The new show in town is the game of data. Running for a while now, it is keeping the audience on edge with its volatile and shifting rules. If one season negated a whole body of data leaving the audience nonplussed, the next brought joy to certain quarters with its data reinterpretation. The latest season has shocked the audience as an entire body of data has gone missing. The article in The Hindu (Editorial page, August 19, 2021), The significance of the ‘there is no data’ answer , which is an incisive review of the latest season of missing data, is enlightening.
On data politics
Data politics is not new. The interconnectedness of power and knowledge and its use by States to control populations has long been expounded by Foucault, Bourdieu and others. Rapid technological innovations in information and communication technologies have further complicated the issue where, through Internet connectivity, both subjects and objects of data are now inextricably intertwined. The spur towards evidence-based policy making or evidence-based budgeting by governments points to the amassing of large, granular level data about citizens by States.
Data-based policymaking or budgeting is meant to facilitate the use of evidence to inform programmatic funding decisions. The goal is to further invest in what works to improve outcomes for citizens. Data-based decisions can redress inter and intra-district inequalities through targeted resource allocations. However, data-based governance pre-supposes the existence of reliable, rigorous and validated data with or without demonstrated impact or outcomes. If governance decisions are to be data centric, there is a need to ensure a good, robust and reliable database.
Data-based policy making
States collect enormous amounts of administrative data. However, these administrative data are often not validated. For example, it is well known that the flow of funds below the block level is often opaque and the data that is submitted by local bodies are generally not validated. The task of trying to match funds with functions at the panchayat level is rather challenging. While there is a critical need to link the databases of various departments, it is not easy as territorial jurisdictions and household-level identifiers are likely to vary from department to department. There is a need to bring some mechanism to homogenise these various data sets with a single identifier; but more importantly, there is a need to validate these data sets through urban local bodies and rural local bodies.
Accurate collection, measurement and interpretation of data are critical for data-based decision making to be successful. However, this is fraught with challenges for as much as data is used, it also gets misused, abused or even manipulated. For instance, absence of data in certain domains does not necessarily indicate better governance. During the novel coronavirus pandemic, some States were not testing enough. Consequently, the data on COVID-19 positive cases were interpreted to seem that some States, especially in South India, were unable to control COVID-19 cases compared to their North Indian counterparts; some with much poorer health indicators as well as infrastructure. In such cases, making resource allocations and decision-making based on data are likely to have adverse impacts.
Similarly, a 2012 academic study on assessing the quality of governance across States had an indicator under a ‘Law and Order’ variable that aimed to measure police behaviour, and the indicator was “Complaints against Police behaviour”. How should such an indicator be measured? What would be an ideal score — low or high? The answer is not simple: it is complex and context-specific and, therefore, should not be interpreted in isolation.
A low score in a poor, backward State does not necessarily indicate that police behaviour is exemplary; it could indicate that people are scared to complain against police behaviour for fear of reprisals. A high score in a State with high literacy and human development index (HDI) can mean that people have enough confidence in the judiciary and the State to complain against police behaviour, thus becoming an indicator of a better quality of governance. Similarly, an issue such as mental health, that comes with enormous social stigma in India, needs careful measurement as higher incidences of mental health (from institutional sources) can indicate better access to institutional care as well as a social context that is less beset with stigma.
Editorial | Data power: On Facebook and BJP
Tamil Nadu’s education data
In the same vein, the recent data on education released by the Union government that shows Tamil Nadu having around 27 educationally backward districts, is baffling. Despite these figures, elsewhere, the same report ranks Tamil Nadu fourth in educational attainment. The literacy rate in Tamil Nadu in Census 2011 was higher at 80.1% compared to the national figure of 73%. While there were inter-district variations in literacy, Dharmapuri district, with a literacy rate lower than the national average, still had 68.5% literacy in 2011. It is problematic to imagine that there has been such a downward slide in the last 10 years as some recent State-level studies have shown further improvements in literacy across districts compared to 2011. Clearly, in this case, the measurement of district-level educational backwardness needs close scrutiny. Such interpretations also highlight the need to supplement the quantitative data with smaller qualitative studies to capture processes, subjectivity, and contextual factors.
As the game of data shows, we are in a data-driven world. While on the one hand, there is a move towards data-based governance and decision-making, on the other, concerned about the ‘statistical vacuum’ due to a number of national statistical bases getting eroded either through delays or data suppression, scholars like Jean Drèze and others have been calling for decentralised systems of data collection processes, with States building their own databases. This requires States to invest heavily in both human and technical infrastructure with built-in quality control measures to ensure an interesting twist to the game of data that is now ongoing.
Kripa Ananth Pur is an associate professor at the Madras Institute of Development Studies, Chennai. The views expressed are personal
