BLOG POSTS ARE PRELIMINARY MATERIALS CIRCULATED TO STIMULATE DISCUSSION AND CRITICAL COMMENT. THE VIEWS EXPRESSED ARE THOSE OF THE INDIVIDUAL AUTHORS. WHILE BLOG POSTS BENEFIT FROM ACTIVE UHERO DISCUSSION, THEY HAVE NOT UNDERGONE FORMAL ACADEMIC PEER REVIEW.
By Justin Tyndall and Joshua Hu
Staying at home and avoiding crowded spaces is an important way for residents to help contain the spread of COVID-19. Measuring the ability and willingness of Hawai’i’s residents to stay at home is difficult. Having more information on resident movements can help gauge the frequency of in-person interactions and the probability of future virus spread.
UHERO is releasing a new dashboard that makes use of detailed data on the movements of people in Hawai’i. The data come from the firm Safegraph, which collects cellular location data from the apps that run in the background of many people’s smartphones. The data are anonymous and represent a selective sample of the population. (Further details regarding the sample and Safegraph’s methodology can be found here). The data can be used to provide a fairly detailed picture of people’s movement around Hawai’i during 2020.
The figure below shows the share of smartphones in Hawai’i that did not leave their user’s home on a particular day. The data exclude tourists. If we assume people normally carry their smartphones with them, the data can approximate the share of people who are staying home all day. Prior to the pandemic, about 20% of people stayed entirely at home on a given day. This share rose sharply in March and April, peaking at 42%. The data show another peak in August, corresponding to the height of COVID infections in the state. Since then, the share of people staying at home has fallen again and is now hovering around 32%.
Because the data are broken down by location, we can provide some information on how rates of staying at home are varying across the state. Residents of O’ahu have stayed home at slightly higher rates than residents of the neighbor islands. We can also explore variation across neighborhoods. For example, stay at home rates for residents living around Waikīkī have recently fallen to 25%, below the state level of 32%. The dashboard can be used to explore the data across islands, zip codes and census tracts. Generally, we did not find huge differences in the rates across neighborhoods. In many ways the effects of the pandemic have been felt differently in rich and poor areas, but we did not find an apparent correlation between the average income within a neighborhood and the stay at home rate for that area.
Given concerns with specific locations becoming the center of outbreaks, we have also begun tracking the amount of foot traffic (proxied by the number of smartphones) at various types of locations. The data can be explored on the new dashboard. For example, retail stores in Hawai’i have recently experienced 60% of the foot traffic that was typical before the pandemic, while foot traffic in hotels is down to 30% of the pre-pandemic level. More location types are available on the dashboard.
The underlying data are released regularly, with a small lag, and we will continually update the data to provide recent information on movement in the state. As we enter the holiday season many residents may face additional desires to travel and gather. At this point, we have data up until Thanksgiving Day, and the relevant data do not show a significant change in the number of people staying at home, suggesting that many may have heeded public health warnings to avoid Thanksgiving gatherings.
Monitoring the dashboard and the underlying data for changes in human mobility could be predictive of future virus spread. Hawai’i has managed to escape the worst of the virus when compared to other states. Our ability to maintain physical distance from one another will be important in determining the virus’s impacts in the coming months.
5 thoughts on “Are Hawai’i Residents Staying at Home?”
Leave a Comment
This information is already being misused, by Hawaiʻi News Now, which casts it as support for questioning the value of staying at home in its headline: Does staying at home help stop coronavirus spread? UHERO set out to find out.
Wow mahalos for this.
Can you clarify which Safegraph field or fields are being mapped here? I used the link at the bottom of the dashboard site to find the “staying home” definition, but I only found durations/distance values (such as median_home_dwell_time) rather than a binary staying home type.
Also, I really hope this doesn’t predict cases because eyeballing the decline in cell phones at home makes it seem like it only took about a 3-5 point drop for cases to surge in June/July if you assume a couple of weeks lag. That’s pretty concerning! I’m assuming this doesn’t correlate given the way the foot traffic data doesn’t seem to indicate that higher traffic = higher cases (also a great dashboard!), but that leaves the question of what caused it unfortunately.
Appreciate your support.
There are two relevant variables here which are calculated for each census block group: `completely_home_device_count’ and `device_count.’ Aggregating each variable to the relevant granularity (i.e. census tract, zip code, island, and state), then dividing the first by the second, we obtain the share of cases which remain home. I will also note that we aggregated data by week to reduce noise and provide a more meaningful and accurate visualization, as the day of the week does impact the percentage of people who stay home.
Ah got it, thanks!
Can you correlate staying at home with incidents of infection by Covid – of special interest are high rises especially older and more congested ones where air flow, social distancing, plumbing, and overall hygiene are issues. Suspect they are dens for proliferation of viruses – remember legionnaires disease.