Avsar Gulay, Ham Roger, Tannous W Kathy
School of Business, Western Sydney University, Parramatta, NSW 2150, Australia.
Int J Environ Res Public Health. 2017 Feb 10;14(2):177. doi: 10.3390/ijerph14020177.
The increasing health costs of and the risks factors associated with obesity are well documented. From this perspective, it is important that the propensity of individuals towards obesity is analyzed. This paper uses longitudinal data from the Household Income and Labour Dynamics in Australia (HILDA) Survey for 2005 to 2010 to model those variables which condition the probability of being obese. The model estimated is a random effects generalized ordered probit, which exploits two sources of heterogeneity; the individual heterogeneity of panel data models and heterogeneity across body mass index (BMI) categories. The latter is associated with non-parallel thresholds in the generalized ordered model, where the thresholds are functions of the conditioning variables, which comprise economic, social, and demographic and lifestyle variables. To control for potential predisposition to obesity, personality traits augment the empirical model. The results support the view that the probability of obesity is significantly determined by the conditioning variables. Particularly, personality is found to be important and these outcomes reinforce other work examining personality and obesity.
肥胖相关的健康成本不断增加以及风险因素已有充分记录。从这个角度来看，分析个体肥胖倾向很重要。本文使用澳大利亚家庭收入与劳动力动态调查（HILDA）2005年至2010年的纵向数据，对影响肥胖概率的变量进行建模。所估计的模型是一个随机效应广义有序概率单位模型，它利用了两种异质性来源：面板数据模型的个体异质性以及体重指数（BMI）类别之间的异质性。后者与广义有序模型中的非平行阈值相关，其中阈值是条件变量的函数，这些条件变量包括经济、社会、人口统计和生活方式变量。为了控制肥胖的潜在易感性，人格特质被纳入实证模型。结果支持了这样一种观点，即肥胖概率由条件变量显著决定。特别是，发现人格很重要，这些结果强化了其他关于人格与肥胖的研究。
