Hello all
I have conducted attitude surveys at the same two points in time in two near
identical regions of an organisation. A team-based intervention was carried
out in the 'treatment' region and a management based intervention in the
'comparison' region. There was no strict 'control' group with no
intervention. Of course many other organisational initiatives were on-going
in both regions throughout. Essentially, attitude scores and objective
output measures in the treatment region improved, while in the comparison
they did not.
The workforce was stable, hence we know the populations for both regions pre
and post were pretty much the same. Unfortunately, because of fears of
breaches of confidentiality, we were not able to identify individuals, hence
we cannot match individuals in the pre and post samples in either of the two
regions. Also, while every employee was sent a questionnaire, response rates
fell from about 70% to 60% in both regions. Hence we know we have two
non-identical, yet very similar, samples pre- and post in each region, with a
certain amount of attrition.
I would welcome advice on the most approriate statistical technique to assess
whether there has been a meaningful and statistically significant effect in the
treatment region, which addresses some of the threats to internal validity I
face.
Thanks
Steve.
---------------------------------
Stephen Young PhD
Department of Psychology
University of Liverpool
c/o 61 Black Griffin Lane
Canterbury
Kent CT1 2DG
Tel/Fax: 01227 471332
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|