Age heaping in Zimbabwe: Evidence from the 1992, 2002, 2012 and 2022 Censuses

Abstract

Age misreporting is a prevalent issue in population-related surveys, attributable to various factors. This study assesses the quality of age-sex data by measuring the extent of digit preference and avoidance in census age distribution. Data from the 1992, 2002, 2012, and 2022 Zimbabwe censuses were utilised. The analysis applied Whipple’s and Myers’ Blended Indices alongside the United Nations Joint Score to quantify digit preference and evaluate data accuracy across the four censuses. The findings revealed notable digit preference and avoidance for ages ending in 0 and 5 across the first three censuses (1992, 2002, and 2012). However, the 2022 census showed an absence of such patterns, indicating significant improvement in age reporting accuracy. Whipple’s indices demonstrated a consistent decline in age heaping for both sexes across the four censuses, though females consistently recorded higher indices for age heaping and a preference for the digit 0 in the 1992, 2002, and 2012 censuses. Myers’ Blended Index for males decreased significantly, from 32.17 in 1992 to 3.35 in 2022. For females, the index increased from 5.35 in 1992 to 19.7 in 2012 before declining to 17.65 in 2022. These findings indicate that data quality improved more markedly for males compared to females. The UN Joint Scores also demonstrated continuous improvement, declining from 40.0 in 1992 to 34.5 in 2002 and 32.3 in 2012, a 6.4% reduction. Trend analysis over the four decades revealed a gradual improvement in overall data quality, although females consistently lagged in this regard. A notable milestone was achieved in the 2022 census with the transition from Pen-and-Paper Interviewing (PAPI) to ComputerAssisted Personal Interviewing (CAPI), which played a critical role in minimising age heaping in Zimbabwe.