overaggregation

Overaggregation is a practice in data analysis and statistics in which information from finer-grained units is summarized into a single, higher-level measure. While aggregation is often necessary for clarity and comparability, overaggregation occurs when the chosen level of aggregation obscures important heterogeneity among subgroups or time periods, potentially biasing conclusions drawn from the data.

Contexts: It arises in statistics, economics, public health, ecology, and data visualization, wherever data are summarized. Common forms include averaging measurements across diverse populations, regions, or time intervals, or using a single overall index to represent a complex system.

Consequences and examples: Overaggregation can mask variation and lead to ecological fallacies, misestimated effects, or inappropriate policy decisions. Examples include reporting a national unemployment rate that hides regional unemployment disparities; presenting a single average test score for a school district that hides performance gaps between schools; or summarizing environmental impacts with a single metric across heterogeneous habitats.

Mitigation: Analysts counter overaggregation by stratifying data, reporting distributional statistics (min, max, quartiles, percentiles), using disaggregated or stratified analyses, and applying hierarchical or multilevel models that respect the data's natural structure. Visualization and interactive dashboards can likewise reveal within-group variation and multiple perspectives.