How we analyzed why millions of California students lack broadband access

Lea este artículo en español.

Have your kids spent the last year struggling to connect to their virtual classes because of unstable broadband connections from hotspots or discount internet plans? What about their peers? Or has logging on to remote school been smooth sailing?

For most Californians, the answer appears to boil down to whether their families can afford to pay for a high-speed broadband connection, according to an unprecedented CalMatters data analysis of state and federal data. 

Combining data from agencies that oversee telecommunications companies and schools, CalMatters built a database of broadband adoption and availability estimates in the neighborhoods of most California public K-12 schools. Now we are making it available to the public.

You can explore broadband access in your school neighborhood using our interactive map. Dig deeper with our downloadable database and read on to learn how we analyzed the data.

CalMatters combined data from the California Department of Education, the California Public Utilities Commission, the California Department of Finance and the National Center for Education Statistics, painting a stark picture of California’s disparate access to broadband just before the pandemic struck.

We obtained data that the Public Utilities Commission collected from internet service providers last spring, which was not previously publicly available. The data shows a snapshot of home broadband access in census tracts across California as of December 31, 2019. It includes most types of residential broadband connections, like DSL, cable, fiber, and terrestrial fixed wireless. It excludes mobile broadband, which is what people use to connect to the internet from their phones or hotspots.

The data shows two elements of access: where residential broadband is available and how many people have signed up for a plan. Our analysis excludes residential connections with download speeds less than 6 megabits per second, or Mbps, and upload speeds less than 1 Mbps — generally too slow to participate in a video call — which is a definition of broadband commonly used by California agencies.

Next, we estimated broadband access within public school attendance boundaries collected by NCES in the 2015-2016 school year. We weighted the number of broadband connections and households with broadband available in intersecting census tracts by the percentage of area overlap with the attendance boundary. To calculate residential broadband adoption and availability rates, we used the California Department of Finance estimates of households in 2020 provided to us by the California Public Utilities Commission as the denominator.

While our analysis looks at broadband adoption among all households in school attendance boundaries, we acknowledge that student households generally have a higher broadband adoption rate than non-student households, according to forthcoming research from the University of Southern California and the California Emerging Technology Fund.

We used California Department of Education data showing the percentage of students eligible for free and reduced priced lunch for each school as a proxy for student poverty.

Our analysis encompasses “traditional” K-12 scools, leaving out preschools, adult, continuation and juvenile schools. We also excluded charter schools from the analysis, because most don’t have attendance boundaries. All told, our database includes 6,923 schools, representing 91% of all non-charter public K-12 schools in the state. The other 9% were schools that were missing from the national dataset of school attendance boundaries.

We used the NCES’s definitions of urban, suburban, town and rural schools, and combined “town” schools, which represented 7% of all schools, into the “rural” classification.

It’s important to note that the Public Utilities Commission’s estimates of residential broadband connections and households where broadband is available are not perfect. They are subject to data reporting errors by internet service providers, natural error inevitable in DOF’s household estimates, and the fact that some census tracts include group living situations with many broadband connections but few formal households, and vice versa. Also, the commission assumes that if an internet service provider offers service to at least one household in a census block, then the entire block is covered by that provider. While this method can lead to overestimating availability rates, the commission goes to significant lengths to verify their estimates. In about 8% of census tracts, we found that estimates for residential connections exceeded household counts, so we capped the adoption rate at 100%. However, where there may be census tracts which significantly underestimate adoption rates, we did not make adjustments.

Download the data

If you end up using the data we want to hear about it. We’re curious to hear what you find, what you think might be missing, and if you find our dataset easy to use. You’re free to use the data in whatever capacity; just let us know through our survey form.

We want to hear from you

Want to submit a guest commentary or reaction to an article we wrote? You can find our submission guidelines here. Please contact CalMatters with any commentary questions: commentary@calmatters.org