MEL
Measuring Prevalence of Child Protection Issues: Evidence from Recent Practice
AUTHOR:
Tarul Jain

Responses to the questions raised during the webinar:

How do these larger trends (trends from secondary sources) help us measure in our target work areas?

The larger trends (from secondary sources), help us compare the average estimates of the geographies (smaller geographies) we are working in, with the overall averages (to be derived from secondary sources), hence providing us with comparable insights, enabling us to take decisions on which geography should be prioritized while planning interventions. It also helps one to analyze the gap and strategize ways to fill that gap.

How many respondents per cluster (let's say village) should be considered under the Network-scale up method?

Determining the appropriate number of respondents per cluster (e.g., village) for the Network Scale-Up Method (NSUM) requires balancing statistical rigor with practical constraints such as time, budget, and accessibility.

Do you account for double counting/ overlap of networks in case of respondents from the same area? (under the N-SUM method)

In the Network Scale-Up Method (NSUM), double counting networks from the same area (from different respondents) is not a concern. The goal is to determine the average number of people respondents know and use this to estimate the prevalence of an issue in a given area.

For example, if person X in a village knows 6 cases of child marriages and person Y knows 4 (even if some of these cases are the same), we average these numbers to estimate a minimum of 5 child marriage cases in the village. This average can then be converted into a percentage based on the total number of children in the village.

However, avoiding double counting within a single respondent's network is important. If person X knows 6 cases of child marriage, those 6 should all be different. Double counting within one person's network leads to overestimates. To prevent this, statistical models can be applied during the final analysis to correct for any such errors.

Refer to this link to learn more about the statistical models to overcome this challenge- Comparing the Robustness of Simple Network Scale-Up Method (NSUM) Estimators

Wouldn’t the NSUM method give rise to memory bias? Respondents would selectively recall memories that are consistent with their current emotional state. Is there a way to address this?

N-SUM does have its own set of biases and one of the biases would be recall/memory bias, where they only refer to a handful of people from their network. There are some ways to address this/ minimize this-

  • Rigorous training of enumerators, to ensure they prompt respondents to think systematically about different segments of their social network (based on the research question in consideration)
  • Use clear and specific questions to reduce ambiguity and help respondents recall relevant information accurately.
  • Considering a fixed reference period while framing the questions (in the last 1 month/ 1 year) also helps minimize the recall bias.

For those of us who are not familiar with the NSUM research method, it will be helpful to know about the replicability/reproducibility of the method. Does it require some sort of changes when being scaled up in different cultural contexts? Examples in this regard will be helpful.

The Network Scale-Up Method (NSUM) must be adapted to fit the context and culture where it is used. NSUM relies heavily on secondary data sources to validate population questions, so these questions need to be tailored to the specific context. Therefore, one needs to ask questions for which secondary data is available.

For instance, in India, asking "How many people do you know in your village who are disabled " may be relevant, as this data can be sourced from the AWWs. However, in another context, it might be difficult. Therefore, customization is crucial for the effectiveness of this method.

While it's important to include children in such studies, what might one do in case they get uncomfortable while sharing something personal? Any firsthand learnings?

While all research studies that involve data collection with human subjects require ethical approval, special measures should be taken to undertake data collection with sensitive respondents like children, crime victims, or marginalized communities.

Some of these measures include-

  • All the questionnaires should be trauma-informed and approved by the Institutional Review Boards (IRBs).
  • Hiring of enumerators and researchers who have prior experience in conducting data collection with similar respondents.
  • Providing thorough ethics training to the enumerators and researchers before data collection. And using the mock interview/role-play approach during training to understand such situations better.
  • Ensuring the availability of counseling services/trauma-informed care experts during data collection.

On June 3, 2024, just two days after International Child Protection Day, team Athena Infonomics organised an insightful panel discussion on measuring the prevalence of child protection issues worldwide, in collaboration with Global Evaluation Initiative. The session brought together experts from various organizations to share their experiences and insights on measuring the prevalence of child protection issues. The discussion highlighted the importance of accurate and reliable data in designing effective interventions and shaping strategies to combat crimes against children.

Our panelists were: Rohan Singh (Senior Manager, MEL, British Asian Trust), Michael Joseph (Senior Research and Evaluation Specialist, International Justice Mission), Timothy Edgemon (Prevalence Estimation Expert), and Anupama Ramaswamy (Associate Director, MEL, Athena Infonomics) The panel was moderated by Vinay (Principal Consultant, MEL, Athena Infonomics).

The webinar underscored the need for collaboration between researchers, practitioners, and the community to develop and implement innovative approaches that effectively measure the prevalence of child protection issues. By sharing knowledge and experiences, we can work towards generating reliable data to inform policies and interventions that protect children from violence and exploitation. To know more about Athena's work in this sector, reach out to Vinay and Tarul.