Reliability of measurement and the methods of estimating reliability.

Author: Yi-biao ZHOU ¹ ; Gen-ming ZHAO
Author Information

1. The Department of Epidemiology, School of Public Health, Fudan University, Shanghai 200032, China.
Publication Type:Journal Article
MeSH: Algorithms; Data Collection; methods; statistics & numerical data; Epidemiologic Measurements; Observer Variation; Reproducibility of Results; Time Factors
From:Chinese Journal of Epidemiology 2003;24(12):1146-1149
CountryChina
Language:Chinese
Abstract:
OBJECTIVETo explore the methods of estimating measurement reliability and their limitations.
METHODSAccording to the given examples, reliabilities of measurement were calculated and limitations of the methods of estimating reliability were analyzed.
RESULTSThe Kappa value of interobserver reliability was 0.793 between two pathologists. Between the two populations with different prevalence rates, the values of Kappa were 0.800 and 0.137 respectively, and with big difference. Cronbach's alpha coefficient of compositive index for A type behavior was 0.55.
CONCLUSIONThe Kappa index and alpha coefficient were both inherently population-specific. Before generalizing to different populations, the reliability needs to be measured.