my problem is this...
- I have a batch of 15 self report surveys of the barriers to research
utilisation in healthcare settings
- they each address the same research question, similar populations and
USE THE SAME SCALE (the 'barriers' scale (Funk et al. 1991).
- the scale is made up of 29 items each of which reports a having a mean
item score, standard deviation and an associated proportion of people
viewing the item as a moderate/great barrier to research.
- the surveys are each (individually) small samples but add up to a
sizeable n=20k or so....
I wish to pool the 15 survey responses to the 29 items to reduce the
confidence intervals associated with the barrier items. And also as it
would be nice to present a reasonably 'tight' overview and influence
anyone else considering doing yet another underpowered survey of self
reported, perceived, behaviour. I think David Covell (1985) showed how
unreliable that was regarding info seeking and utilisation!
I have looked at ARCUS biostat's meta analysis suite but to no avail...
it insists on generating ratio based results.... needing two groups....
hmmm
Does anyone know of HTA reports or similar which have combined results
of survey data... or would it be OK just to simply average the means (&
SDs) from the different studies... maybe I am trying to hard to do
something sexy with these figures - so simple would be good if poss.
cheers
Carl.
*****************************************
Carl Thompson
Research Fellow & Chair of Graduate School
Genesis 6
Department of Health Studies
Science Park
University of York
York, YO10 5DQ
tel +44 (0) 1904 434115
fax +44 (0) 1904 434102
mobile 07939 254 863
e-mail [log in to unmask]
*****************************************
|