Confidence interval scatterplots for evaluating confidence interval performance
Murali Subramaniam, Lawrence M. Leemis · Journal of Statistical Computation and Simulation · 1988
Confidence intervals are widely used to determine an interval, estimated from data, that contains a population parameter with a prescribed probability. Kang and Schmeiser (1986) introduce a scatterplot that is useful in comparing the effectiveness of different confidence interval procedures based on performance measures such as the coverage, the expected interval width and the variance of the interval width. This paper examines the usefulness of this confidence interval scatterplot (CSI) for evaluating approximate confidence intervals, as illustrated by the comparison of simultation output analysis techniques for correlated observations.