Title: Some Misconceptions about Confidence Intervals
Author: Keith M. Bower
Publication Source: American Society for Quality, Six Sigma Forum, July 2003
Abstract
This article discusses three misconceptions regarding the use of confidence intervals, namely:
(1) Overlapping confidence intervals imply “no difference.”
(2) Confidence intervals are always appropriate.
(3) Confidence intervals address the spread of individual values.
Notes
This paper aimed to address a subset of the many misconceptions I had seen from practitioners regarding confidence intervals. I chose the three listed based on both frequency of occurrence and severity of mistake.
My guess is that misconception (1) comes from a two-directional implication that doesn’t exist. Misconception (2) was to highlight the importance of having a stable process for the theoretical assumptions for a confidence interval to make sense. Misconception (3) was to make people aware of “tolerance” intervals, which I feel could be of more use to many practitioners than the rigid use of confidence intervals for population means.
To download Adobe Acrobat (for free) click here.