Data in a post-truth age

Context:

India has been fortunate in inheriting a statistical system from stalwarts like P.C. Mahalanobis and C.R. Rao that has historically demonstrated all three. However, with the growing demand for statistics and increasingly challenging data collection environment, the move by Ministry of Statistics and Programme Implementation (MOSPI) towards developing a National Policy on Official Statistics is most welcome.

Value Edition:

  • The phrase ‘figures don’t lie, but liars figure’ seems to sum up the motif of a post-statistics society. 

Challenges of the policy:

  • It notes increasing data needs, lays down the groundwork for ethical data collection, highlights the importance of data quality and addresses the need for documentation and durable data storage.
  • However, it also remains rooted within the confines of governmental administrative structures and does not directly address the criteria identified by Mr. Spiegelhalter.

In the Indian context, each of these presents great challenge:

Competence

  • Sample surveys, the bedrock of Indian statistical systems, must make explicit choices about who to ask various questions as well as what to ask and how to ask. In a statistical system developed by renowned statisticians and econometricians, it is not surprising that much attention has been directed towards identifying the universe of respondents and sample selection. However, this is only a small part of the challenge. Given the increasing need for statistics in diverse areas, it is important that scholars from many different disciplines be involved.
  • The National Sample Survey (NSS) collects data on occupations and industries of workers. In 2009, it suddenly switched from older codes designed in 1968 to new series of codes developed in 2004. This change makes it difficult to differentiate between farmers and farm managers and shopkeepers and sales managers via occupational codes alone. This leaves out such a large portion of the Indian workforce that it is mind-boggling.

Reliability

  • How surveys are designed and questions are developed has evolved into a science that transcends the skill set usually employed by our statistical systems.
  • The Reserve Bank of India has adopted an inflation-targeting approach that relies on data on inflation expectations of individuals.
  • In a country where ASER (Annual Status of Education Report) surveys repeatedly document extremely low mathematical skills.

Honesty

  • The draft policy as well as many other reports have paid great attention to the fact that data collection is increasingly being done by contractual employees and for-profit organisations. Supervising them and ensuring their honesty remains challenging.

The needs:

  • While improved technology for monitoring fieldwork such as random segment audio recording of interviews and real-time checks for detecting frauds and errors may help increase honesty, there is no substitute for empathy and experience. 
  • Instead of creating a statistical data ecosystem that harnesses the energy of diverse institutions and disciplines in which innovative thinking on data collection and analysis could be undertaken, this tendency towards centralisation may well isolate official statistical systems.
  • This is quite a departure from India’s illustrious history. Mahalanobis was instrumental in setting up both the Indian Statistical Institute (ISI) and what was to become the National Sample Survey Organisation.
  • Most of the early innovations implemented in the NSS emerged from work by academics at the ISI.

Harness diverse energies

  • If we are to revitalise India’s statistical infrastructure, it is vitally important to harness diverse energies from academic and research institutions such as the ISI, the Indian Agricultural Statistics Research Institute, National Council of Applied Economic Research, the Tata Institute of Social Sciences, the International Institute for Population Sciences, the Delhi School of Economics, the Madras Institute of Development Studies and the National Institute of Rural Development and Panchayati Raj.
  • Smaller, technology-savvy private sector organisations may also make important contributions in technology-driven data collection. Around the world, in diverse countries such as China, South Africa, Brazil, the U.K. and the U.S., statistical ecosystems consist of universities, research institutions and government agencies working synergistically.
  • The proposed policy on official statistics is timely and thoughtful but it is also isolationist.
  • Creative thinking about building synergies with diverse communities such as academic and research institutions would strengthen it and reduce the burden on the NSC, leaving it free to devote greater attention to developing quality control parameters and to play an oversight and coordination role.

Conclusion:

  • The draft National Policy on Official Statistics offers a great start for fostering trust in statistics but enhancing its inclusiveness will go a long way towards encouraging competence, reliability and honesty in public statistics.

Source:TH

Leave a Reply