IN NUMBERS: Public Attorney’s Office under Acosta

Category: Best data-driven reporting (small and large newsrooms)

Country/area: Philippines

Organisation: Rappler

Organisation size: Small

Publication date: 11 Apr 2019

Credit: Lian Buan

Project description:

The Public Attorney’s Office poured resources to its forensic unit to file cases linking deaths of dengue patients to the dengue vaccine Dengvaxia despite no scientific proof. PAO was criticized for eroding public trust in vaccination. The story analyzes PAO’s annual reports to show the office’s priorities. Data showed that even though PAO was understaffed for the volume of cases, forensics cases shot up, and in the same period, the number of their clients who had to plead to a lesser offense just to dispose of the case increased three-fold.

Impact reached:

President Rodrigo Duterte withdrew funds from the PAO’s forensic unit.

Techniques/technologies used:

The data was culled from PAO’s website. Relevant information from the annual reports drowned in numbers were extracted by keeping focus on what needed to be established in the story – what has the PAO been doing from 2017 to the present when it became busy for its Dengvaxia work? What has it achieved? Data was visualized using DataWrapper.

What was the hardest part of this project?

The hardest part was to sift through all the numbers on the PAO’s annual reports. To PAO’s credit, it reported all kinds of numbers it could, but it is easy to drown in them because each data set stands on its own. We had to make the comparisons and relate the datasets to each other in order for them to make sense and bring the story forward.

What can others learn from this project?

While Open Data is good, the government has a tendency to just dump unprocessed data. For other agencies like PAO, it does the extra step to process them into reports but even then datasets can still be confusing. It is tempting to just lift datasets as they would still be an accurate report, but it is important for journalists to filter relevant datasets and see how these datasets compare to one another in a given context.

Project links: