Grawe and Braun (1994) proposed a computerized procedure for evaluatin
g therapy effects, which is based on difference scores. However, in th
is procedure a post-pre-difference which is due only to measurement er
ror may look as big as a post-pre-difference based on perfectly reliab
le measurements. Therefore, we propose two simple alternative indices
which rake into account the problem of measurement error. The first al
lows a simple evaluation of the statistical significance of change, th
e second serves to describe the degree of change in units of the prete
st standard deviation. Finally, we present a simple modification of th
ese indices taking into account not only the problem of measurement er
ror, but also the problem of situational and/or interactional effects.