利用多水平方差成份模型探讨个人动物蛋白摄入量的影响因素

以中国８省人口、经济与营养健康调查的资料为例，利用多水平模型探讨了在平衡个人因素后省、县、家庭三个水平上的因素对动物蛋白摄入比例的影响。发现家庭因素如平均收入、饮食习惯、对饮食营养的认识程度等是影响动物蛋白摄入的较重要因素；此外，还发现不同省、不同县动物蛋白的摄入存在差异，揭示出摄入动物蛋白比例的增长还可能与各地区饮食文化和种植特点有关。从而可以为改善我国营养结构提供一些较为客观的依据。对于不同模型的比较认为多水平模型可以有效地用于当前许多大型卫生调查中具有层次结构的数据的分析，克服普通回归模型的不足

Studies on the factors affecting the individual''s animal protein intake by a multi-level model

In this article, the influence of factors related to different levels (seg, provincial, county and household) on the proportion of protein intake from animal food was evaluated by using multi level modeling. It was found that after being adjusted by individual factors, the factors related to the household, such as household income, dietary habit and knowledge of nutrition were the most important factors affecting the individual's showed animal protein intake. In addition, the results showed that the difference of animal protein intake in various places depended not only on the increase of household income, but also dietary behavior and agriculture crops in different areas. Comparison of diffiferent models indicates that in analyzing the data from large scale health survey with hierechical structure, multi level modeling is strongly recommended.