Fig. 2

Correspondence between empirical data and the output of the calibrated models over multiple calendar years. The fitted model’s estimates of population dynamics correspond closely to multiple alternative datasets from India. A representative sample of model fits to Indian datasets (listed in Table 1) are illustrated here, including (a) fertility, (b) mortality, (c) educational attainment, (d) migration, and (e) population size fits. Model fits to the complete datasets for all population cohorts are provided in Additional file 1: Figures AF1–S4. Here, a random subsample of fits are provided as illustrations of how the model was fitted to disaggregated data for various calendar years and birth cohorts, where education category (“ed category”) for each cohort was categorized as 0: no education, 1: >0–6 years education, 2: >6–12 years, and 3: >12 years. Gray shaded areas are results of 10,000 repeated samples from the posterior joint distribution of the fitted model (Fig. 1c), with samples from the interquartile range as black lines and data displayed as dashed blue lines or circles reflecting the 95 % confidence intervals of the input datasets. The solid vertical bar in the population size figure (e) reflects the point at which UN data transition from recorded values to model-based estimates