Why global averages often fail to capture the true reality of local crises
In short: When studying obesity in the US or COVID-19 in Jakarta, a single national average can be dangerously misleading. Global regression models often miss the critical, localized patterns that define public health. To find the truth, researchers are turning to local regression to uncover hidden spatial and temporal dependencies.
In large-scale studies, such as those examining the relationship between food deserts and diabetes across 3,10 8 US counties, global regression models can show non-significant associations. For instance, while global models might suggest fast-food density has no significant link to obesity, geographically weighted regression reveals that in specific areas, like western New York and northwestern Pennsylvania, a positive association does exist. This highlights the importance of accounting for spatial heterogeneity.
The same principle applies to infectious disease control. During the COVID-19 crisis in Jakarta, Indonesia, researchers found that a global regression model could not adequately capture how human mobility restrictions impacted transmission. By modifying the model into a local regression approach that accounts for spatial and temporal interdependence, they found that the impact of mobility on the log relative risk of COVID-19 varied significantly across Jakarta's 44 districts, ranging from –4.445 to 2.353.
Technically, local regression—including methods like LOESS and LOWESS—operates by fitting simple models to localized subsets of data. Unlike classical global procedures, the analyst does not need to specify a single global function for the entire dataset; instead, the method builds a function point by point. While this provides immense flexibility and can handle complex, non-linear patterns, it comes at the cost of increased computational intensity, making it a technique that truly leverages modern computing power.
Opens the Intellect+ feed with this card first. Swipe up for the next one.
Sources
- Summary by Intellect · source: Local regression
- Also: Food deserts exposure, density of fast-food restaurants, and park access: Exploring the association of food and recreation environments with obesity and diabetes using global and local regression models.
- Also: Does the Restriction of Human Mobility Significantly Control COVID-19 Transmission in Jakarta, Indonesia? Global Versus Local Regression Models
To cite this card, link to https://intellectplus.co.uk/app/health-local-regression.
How we source, check and correct cards: Editorial standards.
General information, not professional advice. Spotted an error? Flag the card or see corrections and complaints.
Intellect Plus is a visual educational discovery platform for exploring fascinating facts and ideas. Brain Food, Not Brain Rot. What is Intellect Plus?
Related
More in Epidemiology · More in Health · Latest · A–Z · All topics
