Designing for Data Use
This blogpost is based on a talk I gave at the Evaluation Community of India’s recent “EvalFest” at the Indian Habitat Centre in February. It advocates for using “user-centred” approaches for promoting data use, including understanding- and even creating – new influence pathways. It also highlights that different decisions have specific information needs and these should be met by the evaluator.
During last week’s EvalFest, a sticky point emerged…and then emerged again.
What happens when you do everything right – conduct a rigorous evaluation, engage stakeholders at every level, get fantastic results and then….the government completely ignores your data? Speakers from JPAL and Breakthrough, among others, mentioned this situation.
What Can We Learn from Design Research?
I live in Bangalore where I miss a lot of the national level policy dialogue. But I am geographically blessed in that Bangalore is home to many R&D centres for healthcare – such as Phillips Innovation Campus and GE Healthcare. This means that on occasion I get to work with designers and engineers – people in the process of building new healthcare products. I have found they do a lot of research too – they need to understand the context, the market and perceived needs if they are going to design a new health care product for the facility or the home. There are some common tools we share, but there are things they do much better, and I think we should learn from them.
One thing is they are always guided by the needs of the user. User needs is the guiding star in everything they do.
Evaluations are a Decision Making Support Product?
In evaluation (and M&E generally) what are we doing except designing products for decision-makers to make better policy and management decisions? Our data cannot transform anything unless decision-makers use it. We often don’t think of it that way because we like to consider ourselves working towards a higher cause (building the global knowledge base, helping eliminate poverty…). However, by failing to think this way, we are undermining our own potential as a change-maker – as a transformative leader (the role set out for us in the keynote session). We need to make products that are used.
Obviously, management and policy decisions are made in a complex environment with many different influential factors. But we want our data to be the main deciding factor.
What would this take? We need to make sure our data is relevant, and that it’s meeting a perceived need. And it’s not just one decision maker – to implement a new innovation, or take new data into account – decisions need to be made at many levels of the health system; the national level, the state level the district and the health worker level (as other speakers highlighted). We need to make sure we provide the right information for people at each level of the health system.
Creating a User Profile
One trick I have learned from designers is to create a user profile – this is a profile of a person who would use your product. To create a really good product (data!) we need to truly empathize with the decision maker (data user!)– really understand the challenges they face, and the pain points in their day.
These are some of the questions you can consider:
What decisions does he have to make?
What does s/he care about?
How flexible is s/he?
What does their day look like?
Does he have the bandwidth to respond to new information?
What would it take for him/ her to take on something new?
How much autonomy does s/he have?
Who else is involved in making decisions?
What are their fears? What are their vulnerabilities?
What mental or practical barriers are in their way? What obstacles cause friction?
How does your data help him/her make his life and job easier?
But also consider:
How will they benefit personally from adopting your research findings? (materially or emotionally)
How will this help their sphere of influence?
How will it help humans or the planet?
I find creating something like this is a fun activity for a research or project team – make sure everyone understands the activity is confidential and so everyone can feel free to highlight the idiosyncrasies of the decision makers you are targeting.
Data Use Influence Pathways
We also have to recognize that there are different influence pathways to data being used, including interpersonal and collective mechanisms. Not many decision makers have the autonomy to make decisions on their own, even if they have great data to back it up. Understanding the influence pathways means not just understanding the decision-maker – but also the context in which decisions are made.
Sometimes this means we might have to introduce new influence pathways. For example:
In Uttar Pradesh, CRS project staff went to meetings at the PHC level and the district level and talk them through the health worker performance data that emerged from the ReMiND app. This hand-holding helped them establish a habit of data use.
The Ananya project in Bihar instituted monthly meetings at the PHC level to review routine data – to track the facility and health worker performance. The approach adopted in the meetings was non-punitive, focused on problem-solving. By taking such an approach, new insights can be gathered from different stakeholders about how to interpret and make the right decisions from the data.
Different Information Needs for Different Decisions
Of course, we do not just have to understand the user of the data (ie. the decision maker), we also need to understand the data needs of each and every different decision. Even a simple decision like scale up (a key goal of many people presenting their data at EvalFest) has many different angles – do you want the government to adopt the program, to fund the program, to give permission for you to scale up – in each case the information needs will be different.
A big thank you to the EvalFest organisers for having me at the event.
If you liked this…I have written before about user-centred design as a tool for public health here, here and here. I’ve written before about M&E in public health here and here. And I have written about data for decision making here, and routine data here