Does Your Mapped Variable Add Up? Statistical Guardrails for Web-Based Cartogram Generation

Abstract

Data and code accompanying the paper. Area cartograms are appropriate only when area represents a meaningful total rather than a derived quantity such as a density, rate, or percentage. This repository contains subnational (admin-1) data for 3,591 administrative divisions across 34 geospatial variables (17 intensive, 17 extensive), an R analysis pipeline implementing a generalized additive model that tests for extensivity violations, and replication scripts for the benchmark results reported in the paper. You may find the latest code on GitHub in our repository: mgastner/cartogram-extensivity-guardrail.