What Women Want
The groundbreaking What Women Want campaign asked women and girls around the world one simple question: What is your one request for quality reproductive and maternal healthcare service?
The 1.2 million open-ended responses White Ribbon Alliance received have been categorized, analyzed, and are now available at your fingertips through the What Women Want Interactive Dashboard, a unique tool that connects women’s voices with the almost infinite possibilities of digital analysis.
The Dashboard offers an unprecedented level of access to women’s demands to allow anyone, anywhere to see exactly what women want when it comes to their healthcare, and to meaningfully respond.
We asked, we listened, and now we’re excited to act – with all of you – in delivering on these demands to make sure that what women want,
is what women get.
Frequently Asked Questions
What is the What Women Want Interactive Dashboard?
An interactive visualization tool that puts the over one million demands from the What Women Want campaign directly at your fingertips.
What makes the What Women Want dashboard unique?
The Dashboard allows you to explore women’s demands and compare countries to one another. Unlike most dashboards, it allows you to see each individual respondent’s exact words. This is because every woman counts.
As well as allowing you to drill down by geography, category and age, the Dashboard relies on cutting-edge natural language processing technology (NLP). Women’s demands have been automatically categorized using a neural network which can learn from a few examples and generalize across, allowing for a deeper understanding of responses than ever before.
Users can filter survey responses to include or exclude specific words such as “blood”, and even see multi-word phrases such as “blood centers needed” (called N-grams).
This ability to easily filter terms and categories provides White Ribbon Alliance team members, public health advocates, policy makers, service implementers, the media, and the public an unprecedented level of detail, while still protecting an individual’s privacy, a critical step forward in transparency in global public health.
How do I use the What Women Want dashboard?
When you first open the Dashboard, you will see a series of graphs showing an overview of women’s demands, broken down by features such as age or geography, as well as a text panel showing a selection of the demands. A tutorial box pops up when the dashboard is opened, explaining how to use the Dashboard controls and interpret the graphs. If you close the tutorial, you can open it again at any time by clicking on the “Show tutorial” box.
The most important panel in the Dashboard is located at the top left and contains the Dashboard controls. You can click on the controls to view responses from a particular country or about a certain topic. The Dashboard also allows you to compare segments of respondents, such as various countries or age groups, with the “Compare to…” tab.
We recommend you go through the step-by-step tutorial, watch the how-to video on Youtube, and explore all the possibilities.
How did you determine the categories in which you put women’s and girls’ responses?
Categorizing a million answers to an open-ended question was far from simple. Using Hamara Swasthya Hamari Awaaz, or “Our Health, Our Voice,” the original campaign from India that reached 143,556 women and girls in 2016, as a starting point, we refined the categories for a global frame. We initially had about 50 categories related to key areas identified for improved outcomes in maternal and reproductive health, but women’s and girl’s answers took us well beyond that.
Mobilizers sent suggestions for new categories as they heard from women; other categories were created or consolidated based on volume and recommendations from technical and country experts.
Early supporters of the What Women Want campaign will notice that our original categories were condensed to 39 for the Dashboard. This is due to partner feedback and suggestions to link key themes for programming and policy making purposes. For example, the Respectful and Dignified Care category now includes requests from the former code Ethical, Lawful, Non-abusive and Secure care.
And, because the Dashboard allows users to search within the response repository for specific keywords and phrases, users now have the power to go beyond White Ribbon Alliance’s chosen categories.
How did you analyze results? Who was involved in the analysis and what were their qualifications?
A core group of reproductive and maternal health and rights experts from White Ribbon Alliance refined the codes and hand-coded approximately 800,000, or 70%, of the responses, often sitting together and discussing women’s and girls’ answers to ensure that everyone was categorizing responses in the same way. To ensure consistency in the scoring, these required extensive internal reviews due to the qualitative nature of the findings.
The remaining 30% were coded with the help of a data engineer who used Google’s pre-trained BERT model and fine-tuned it as a text classifier, resulting in a micro-averaged precision recall score of 83.9%.
How accurate are the categories?
Although White Ribbon Alliance team members followed a rigorously reviewed process for categorizing the survey responses, with more than one million women participating in the survey, it is possible that a small number of responses may have been miscategorized.
In a dataset this large, some variation is expected. From our own quality control process, we are satisfied that more than 98% of responses have been categorized correctly.
As we are always working to improve the usability and accuracy of the Dashboard, we welcome your input and support on categorization errors.
To submit corrections, please use this FORM.