Showing posts from November, 2020

Fixing Election Maps using Street Lamps

Update: this stress fueled post won 1st place in the Mapbox Election Mapping Challenge ( "Best Data Mashup"). Be sure to check out the other cool projects that were entered!  As election night nears, we're about to be bombarded with maps. Almost all of them will be highly misleading. Using precinct level data from the 2016 eleciton, I explore a fundamental flaw in the conventional approach to election mapping, and outline a new method for correcting this bias using satellite images of the earth taken at night.  The problem with normal election maps When presenting election results, standard practice is to do so at the level of administrative units (e.g. states, counties, precincts), coloring their whole area according to voteshare. You've come across this type of map before: The New York Times boasts that this is an "Extremely Detailed Map of the 2016 Election"  because it uses precincts, which are smaller than the more commonly used counties. But this typ