Adjustment of Indian Census Data: A Process for Removal of Bias in Age Reporting

The study aims to smooth out heaping in age reporting of Indian Census data for the last three censuses. Here, data smoothing employing cubic spline interpolation and smoothing spline by leave-one-out cross-validation has been used to obtain smoothed adjusted values of single-year age returns. Smoothed values obtained by cubic spline at various knots and smoothing spline were re-evaluated for the presence of approximation errors by using Mayer's Index, Whipple's Index, and Total Modified Whipple's Index. Smoothed single-year age returns obtained by cubic spline interpolation and those obtained by smoothing spline were free from approximation errors. Therefore, it is recommended that the adjusted data, which is almost unaffected by digit preferences or avoidances, be used as a base population for the estimation of various demographic parameters.