Fieller's theorem vs. The delta method for significance intervals for ratios

1 January 1975

journal article
research article
Published by Taylor & Francis in Journal of Statistical Computation and Simulation

Vol. 3 (3) , 265-274
https://doi.org/10.1080/00949657508810091

Abstract

Fuller's Theorem and the Delta Method are both proposed as methods for obtaining a significance interval for the ratio of two parameters. The relative difference between Fieller and Delta Method lower limits and the relative difference between Fieller and Delta Method upper limits for 95% significance intervals are examined conditional on the observed sample coefficients of variation of the estimators of the unknown parameters and correlation between the estimators The behavior of relative differences are discussed in terms of maximum values of relative differences, conditions for which relative differences are greater than 100% and for which they are less than 10 %. Several rules of thumb for deciding when to approximate a Fieller interval by a Delta Method interval are discussed and evaluated.

Keywords

This publication has 4 references indexed in Scilit:

Fiducial probability
Statistische Hefte, 1967
The Biological Standardization of Insulin
Journal of the Royal Statistical Society Series B: Statistical Methodology, 1941
THE COMPARISON OF DOSAGE‐MORTALITY DATA
Annals of Applied Biology, 1935
THE CALCULATION OF THE DOSAGE‐MORTALITY CURVE
Annals of Applied Biology, 1935