Recently, I detailed the Broadband Connectivity Index we are building into the TPI Broadband Map. The BCI is derived using a principal components analysis that explicitly takes into account many factors, not just availability, though of course that is important. The index can help policymakers prioritize areas for buildout (the “where”), while the components of the index can help identify the type of assistance (the “what”) that might be most useful.
This post uses Florida to demonstrate how a connectivity index can be useful. This analytical exercise comes with two important caveats, however. First, we calculated the connectivity index at the county level. Counties are far too large to provide actionable information to state policymakers, who already have a deep understanding of which counties are well off and which are not. A more useful analysis would be at a smaller geographic level, ideally Census Block. A Census Block analysis would help policymakers identify very specific areas on which to focus. Second, while the connectivity index helps to smooth out errors in each individual dataset, the breakouts by index components are only as good as the underlying datasets. In order to be actionable, it would be important to check results with complementary information. It is far more cost-effective, however, to do intensive data collection in Census Blocks identified by the index as a starting point than it is to collect that data for the entire state, particularly when it may not be necessary in some areas.
The figure below maps the results, and the table below the map ranks Florida counties from lowest BCI rating (Dixie County) to the highest BCI rating (St. Johns County).
The following figure shows the components of the connectivity index for each county, as ordered by lowest to highest BCI rating. Each “spoke” on the radial charts shows the value of the component for that county as a percentage of the maximum index component rating of any county in the state. That is, rather than showing the data values, it shows how well the county fares relative to the best county in that state.
The radial charts show that while the measures are correlated with each other—having a high (or low) score in one area means the county is likely to have a high (or low) score in another—the correlation is not perfect. For example, Hardee County has, according to the FCC, the highest share of population that can access connections offering 25/3 service, but has a relatively low adoption rate and also among the lowest share that can access 100/25 service. In contrast, St. John’s County, which has the highest BCI rating, does not have the highest of any of the measures by themselves.
To reiterate, this example is not intended to tell Florida policymakers anything they don’t already know about their counties. Florida’s own Office of Broadband is doing extensive work to better identify and understand the state’s unique situation and responses. Rather, we show how multiple datasets can be combined to yield new insights with a method that generates a comprehensive snapshot of connectivity data through the development of a Broadband Connectivity Index. The states may find it useful to take a similar approach at smaller geographic levels that takes into account a state’s own data to most effectively spend the resources available to it.
Access to connectivity and the digital divide are multi-faceted policy problems that require multi-faceted analytical methods. The Broadband Connectivity Index may be a powerful approach to integrate what we know in separate datasets to better focus the priorities and attentions of state officials with limited time and the important task at hand to fund broadband infrastructure projects in their local districts.