Robustness of Confidence Intervals

A 95% confidence interval for a parameter is designed so that 95% of possible samples of data would result in an interval that would include the actual value of the parameter. But this rate of coverage of the parameter only holds if the conditions for which the confidence interval formula was developed are true. In this video, we’ll examine how close the coverage rates of our confidence intervals for proportions and means are to what we say they are, even if these conditions don’t hold.

Stream the video without the embedded quiz questions by clicking on the video link below. Closed captions are available.

Notes on the video: Robustness of Confidence Intervals

A point to consider for this video:

This video demonstrates the coverage rates of confidence intervals using two applets. The applet used for confidence intervals for proportions (external link, opens in new window) was developed by Allan Rossman and Beth Chance (external link, opens in new window). The applet used for confidence intervals for means was developed by the Web Interface for Statistics Education (WISE) project (external link, opens in new window) but unfortunately is no longer accessible. We are grateful to the WISE project and to Professors Rossman and Chance for permission to use their applets.