What we learned as technical partner on largest gender study

Lessons learned from collaborating on GMMP, the largest study on gender equality in media.

Another notable change was how the study, which depends on thousands of volunteers, was done. In April 2020, when the research was officially slated to begin, the Covid-19 pandemic had taken hold. Governments imposed lockdowns and physical meetups were out of question. This meant the media monitoring could not go ahead as planned and had to be moved to September.

“The robustness of the platform gave us a relatively smooth analysis process for a project of this scale thanks to the very innovative adaptations made by the tech team,” said Jean Githae, Data Analyst at CfA.

Given the different skill sets and the fact that members work remotely, the first task was to dockerise the platform which made it easy for all developers to add their contributions. It meant changes could be deployed quickly and allow for feedback from the team. On top of that, we took advantage of our weekly meeting to demonstrate our progress for that week.

We then added containerisation which greatly improved our developer experience and easy deployment of the system.

The initial GMMP platform was programmed in Python2.7 and Django 1.7 and CfA had to upgrade the whole system to Python3 and Django 2.2.

One of the most compelling reasons for the upgrade was that support for Python2 stopped in January 2020. The new features in Django also helped us write cleaner code and fix security bugs in the previous versions.

We also implemented frontend changes including GenMap integration using JavaScript.

Telling stories with visualisations

We then created a Python script that extracted datasets from the GMMP report and converted them into the Wazimap data format.

The final bit of the integration was to customise the Wazimap frontend code to work with our newly deployed Wazimap instance. Finally, we customised the frontend to match the GMMP report theme.

How we made our coding sheets accessible

While the data capture has traditionally been done offline, the new platform presented a challenge as different teams have varied resource capabilities and contexts.

To facilitate the virtual monitoring, CfA developed a spreadsheet, which could be downloaded before the monitoring date.

We extracted data from excel sheets and added them to the database using data manipulation libraries such as Pandas. This saved teams from having to manually enter data into the platform.

In addition, CfA developed an API which would upload the responses from the coding sheets to the platform.

Training teams

Leading up to the monitoring day, training sessions were organised for the different regions. The CfA tech team shared video tutorials, which gave a step-by-step guide on the platform features, and trained coordinators from Latin America, the Middle East, North Africa and Asia Pacific on how the platform works.

Our most memorable lessons weren’t technical

While there are many technical lessons the team learned from building and managing this kind of software platform, surprisingly, the most memorable ones turned out to be non-technical.

It’s a team game: By creating a small but self-managed team consisting of a product owner, a data analyst and software engineers, the development of the GMMP platform was smoother than expected. The team was in constant communication over Slack, had weekly meetings where new features were demonstrated and feedback given, and together, next steps were decided and prioritised.

Real user testing is priceless: While the implementation team did their best to anticipate what users behaviour would be like, the actual feedback received from users still managed to surprise. This included features they enjoyed and the ones they didn’t quite like. The final version of the platform used on the monitoring date would have been quite different if it weren’t for the feedback received during real user testing and training.

Support is everything: By making sure the platform supports multiple languages and training material (guides and videos) were readily available, many more participants were eager to use the platform, compared to 2015.

We also made sure the implementation team were generally available over Slack, email or even WhatsApp to address any issues from participants across multiple time zones.

Related posts:

Dallas Cowboys star quarterback Dak Prescott had by far his best game of his young career last Sun. against the New York Giants. The third-year quarterback went 25 of 32 for 405 yards, four…

Sometimes it happens that we come home after work and all we dream about is to lie down on the couch, order a pizza and lie like that for the rest of the day. There’s nothing wrong with that…

leaving only wet sand, quiet and tower-less, on which I build anew once I’m rooted again with courage. Though I can’t make a steel fortress my dominion, I am much more than the fluid material that…