RESEARCH ARTICLE


Why Confidence Intervals Should be Used in Reporting Studies of Complete Populations



Matthew D. Redelings1, Frank Sorvillo1, 2, Lisa V. Smith1, 2, *, Sander Greenland2, 3
1 Los Angeles County Department of Public Health, Office of Health Assessment and Epidemiology, USA
2 University of California Los Angeles, Department of Epidemiology, USA
3 University of California Los Angeles, Department of Statistics, USA


Article Metrics

CrossRef Citations:
3
Total Statistics:

Full-Text HTML Views: 250
Abstract HTML Views: 517
PDF Downloads: 526
Total Views/Downloads: 1293
Unique Statistics:

Full-Text HTML Views: 139
Abstract HTML Views: 311
PDF Downloads: 419
Total Views/Downloads: 869



Creative Commons License
D. Redelings et al.; Licensee Bentham Open

open-access license: This is an open access article distributed under the terms of the Creative Commons Attribution 4.0 International Public License (CC-BY 4.0), a copy of which is available at: https://creativecommons.org/licenses/by/4.0/legalcode. This license permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

* Address correspondence to this author at the Los Angeles County Depart-ment of Public Health, Office of Health Assessment and Epidemiology 313 N. Figueroa Street, Suite 127 Los Angeles, CA 90012, U.S.A; Tel: (213) 240-7785; E-mail: lismith@ph.lacounty.gov


Abstract

Public-health reports sometimes leave out confidence intervals when data are presented for an entire popula-tion. A rationale cited for this practice is that population statistics are measurements rather than estimates; hence there is no need to consider random error because the statistics show exactly what occurred. We argue that this reason does not justify leaving out interval estimates. Targeting intervention in areas with high disease rates can be justified only on the assumption that the excess would continue in those areas; in that case, at the very least, we need to allow for random fluc-tuations over time. Thus, we recommend that interval estimates be reported even when the entire population is observed.

Keywords: bias, confidence intervals, population studies, random error.