In his blog, the Nonprofit Chronicles, Marc Gunther writes:
How do feedback loops differ from conventional monitoring and evaluation (M&E)? One attendee told me that feedback loops are the equivalent of diagnosing and treating a disease; a conventional evaluation is more like an autopsy, and thus of limited value to the patient.
This leads us to our question in Tunisia:
Can info culled from big data help us monitor (read: diagnose and treat) in real-time the achievement of Global Goal 16 (read: the patient)?
We’re here working on piloting the localization of achieving Goal 16 (the one on peace, justice, and good governace) alongside the Tunisian National Statistics Institute.
Our aim is to explore how non-traditional sources of data like social media can contribute to the establishment of a baseline – and continued monitoring of our progress.
We hope to find a way in which this data can complement traditional statistics to monitor citizens’ perceptions and attitudes.
Not much work had been done on SDG monitoring through big data – up until now.
Working with UN Global Pulse, we begin with an analysis of the goal’s first target: corruption.
We started by conducting a network analysis on web and social media, looking for content that included relevant corruption-related keywords in Arabic, French, and English.
A content analysis was then done to determine the posts’ relevance and whether or not the tone was positive or negative.
Social media is fascinating because it is ephemeral; however, this also makes it difficult to assess.
In a traditional household survey – such as Tunisia’s 2014 Household Survey on Governance, Peace and Democracy – we get a static snapshot of particular time and place.
How can these two things work together?
We decided to compare the results obtained through our analysis of social media and the household survey. Since this survey included questions regarding citizen perception of corruption, it could be possible to see what crossover there was.
And this was what was so fascinating: For the same timeframe, both the survey and the social media provided the same perception on corruption: 70 percent of people saw corruption as a negative and problematic force in Tunisian society.
Echoing Gunther’s post, we want to argue that social media analyses can be regarded as a sort of heart rate monitor providing a diagnosis based on parameters like variability and activity over time.
From this perspective it can provide a significant value added to complement traditional data sources.
Kamel Abdellaoui presented these early results at the Global Conference on Big Data for Official Statistics in October in Abu Dhabi.
His presentation raised great deal of interest especially given the ongoing discussion on monitoring SDGs through use of big data.
However, there are still challenges to confront such as resistance from different partners, including civil society on use of big data.
Resistance also seems to emanate from privacy issues related to social media, as well as the lack of technical rigor and the precision that traditional household surveys have.
We expect to continue exploring other targets and begin analyzing the results to see how they can complement household survey and administrative data in developing a baseline study on Goal 16 in Tunisia.
We will also continue to explore other social media analysis tools.
From what we have seen until now, it is quite clear that for statistical purposes, some tools are more suitable than others.
The broader question is this:
If we can measure this in real-time, for what other Global Goals could this approach also work?