Notes from Kline, Rex

  • Published 2014


Notes that many statistics have more complex sampling distributions than do means, for example, , the standardized difference between means. Briefly discusses ways to construct confidence intervals about such statistics. Mentions confidence interval transformation (Steiger & Fouladi, 1997) – the statistic is transformed into normally distributed units, lower/upper limits of the interval are obtained by subtracting/adding ZccSE to the transformed statistic, resulting in a confidence interval for the transformed statistic. The limits are then transformed back to the original metric. Construction of a CI about Pearson r is given as a well-known example of this approach.

Cite this paper

@inproceedings{2014NotesFK, title={Notes from Kline, Rex}, author={}, year={2014} }