The concentration-coverage curve: A tool for ecological studies

Warning View the most recent version.

Archived Content

Information identified as archived is provided for reference, research or recordkeeping purposes. It is not subject to the Government of Canada Web Standards and has not been altered or updated since it was archived. Please "contact us" to request a format other than those available.










by Philippe Finès

Background

A selective approach may be used in an ecological study where the aim is to choose a subset of units of analysis (UAs) and produce interpretations about a population of interest (PI) based solely on those UAs.  The results for the PI will be reliable if that population is concentrated in the selected UAs and rare in other UAs.  This article presents a graphical tool that helps determine whether these conditions are satisfied.

Data and methods

Data on the Inuit and Métis ancestry populations from the 1996 Census of Canada are used for illustrative purposes.  Based on a classification statistics table, a concentration-coverage curve can be created for a given PI. The shape of the curve indicates whether it is possible to choose a threshold that will yield both adequate concentration and adequate coverage of the PI.

Results

The concentration-coverage curve shows that, among Aboriginal peoples living in rural areas, the Inuit population is classifiable, but the Métis population is not.

Interpretation

This method can be applied to any ecological study focussing on the proportion of individuals sharing a single characteristic defined by a binary variable.

Keywords

classification, demography, ecological studies, population distribution

Findings

In ecological studies, individuals are grouped into units of analysis (UAs) rather than being considered separately. In certain instances, a researcher may wish to employ UAs to make inferences about a population, but there is a potential for error if the population of interest (PI) is not sufficiently represented in those UAs. [Full text]

Authors

Philippe Finès (613-951-3896; Philippe.Fines@statcan.gc.ca) is with the Health Information and Research Division at Statistics Canada, Ottawa, Ontario, K1A 0T6.