Remotely researching the world’s most marginalised communities

  • Latin America, South Sudan, Uganda, Philippines, Rwanda, UK

Objective

Qualitative data is typically collected from marginalised communities via face-to-face methodologies. However, the global pandemic, and its social distancing demands, has left the International Development sector unable to understand those most in need of support.

For those with a good level of digital literacy, a smartphone, access to the internet and enough data, qualitative data can be easily remotely collected via app and web based solutions. This accounts for 45% of the global population. However, for the remaining 55%, reaching and understanding their needs was a big, unsolved problem. The techniques deployed to reach the 55% were not effective or secure: phone call interviews recorded as a ‘whole interview single MP3 file’ using a second mobile phone, resulting in 100s of 30-60 min indecipherable audio files sent via insecure emails, creating a laborious, slow analysis process comprising listening to all interviews in their entirety before coding, tagging, or insight extraction could begin.

In partnership with CARE International, we decided to tackle this problem, and create a research approach that can remotely reach and qualitatively understand marginalised communities - those without good level of digital literacy, a smartphone and access to the internet, and rapidly analyse the results enabling fast, accurate decision making.

Exclusion drivers

Respondents had no /low internet access, no access to smartphones, no / low digital literacy.

Confirm that a respondent is in a safe space, without visual aid created safety risks for the respondent - particularly when talking about sensitive topics such as SRH or GBV.

Audio files are typically laborious to analyse, which slows down decision making in often urgent situations.

The product: Fatima

To solve this urgent problem, we built Fatima, a qualitative data collection tool designed to remotely collect qual data from marginalised demographics. It collects data via a simple phone call from any person with a basic feature phone, no matter their location, at no cost to them. No internet, smartphone, training or data credit required.

The respondent receives a phone call at a pre-agreed time from a pre-agreed number. The interview begins by confirming respondent identity via a pre-agreed password. The interviewer then asks a series of questions to confirm that the respondent is in a safe, quiet space.

Consent is then collected via a segmented step-by step-process of the interviewer sharing the consent, and the respondent correctly answering a set of questions to demonstrate they fully understand what they are consenting to. If the respondent is under 18, guardian consent will also be captured following the same process.

To ascertain respondent safety we created a safe space assessment step, comprising a number of questions the interviewers ask the respondent to better understand their environment. Throughout the interview, the interviewer re-confirms their safe space status.

The call is recorded and the interview is automatically uploaded to a secure data platform, where the findings are segmented, transcribed, translated, and tagged — filterable by respondent, question, and topic.

To accelerate analysis and decision making, Fatima also offers a brand new Machine Learning function that programmatically analyses the data, and provides a set of scores, which act as complimentary data sets that aren’t usually extracted using manual analysis. These data sets fall into 2 categories:

Survey insights - a set of scores highlighting how the research is performing including respondent fatigue, and question resonance. This data enables researchers to create better, and more relevant surveys, results in more impactful results.

Respondents insights - a set of scores providing additional insight into the respondents / responses. These included quantified scores for ‘knowledge’ and ‘attitudes’ on any subject allowing users to quickly understand how knowledgeable respondents are on a particular subject, and if they have positive or negative attitudes towards this / these subjects. This data can be segmented by demographic.

Fatima Machine Learning is a ground-breaking new offering to complement and supercharge the manual analysis process, and is currently only available in English, but has the potential to roll out into any language.

Impact

Fatima is a world-first in its seamless ability to reach and understand those without access to the internet or a smartphone, and programmatically segment and visualise the findings. Since launching August 2020, Fatima is now live and providing rapid access to hard to reach communities in Uganda, Philippines, South Sudan, Latin America and the UK.

More information about Fatima can be found here.

We've launched a newsletter:Subscribe