la diaria datos - open/collaborative exit polls

2020

la diaria datos – open/collaborative exit polls

Category: Innovation (small and large newsrooms)

Country/area: Uruguay

Organisation: la diaria

Organisation size: Big

Publication date: 27/10/2019

Credit: Natalia da Silva, Juan José Goyeneche, Ana Coimbra, Lucas Silva, Ana Tuduri

Project description:

2019 elections in Uruguay were particularly unpredictable with new political parties, eroded trust in polls due to large differences with results in previous elections, and no clear favourites for winning with very small margins in polls leading to the election.

In that context, la diaria (a cooperative owned, subscription-based newspaper) decided to leverage it’s heavily active community of subscribers to crowdsource real-time exit poll data through an app, gathering precise, high quality data, that thanks to a collaboration with the University of the Republic (UDELAR) allowed to create an indepent exit poll and live, real-time coverage of the election night.

Impact reached:

First and foremost, this allowed la diaria to have first hand, independent access to real-time data of the highest quality, in a context of distrust of traditional exit polls. And to be perfectly clear about the quality, this crowdsourced, technology-aided effort was beyond successful, achieving the closest results to the actual vote compared to evey other exit poll (Link 3).

It also enabled deeper analysis of voter behaviour, leading to insights such as detecting that the left-wing candidate was able to capture 27% of the far-right vote in the runoff vote. A larger percentage than the one captured from center-right and right wing parties (Link 5).

The availability of real-time data also allowed to form partnerships with other media to explore new formats. The exit polls results were available through the newspaper’s website and social media presence, but also through live TV coverage in partnership with TevéCiudad (local government TV channel) and live radio coverage through the University’s UniRadio. These transmissions included political commentary from la diaria’s journalists, and professors from the University’s Faculty of Information and Communication (Link 7).

For la diaria, the biggest impact might have been the real-world proof of the potential of it’s community of subscribers and allied organizations, to create powerful and impactful reporting. Data and technology have been part of the newspaper strategy for many years, as well as community, which has been fundamental for sustaining the project, and a source of much of the quality content published by la diaria. But for the first time, this success has proven that the community, aided by technology can help achieve a feat as surprising as beating opinion polling companies at their own game.

Techniques/technologies used:

The statistic sample was randomly selected among electoral circuits, it includes three strata shaped on available open data from 2014 elections; one with 3.000 circuits where the incumbent party had majority, one with 3.000 circuits where the sum of opposition parties had majority, and one with 1.000 circuits with new voters.

On every one of the 150 electoral circuits randomly drawn from the sample, a volunteer la diaria subscriber was present when the ballot box was opened, and uploaded online the first 50 opened votes in reral-time, through a purpose built app developed by la diaria’s IT department. This level of coordination with volunteers and the drafting among subscribers was only possible thanks to the open source CRM also developed by the IT department and the in-house call center.

All data is available as open data (Links 1,7), including data by circuit (circuitosLD.csv), individual votes (votosLD.csv), sample information (Muestra_ld_con_st_y_orden.csv) and two files with the partial results obtained through the evening, (porcentajeAcumLD.txt y totAcumLD.txt). The Rstudio project (LD.Rproj) and Rmd (reporteLD.Rmd) are also open and results fully replicable using R and Rstudio.

What was the hardest part of this project?

This project couldn’t have existed without a vibrant community that has been growing, sharing and learning over 14 years since the foundation of la diaria (through community crowdfunding, incidentally). This community and this project have allowed for a new paradigm of collective building of the newspaper and reporting, evolving beyond financial support, beyond being a prime source of experts for consultation in the newsroom, beyond generating a substantial percentage of the published content, and into allowing the paper to punch way above its weight, with a country-wide, real time collective effort involving over 200 people, that would have been logistically and financially impossible without the combined support of community and technology.

We believe it should be selected because it proves beyond any doubt that not only independent journalism is possible, but it can also aim much higher than traditional reporting, and develop complex, logistically challenging initiatives that allow complete freedom from corporate sources of information, through strong communities, open source principles, and collaboration with key partners.

What can others learn from this project?

Hopefully, this helps demonstrate that innovation in the newsroom does not come exclusively from the use of sophisticated technologies, but also from exploiting the power of collaboration with the same people that are interested in what journalism has to say. In this particular case, innovation also comes to disrupt a long-standing status quo, where corporate players have had dominance for decades, with an absolutely opaque operation.

This has also been possible with very limited resources, which we hope positively impacts it’s chances of replicability in other countries and contexts.

And most importantly for replicability, all components of this project are open sourced and available for others to use or improve on. From the software used to collect data, and the analysis done with it, to the CRM used to coordinate volunteers.