Z scores derived from a regression equation in one group applied to other groups I work in the developmental psychology arena. 
Often papers report using regression to derive a function of age onto a score in a typical group and then applying this function to a second atypical group, saving the residuals then transforming these to z scores in respect to the typical groups mean and standard deviation. 
I have derived the regression function intercept and slope, for example 5.5 + (.05*age)and applied it to all children. Then computed the residual (actual - predicted) for all children. I then applied the z score equation (z = x - mean /sd) using the mean and SD of the typical groups' raw residuals. This results in a mean of zero and sd of 1 for the typical group and mean z scores much lower (as might be expected)in the atypical group. 
The alternative was not to use the mean and sd of the typical group and generate the z scores from all children in the group, whereby the mean z scores of the TD group are now higher than zero and the z scores reflect the distance from the mean of only my sample. 
My question is whether this is what others mean when they use this to standardize the sample and is there anything incorrect in either way of calculating these scores? 
Thanks in advance. 
 A: There are, I think, multiple uses of the term "standardize" in this sort of situation.
For diagnostic purposes, on a test that has been standardized on some group, we use that mean and sd. For example, IQ tests are normed for different ages to have a mean of 100 and sd of 15 (or 16 for some tests). We can then compare either an individual or a group to that mean and sd. 
But if the test isn't normed (perhaps it is a new measure) or if our purpose is not to compare to a normed group, then I think either of the methods you describe could be called "standardizing". That is, you have tested two groups. You can standardize within each group or across the two groups. Or, if you want to attempt to replicate the first purpose with a new measure, you could norm it within the typical group and use those norms on the atypical group.
Authors should state more detail than just "we standardized the scores". They should state what mean and sd they used to do the standardization. I've seen this stated in many articles, but some authors are careless about it. 
