How to read India's social statistics without being misled

This series has quoted many numbers: 26 crore students, 3.1 per cent unemployment, 25.47 billion UPI transactions, 99.82 per cent literacy in Sikkim. Each is true in its own terms, and each can mislead if its terms are ignored. This note gathers the rules for reading them, using only figures already established in this series, so that readers can check any social statistic they meet against the same tests.

First, ask what was counted. UDISE+ counts schools, students, and teachers: it measures access and inputs. ASER tests children at home: it measures learning. The Periodic Labour Force Survey counts people in the labour force who are working or seeking work: it measures unemployment, not job quality. A state can therefore report full enrolment, low unemployment, and weak learning at the same time, without contradiction, because the three surveys count three different things. Confusing enrolment with learning, or employment with good employment, is the most common error in social statistics.

Second, ask who is missing. A literacy benchmark of 95 per cent is treated as “fully literate”, which means 5 per cent of adults are, by definition, outside the headline. A census that excludes Manipur until further declaration is incomplete by design, however good its reasons. A TB-free panchayat certification reflects programme criteria, not a guarantee that no case will appear. Every statistic has an edge, and the people at the edge are usually the people policy most needs to reach.

Third, ask when and by whom. A figure from the PLFS calendar year 2025 published in March 2026 is not a figure for 2026. A screening total of 28.1 crore counts tests, not cures. A government-reported decline in TB incidence is an estimate from a model, not a headcount of cases. And when two government releases give different panchayat counts, the honest response is to report both with their dates, not to pick the larger one. Statistics do not speak for themselves; they speak for the survey that produced them.

None of this means distrusting data. It means using data the way a civil servant must: triangulating sources, reading definitions, and refusing to let a single headline number settle a debate about schools, jobs, or health.