How is the Digital Distress Indicator Computed?

The Digital Distress Indicator

The Digital Distress Indicator is comparative measure of the level of digital readiness of various regions, such as school districts or counties. A good introduction is found in Digital Distress: What is it and who does it affect?. The Connectivity Explorer has adapted this measure to rank school districts and counties within a single state.

The Digital Distress Indicator is computed from Census ACS-5 data. It uses two measures, subscriptions and access .

The subscription measure is computed from two Census estimates:

  1. the fraction of households having only a cellular data plan, plus
  2. the fraction of households that do not have any Internet subscription.

The access measure is computed from two other Census estimates:

  1. the fraction of households having only a mobile device, plus
  2. the fraction of households that do not have any devices.

Each pair of measures is added together, and a z-score is computed for each sum. The z-score measures the distance from the value to the average (mean) of all the values.

Finally the two z-scores are added together, and the result is normalized on a range from 0 to 100, where 100 indicates the highest degree of distress. For places where there is only a single school district, like Hawaii or Puerto Rico, that district gets a DDI of 100.

1 Like

Gallardo’s work is consistently excellent, and this is no exception.
I do think, however, that there is a missing factor in this analysis - choice. I helped a team of students at UW develop a similar measure a few years ago that included choice as a component.
A household that has both a mobile and an internet subscription, but where both are provided by lightly-regulated monopolists is also experiencing a degree of digital distress, though admittedly a lesser degree.

Hi Will -

Roberto and I talked about this after your note. You’ve triggered two thoughts.

The first concerns how to defined choice in this context. Roberto and his group were working with primarily Census and other economic data. If you fold in the FCC data (along with a definition of choice) we could build a new indicator that captures your idea. Would the I3 notion of When is a block “well-served?” be useful here? Does anyone have another measure that we might use to capture “choice”?

The second thought is that Roberto Gallardo and his collaborators have shown us how to create families of distress indicators based on choosing appropriate measures for forms of distress. So the Digital Distress + Choice is a new indicator, but a Digital Distress + the CDC’s Social Vulnerability Index might be another measure of risk for populations. The hard part, of course, is to define and field-test indicators that properly capture their intended purpose.