中国中部六省预期寿命时序加密估算研究

年龄组死亡率是利用年龄分组人口数据计算预期寿命的关键参数,而非采样年份的统计年鉴中年龄分组死亡率缺失导致无法计算预期寿命。针对该问题,本文将人口普查数据与统计年鉴数据融合,首次提出一种基于拉格朗日插值的中国省级预期寿命时间序列加强密集度（时序加密）的算法,以解决非采样（即未进行人口普查或1%人口抽样调查）年份省级预期寿命的估算问题。以中国中部六省为例,在所选取年份省级预期寿命估算实验中,绝对精度表明年龄分组人口比例线性插值计算的精度明显高于人口比例抛物线插值和直接插值算法的精度,故为推荐算法。本研究为高时间分辨率下省级预期寿命值的获取提供了一个新的可行思路,为分省较精确地进行预期寿命趋势分析奠定基础。

Time-series estimation of provincial life expectancy in China: A case study of six provinces in central China

Age-specific mortality rate is a key parameter to estimate life expectancy based on age-group population. However, it is impossible to estimate life expectancy in non-sampling years (i.e., without census or 1% population sampling survey) due to the loss of age-specific mortality rate in statistical yearbooks. To estimate time-series life expectancy at China's provincial level in the non-sampling years, this paper firstly proposes a time-series estimation algorithm based on Lagrange interpolation by combining census data with population data from statistical yearbooks. We selected six provinces in central China as study areas and estimated provincial time-series life expectancy in non-sampling years by four algorithms, i.e., linear interpolation and quadratic polynomial interpolation in direct and indirect ways. And the absolute accuracy of estimating time-series life expectancy indicates that the accuracy of linear interpolation for proportions of population by age group (i.e. indirect method) is significantly higher than that of quadratic polynomial interpolation (i.e. indirect method) and time-series interpolation of life expectancy (i.e. direct method) based on two methods, which is proposed as a recommendation algorithm. This study provides a new and feasible way to acquire the provincial time-series life expectancy in non-sampling years, which lays a foundation for the more accurate trend analysis of life expectancy in China.