comparison: single case - small group

[Marita Boehning]

Dear Info-CHILDES,

  statistically it seems to rather hard to find a solution to the
following problem but maybe I haven't looked in the right places.  Can
anyone help me out with this one:

Say, you investigate a single case's performance (let's say a child with
SLI) on an experiment or test (not standardized).  Then you want to
compare the performance with a group of normally developing children.
However, your group of controls is not very large (N= 7 or a bit
larger).  It is very likely that the groups' performance will not be
normally distributed as the group is too small.
Usually one can compare single cases with groups using z-score or
T-score. However, the use of z-scores or T-scores is only "allowed" if
the group's performance is normally distributed.  The only other method
I could think of or find in books is to use percentiles. These are not
very informative as it does not take SDs and/or mean into account. Thus,
does not give us a very deep insight into group vs. single case
behavior.
Does anyone know of another method other than percentiles?
Thank you so much for your suggestions.

Marita Böhning


--

******************************
Marita Boehning
Department of Linguistics
University of Potsdam
P.O. Box 60 15 53
D - 14415 Potsdam
Germany
Phone: +49 331 977 2929
Fax:  +49 331 977 2095
*****************************