.svg)
Foot Traffic Data for Public Health Research
Public health questions often turn on where people gather, for how long, and who else is present, none of which appears in survey responses or vital statistics. Foot traffic panels record visits to individual clinics, pharmacies, parks, bars and grocery stores at a daily or weekly cadence, along with the home census block group of the people making those visits. That allows exposure, utilisation and behavioural response to be measured as they happen rather than as recalled months later by a respondent. The trade is that these measures are built from a sample of mobile devices, so they describe relative movement across places and weeks well and absolute population counts less well.
How public health researchers use foot traffic data
Behavioural response to outbreaks and public health guidance
A standard design pairs weekly visit counts at points of interest with the timing of a mandate, advisory or case surge, and estimates the change in visits relative to comparable places that were not covered, often splitting the effect by the income or racial composition of the visitors' home block groups. The variable is the visit itself rather than a stated intention, which removes the recall and social desirability problems that make self reported distancing hard to interpret. What breaks is representativeness: the panel is a self selected sample of smartphone owners, so compliance estimates for older and lower income populations depend on inferred home locations rather than observed demographics.
Healthcare access, utilisation and catchment areas
Accessibility measures such as floating catchment area models normally assume that patients use the nearest provider within a fixed travel radius, an assumption that foot traffic data replaces with observed flows from residential block groups to clinics, pharmacies and hospitals. That makes it possible to distinguish proximity from realised use, and to show where a facility that looks accessible on a map draws almost nobody from the neighbourhood around it. The caveat is that a visit to a medical building is not evidence of a visit for care, and small or low volume facilities are frequently suppressed or missing from point of interest registries entirely.
Place based exposure: alcohol outlets, food environments and injury risk
Routine activity designs pair visit counts and dwell times at specific venue categories with incident records geocoded to the same blocks, testing whether injury, violence or overdose risk tracks the number of people converging on a place rather than the number who live there. Foot traffic supplies the denominator that residential population cannot, since a block group with few residents may host thousands of nightly visitors. The limitation is that counts are modelled up from a device sample to a population estimate, so absolute footfall carries real uncertainty and the credible signal is variation across comparable places and time periods.
Foot traffic datasets available on Dewey
Advan
Visit counts and visitor home origins for US points of interest, which public health researchers use to build continuous exposure and mobility panels around policy changes, outbreaks and local shocks.
View datasetSafeGraph
A point of interest registry with visit and spend patterns, useful when a study needs stable place identifiers and category codes to define clinics, pharmacies, parks or alcohol outlets before measuring visits to them.
View datasetVeraset
Device level mobility records for researchers who need custom geofences, stay points or dwell times that an aggregated visit panel cannot provide, with the handling and disclosure obligations that raw location traces carry.
View datasetMore published work using foot traffic data
Samuel Rueda · Working Paper · 2024
Szandra A. Péter, Travis Gallo, Jennifer Mullina, Amira Roess, Gabriela Palomo-Munoz, Taylor Anderson · Nature · 2025
Lee, Serin, Zabinsky, Zelda B, Liu, Shan · Health Care Management Science · 2025
.svg)