ABSTRACT

In this research, we evaluated the relationship between obesity rates and altitude using a cross-county study design. We applied a geographically weighted regression (GWR) to examine the spatially varying association between adult obesity rates and altitude after adjusting for four predictor variables including physical activity. A significant negative relationship between altitude and adult obesity rates were found in the GWR model. Our GWR model fitted the data better than OLS regression (R² = 0.583), as indicated by an improved R² (average R² = 0.670; range: 0.26–0.77) and a lower Akaike Information Criteria (AIC) value (14,736.88 vs. 15,386.59 in the OLS model). These approaches, evidencing spatial varying associations, proved very useful to refine interpretations of the statistical output on adult obesity. This study underscored the geographic variation in relationships between adult obesity rates and mean county altitude in the United States. Our study confirmed a varying overall negative relationship between county-level adult obesity rates and mean county altitude after taking other confounding factors into account.

摘要

在这项研究中，我们使用跨县研究设计评估了肥胖率与海拔之间的关系。在调整了包括体育活动在内的四个预测变量后，我们应用地理加权回归 (GWR) 来检查成人肥胖率与海拔之间的空间变化关联。在 GWR 模型中发现了海拔和成人肥胖率之间的显着负相关关系。我们的 GWR 模型比 OLS 回归 ( R ²  = 0.583) 更适合数据，如改进的R ² (平均R ² = 0.670; 范围：0.26–0.77）和较低的 Akaike 信息标准 (AIC) 值（14,736.88 与 OLS 模型中的 15,386.59）。这些方法证明了空间变化的关联，证明对于改进对成人肥胖统计输出的解释非常有用。这项研究强调了美国成人肥胖率与平均县海拔之间关系的地理差异。在考虑其他混杂因素后，我们的研究证实了县级成人肥胖率与县级平均海拔之间存在不同的总体负相关关系。