Reporting Error in Age Data of India: A Multivariate Statistical Analysis Approach

This chapter reported the error in age data in India and its analysis in multivariate statistical approach. Age is an important study variable in demography and epidemiological studies. It is a socio-demographic variable related to the host in descriptive studies and also a commonly assessed risk factor in analytical studies. It is generally widely accepted that education enhances the accuracy of age reporting. However, an attempt has been made in the paper to analyse the age reporting error in relation to different socioeconomic, cultural, and developmental factors, in addition to the literacy factor, in order to discover the various unique factors and their commonalities in influencing the variations in the age reporting error in India. Ages reported in censuses in developing countries are subject to errors and bias resulting in uncertainties in population estimates and age distributions. In order to study this kind of detailed statistical analysis, National Family Health Survey (NFHS) data would be appropriate in the sense that this is a huge survey covering all the states of India on a random sample basis. In those reports a measure of age reporting error separately for male and female populations has been given by Myers’ indices. Due to the large number of variables that may or may not be significant, all variables were subjected to a backward regression process at the beginning of the analysis, where a number of variables were deleted as they were found insignificant at a certain level of significance. The remaining variables were subsequently analysed using commonality analysis, which yielded some intriguing results. Nonetheless, male literacy played a substantial influence in enhancing the quality of age statistics. In female part, scheduled tribe population has some influence on the reporting error. Mean household size also played some role in influencing the age reporting error. While studying the pattern between the two periods of time gap of about six years, male data in earlier period of 1992-93 showed illiteracy as the only significant factor influencing age reporting. Other variables were deleted from backward regression process.