On 1 Apr 2003 06:23:44 -0800, [EMAIL PROTECTED] (VOLTOLINI) wrote: > Hi, my doubt is simple...... > > > When to use the Standard Deviation (SD) and the Standard Error (SE)? > > I preffer to use the SE because when comparing groups of data the box plots with SE > give me the same results as the tests. Then, it seems that using a box plot with SE > the conclusion is more powerfull. >
A "standard error" is the standard deviation of a *statistic*. The term, Standard deviation, is often (but not always) restricted to the description of raw scores. The context where we have to choose between SE and SD is usually the simplest one: for the sample and its mean. Do we say, Here is the (a) SD: the standard deviation of the sample, or (b) SE: the Standard error of the sample (SD of the sample's mean, often SDi/ sqrt(Ni) ). Individual variations in scores require the SD. Sample means and their contrasts call for the SE. Especially if your context does not strongly suggest one or the other, and your scale units are arbitrary, and your sample Ns are equal, it might not matter which you show. -- Rich Ulrich, [EMAIL PROTECTED] http://www.pitt.edu/~wpilib/index.html . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
