Some of those who think that estimation of the size of effects is more important than the testing of a nil hypothesis of no effect argue that we would be better served by reporting a confidence interval for the size of the effect. Such confidence intervals are, in my experience, most often reported in terms of the original unit of measure for the variable involved. When the unit of measure is arbitrary, those who are interested in estimating the size of effects suggest that we do so with standardized estimates. It seems to me that it would be useful to present confidence intervals in standardized units.
As an example, consider the research described on pages 209 and 210 of the fifth edition of David Howell's Statistical Methods for Psychology. Homophobic and nonhomophobic men are compared on a measure of sexual arousal when viewing a video containing explicit man to man erotic content. The unit of measure for the arousal variable is not meaningful to most persons, so Howell recommends that the effect size be presented in standardized units. The effect size, by the way, was .62 (with homophobic men being more aroused), which strikes me as being a medium to large sized difference. Howell opined that a 95% confidence interval is not very informative in this case, because the unit of measure is arbitrary. The confidence interval extends from 1.46 to 13.54. That really does not tell us much, does it -- other than that the value of zero is not included. Howell suggested that the confidence interval was less useful than a standardized estimate of effect size in this case. Now consider standardizing that confidence interval. I simply divided each of the confidence limits by the pooled standard error (12) to obtain a confidence interval extending from 0.12 to 1.13. That is, in standard deviations units, we are 95% confident that the mean is somewhere between 0.12 (small, maybe even trivial) to 1.13 (huge). IMHO opinion, this is much more useful than just the standardized point estimate of effect size or the nonstandardized confidence interval. Why don't people report standardized confidence intervals in situations like this? +++++++++++++++++++++++++++++++++++++++++++++++++ Karl L. Wuensch, Department of Psychology, East Carolina University, Greenville NC 27858-4353 Voice: 252-328-4102 Fax: 252-328-6283 [EMAIL PROTECTED] http://core.ecu.edu/psyc/wuenschk/klw.htm ================================================================= Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =================================================================