• 3 min read. Content may take a few seconds to load.

Mobile Devices Visits & Devices Sampled

You might think just about everyone carries their mobile device wherever they go in the State of Hawaii. Even if that were true, it does not mean that everyone who carries a mobile device with them is part of the SafeGraph panel sample of mobile devices. SafeGraph data comes from a dynamic panel of mobile of devices. As you can see below, in any given month the SafeGraph panel sample might track as many as 250,000 mobile devices in the State of Hawaii, of which maybe 130,000 are Hawaiian “home” devices, representing about 10% of the population of the entire State of Hawaii.

Whenever you compare mobile device visits at two different points in time, you have to take into account the fact that the sample size is different. In addition, since some parts of the population did not have the same chance of being selected into the SafeGraph mobile device panel as other parts, the sample is not going to be perfectly representative of a population of interest, which you also have to take into account.

Mobile Devices Visits & Devices Sampled Statewide

The Data: The first graph shows you the total number of devices (total_devices_seen) and subtotal of Hawaiian “home” devices (total_home_visitors) in the SafeGraph panel that visited any point of interest or home location on a given day between January 1, 2018 and June 30, 2021. The first graph also shows the number of visits (total_visits) that these devices made to over 40,000 points of interest in the State of Hawaii. The second graph shows the total number of point-of-interest visits observed in the the State of Hawaii (num_visits), the number of unique visitors observed with at least one point-of-interest visit in the specified state (num_unique_visitors), and the number of distinct devices observed with a primary nighttime location in the specified census block group (number_devices_residing).

What to Look For: The two graphs below give you an appreciation of the size of the overall SafeGraph panel sample as well as the fact that the panel sample is always changing, on a monthly and even daily basis. After Feburary 2020, the size of the SafeGraph panel sample obviously drops precipitously, not merely because of an anomaly in sample size but due to the COVID-19 public health emergency.

Scroll over the graphs below.

Hawaiian Mobile Devices Sampled as Percentage of the Population of Hawaii

The Data: The histogram below uses data on the number of distinct devices observed with a primary nighttime location in a Hawaiian census block group (number_devices_residing), otherwise known as Hawaiian “home” devices. A mobile device is determined to be a Hawaiian “home” device by analyzing the last six weeks of dwell times during primary nighttime hours from 6 pm to 7 am. If there is enough evidence the device had a primary nighttime location in the State of Hawaii, then the device can be assigned to one of 872 Hawaiian census block groups.

What To Look For: Some places are more highly sampled than others. You can also see that the size of the SafeGraph mobile device panel sample changes from month to month, measured as a percentage of the 2016 American Community Survey census block group population of the State of Hawaii, where we assume that the resident population is the same each year. Since January 2018, the SafeGraph mobile device panel has captured an average of 7% to 12% of the Hawaiian “home” devices, representing anywhere from 70,000 to 130,000 Hawaii residents. That total does not include non-resident or visitor mobile devices, which can represent almost as many mobile devices as Hawaiian “home” devices in State of Hawaii at any given time. Notice also the red line (mean value) is always to the right of the green line (median or “middle” value in a sorted list of numbers), and that the larger the panel sample, the further apart the two lines appear. Location errors like GPS drift can associate mobile devices with census block groups that have a very small or even zero population, which creates a long tail of outliers with significantly higher than average values that tend to pull the mean value (red line) further apart from the median (green line).

See histogram below.


SafeGraph variable names are in parenthesis. For detailed information on how SafeGraph variables were derived, see the SafeGraph documentation. SafeGraph data was processed for the State of Hawaii using PostgreSQL and PostGIS then added into R. Graphs and bar charts developed in R included use of the dygraphs and ggplot2 packages.

Tell us what you think.

Click Here To Go To Your Next Course