Digital economy boundaries and measurement · Lesson 3

Access, use and quality

5 min read

Lesson objectives

  • Compare access, use and quality indicators with population and time denominators.
Reading tools

Build a compact dashboard that separates the opportunity to connect, actual use and task quality, stating the population, period and source behind each number. Network coverage does not imply that every person uses a service. An account does not imply that its holder completed a task. The easiest available proxy cannot simply replace the outcome a decision maker needs.

A rate needs a numerator and a compatible denominator. “Households with suitable connectivity” uses all households in its defined population. “People who used the internet during a stated period” uses the individuals in the relevant population or sample, not the number of subscriptions. One person may hold multiple subscriptions. Quality needs its own operational rule, such as completing a task within an acceptable time or succeeding on the first attempt. Connection speed alone will not reveal a language barrier or difficulty filling in a form. Before ranking two areas, compare the age groups, periods, data-collection methods and meaning of “use.”

Read an indicator dashboard by unit and collection method before ranking its values. Give each measure a row naming the observed unit, period, collection route and population missing from it. Subscription counts may contain several connections per person; a use survey may miss unreachable residents; booking success logs cover only attempters. When a series changes method or period, mark a break rather than draw a continuous trend attributed only to behaviour. Before comparing areas, inspect sample size, the meaning of suitable connectivity and the time of collection. Then specify the next evidence needed: interviews with non-attempters or logged form errors, depending on the gap the current indicator cannot explain. The dashboard is a decision aid with bounded reach, not a single ranking.

The International Telecommunication Union publishes connectivity indicators and explains that some estimates use models where survey observations are missing. This is a methodological point, not a claim that the invented case figures below come from ITU. We label training data as hypothetical and use the source to teach reading methods, rather than lending borrowed authority to invented numbers. A shift in an indicator's definition or source can produce an apparent trend even if user behaviour has not changed.

When designing a local dashboard, put a definition record next to every rate: data owner, collection date, missing-value count, unit and next review date. If access comes from a household survey while task success comes from an administrative log of individual attempts, do not draw them as one homogeneous series or multiply them to obtain a supposed population result. A useful display can separate opportunity measures from outcome measures and mark the link between them as a hypothesis requiring suitable data. Representation matters too: a service log omits people who never attempted the task and therefore cannot explain their non-use. A decision about inclusion may need a fit-for-purpose sample of non-users alongside system records. Show uncertainty or the sample's limits rather than ranking areas on a small difference unsupported by collection methods. The dashboard becomes a way to ask and test questions, not an ornament of percentages.

Lesson summary

Apply Access, use and quality to the case data and distinguish the result from an assumption before deciding.