Efficient Sampling Strategy that Preserves Homogeneity and Statistical Power

Develop an improved sampling strategy for selecting addresses within census block groups that simultaneously ensures statistical significance of aggregated metrics and preserves the geographic, demographic, and socioeconomic homogeneity of reported statistics when assessing CAF serviceability and compliance.

Background

The analysis aggregates metrics such as serviceability and speed at the census block group level. In some areas, limited address counts and query hit rates constrained sample sizes, reducing statistical significance.

While coarser aggregation could increase statistical power, it would mix heterogeneous geographies and demographics. The authors therefore explicitly leave open the task of designing a more efficient sampling approach that balances significance with the need to preserve local homogeneity.

References

We leave the exploration of a more efficient sampling strategy that strikes a balance between the statistical significance of aggregated metrics while preserving the geographic, demographic, and socioeconomic homogeneity in reported statistics for future work.

The Efficacy of the Connect America Fund in Addressing US Internet Access Inequities (2405.18657 - Manda et al., 28 May 2024) in Section: Limitations, Aggregation at finer spatial granularity