which organizations are interacting to support health workforce developments and the delivery of maternal and HIV treatment services? What resources do these agencies exchange? How satisfied are the partners with their relationships?

When I applied to MSSR for enrollment on the Social Network Analysis (SNA) course in June 2013, I wanted to pick up basic skills to analyze our data. My responsibility was to ensure that we have the necessary skills to conduct our research. The consortium has an objective to build the capacity for health systems research in partner institutions. To this effect, the consortium has a competitive Capacity Development Fund to which I was the first successful applicant.

The course exceeded my expectations. It exposed me to sophisticated approaches and uses beyond the basics that I needed. There were three classes running concurrently; the Introductory Group, the Intermediate and the Advanced. Ours was the introductory group with 12 trainees – mostly students on PhD and Masters programmes in the UK. I was among the few persons who were not students.   I also had a dataset from my research and this made the training more applicable to my purpose. In retrospect, this was one of the reasons this course was most remarkable for me. Other attendees at the course were hoping to use the skills in future. If I had to advise the course organizers about one thing, it would be to encourage future participants with research data to come with it and get  support to analyze it during the training.  

Nevertheless, it was exciting to interact with the facilitators; among them Prof Martin Everett, one of the gurus in SNA. He has written a lot of books, papers and he co-developed one of theanalysis software packages for SNA. The Mitchell Research Centre at MSSS has interesting and regular SNA seminars and working papers that I have found very useful for my ongoing learning.

As I prepare this blog, nine months have elapsed. Establishing a “theory of change” to link my training to policy or practice impacts is a little easier given the benefit of time and hindsight. This is a point research funders should not lose sight of.

First, since the training in July 2013, we have been able to analyze our data, validate our findings in the study districts, shared findings with the Government officials and submitted a manuscript for publication. We are now using the skills to explore more hypotheses with our data. 

Second, I have taken advantage of my teaching (organizational research methods) of graduate MPH students to introduce SNA. Although the one-day orientation to SNA is unable to cover much, my students have appreciated the exposure to a new world of scientific research enquiry.   In future, when the necessity to use the SNA method arises, they will know where to shop for  skills or find potential collaborators. At our annual consortium meeting in Liverpool last September, I was able to demonstrate the utility of SNA skills to all members of consortium at one of the sessions.

The third and most pleasant surprise came from the Ugandan Ministry of Health. When we shared our preliminary findings at the October (2013) meeting of the Sector Monitoring, Evaluation and Research (SMER) Technical Working Group, their interest was more about the method than the findings of the study. The Ministry was grappling with two problems that our study seemed to provide some answers to: 1) weak effective coordination of NGOs at the district level and 2) lack of central information about what the different actors were doing at the district level. Although our study provided preliminary findings about these two issues, the Working Group was impressed more about the use of SNA in generating visual graphs that were able to capture the density of actors and relationships among them in the three study districts. In particular, they recommended that SNA should be used by the Health Ministry to understand why some districts are performing well or poorly on the District League Table that the Ministry publishes every year. In their recommendation, SNA was seen as a useful approach to generate explanatory insights or mechanisms as to why some post-conflict districts like Gulu are always performing well on the league table while many districts not affected by conflict were at the bottom. Among others factors, our findings showed a higher density (3 times more) of agencies that were supporting health service delivery in Gulu relative to other study districts.  As we work with the Ministry to explore how to integrate SNA into the District League Table, I cannot help but think this chain of events represent the most enduring impact of a5-day training in SNA at Manchester!

Follow this link to free SNA book: