Local cost surface models of distance decay for the analysis of gridded population data

Christopher D. Lloyd*

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

1 Citation (Scopus)

Abstract

Summary: The paper evaluates some proposed improvements to the analysis of gridded population data, using as a case-study the religious segregation that is observed in gridded population data from Northern Ireland: first, the use of cost surfaces rather than simple Euclidean (straight line) distances to represent the interactions between gridded geographic areas; second, a method for creating gridded cost surfaces that takes account of vector features (such as roads and physical obstructions); third, the limitation of cost surfaces to a tightly defined 'local' set of areas, with a view to reduce computational overheads significantly without adversely impacting the accuracy of subsequent results. The results suggest that all three improvements have merit. The paper further explores the effect of using log-ratios rather than percentages (minimal) and of local rather than global measures of segregation (which allows for considerably greater insight into population characteristics). Although the case-study and results apply specifically to gridded population data, the results of the paper have wider implications for the analysis of any type of zonal data.

Original languageEnglish
Pages (from-to)125-146
JournalJournal of the Royal Statistical Society. Series A: Statistics in Society
Volume178
Issue number1
Early online date12 Dec 2014
DOIs
Publication statusPublished - Jan 2015

Keywords

  • Auto-correlation
  • Census
  • Neighbourhoods
  • Northern Ireland
  • Spatial statistics
  • Travel time

ASJC Scopus subject areas

  • Statistics and Probability
  • Social Sciences (miscellaneous)
  • Economics and Econometrics
  • Statistics, Probability and Uncertainty

Fingerprint Dive into the research topics of 'Local cost surface models of distance decay for the analysis of gridded population data'. Together they form a unique fingerprint.

Cite this