Multilevel logistical regression models for the association between the presence of depressive symptoms and contextual variables: income inequality and socioeconomic deprivation‡.

‡ Model 1: Random intercept at locality level; random intercepts for localities and municipalities; random intercepts for localities, municipalities and states.†OR = Odds ratio; CI 95%: 95% Confidence Interval.¥Model 2 versus Model 1.£Model 3 versus Model 2.*pLRT: Likelihood Ratio Test.AIC: Aikaike Information Criterion.n.s.: non-significant.+Mexico is organized politically and administratively into states, which are composed of municipalities, the smallest administrative units in the country. In turn, municipalities are composed of localities, which are defined as “any place occupied by one or more buildings utilized as housing - inhabited or not – and which are designated by a name recognized by law or custom.” Rural localities are those with less than 2500 inhabitants and which are not municipal centers. Source: National Geostatistical Framework 2010. National Institute of Statistics, Geography and Data Processing.Multilevel logistical regression models for the association between the presence of depressive symptoms and contextual variables: income inequality and socioeconomic deprivation‡.

‡ 模型1：仅包含居民点层面的随机截距（random intercept）；包含居民点与市级行政单位的随机截距；包含居民点、市级与州级行政单位的随机截距。† OR=比值比（Odds Ratio）；CI 95%：95%置信区间（Confidence Interval）。¥ 模型2与模型1的比较。£ 模型3与模型2的比较。*pLRT：似然比检验（Likelihood Ratio Test）。AIC：赤池信息准则（Akaike Information Criterion）。n.s.：无统计学显著性。+ 墨西哥在政治与行政上划分为州，州下设市级行政单位——该国最小的行政单元；市级行政单位又下辖居民点，居民点被定义为“一处或多处用作住宅（无论是否有人居住）的建筑聚集场所，且拥有法律或习俗认可的官方名称”。农村居民点指人口不足2500人且非市级行政中心的区域。数据来源：《2010年全国地理统计框架》，国家统计、地理与数据处理研究所。针对抑郁症状检出情况与收入不平等、社会经济剥夺两类背景变量之间关联的多水平logistic回归模型（multilevel logistic regression models）‡。