中国婴儿死亡率影响因素的岭回归分析

Ridge Regression Analysis of Influencing Factors of Infant Mortality Rate in China

摘要: 目的：寻找中国婴儿死亡率的影响因素，分析人均GDP对婴儿死亡率的影响及其关系转变。方法：将1961~2018年分成6个阶段，以各阶段的婴儿死亡率为因变量、各经济社会指标为自变量，进行阶段性岭回归分析，回归系数t检验差异有统计学意义的自变量为婴儿死亡率的影响因素，判断人均GDP对婴儿死亡率有无影响确定二者关系的转变，比较标准化回归系数绝对值大小判断各自变量对因变量的影响程度。结果：6个阶段的岭回归方程F检验差异均有统计学意义(P < 0.05)，1961~1970年人均GDP、2000~2010年人均当前卫生支出的回归系数t检验差异均无统计学意义(P > 0.05)，从1970年起5个阶段的人均GDP、2010~2018年人均当前卫生支出的回归系数t检验差异均有统计学意义(P < 0.05)，其他纳入回归方程自变量的回归系数t检验差异均有统计学意义(P < 0.05)。从1961~1970年起6个阶段按标准化回归系数绝对值最大的自变量依次为农业用地、人口密度、中小学女生与男生的入学比例(1980~1990年)、能源使用量、高等院校入学率、中小学女生与男生的入学比例(2010~2018年)。结论：不同阶段婴儿死亡率影响因素各有差别。人均GDP与婴儿死亡率关系确实存在从无影响向有影响的转变。人均GDP并非一直是中国婴儿死亡率的影响因素，且不是首要影响因素。不能放大也不能忽略经济因素的作用。

Abstract: Objective: To find out the influencing factors of infant mortality rate in China, and analyze the im-pact of per capita GDP on infant mortality rate and its relationship transformation. Methods: It was divided into six stages from 1961 to 2018. With infant mortality rate at each stage as the dependent variable and economic and social indicators as the independent variable, the stage ridge regression analysis was carried out. The independent variable with statistically significant difference in re-gression coefficient t test was considered as the influencing factor of infant mortality rate. The transformation of the relationship between per capita GDP and infant mortality rate was deter-mined by judging whether per capita GDP has or has no impact on infant mortality rate. The degree of influence of each variable on the dependent variable was judged by comparing the absolute value of standardized regression coefficient. Results: The difference of ridge regression equation F test was statistically significant in six stages (P < 0.05). The difference of regression coefficient t test of per capita GDP from 1961 to 1970 and per capita current health expenditure from 2000 to 2010 were not statistically significant (P > 0.05). The difference of regression coefficient t test of per capi-ta GDP from 1970 and per capita current health expenditure from 2010 to 2018 were statistically significant (P < 0.05). The t-test difference of other regression coefficients included in the inde-pendent variables of the regression equation was statistically significant (P < 0.05). From 1961 to 1970, the independent variables with the largest absolute value of standardized regression coeffi-cient in the six stages were agricultural land, population density, enrolment ratio of girls and boys in primary and secondary schools (1980~1990), energy use, enrolment ratio of colleges and univer-sities, enrolment ratio of girls and boys in primary and secondary schools (2010~2018). Conclusions: The influencing factors of infant mortality rate at different stages are different. There is indeed a shift in the relationship between per capita GDP and infant mortality rare from non impact to im-pact. GDP per capita is not always the influencing factor of infant mortality rate in China, and it is not the primary influencing factor. The role of economic factors cannot be amplified or ignored.