I have been a receptive member of this mail-list for some time and enjoy
the discussions - now a first contribution so forgive any naivete.
Working in an area where case populations are frequently small and the
variance in many assessments is incredibly wide due to the heterogenity of
the populations (rehabilitation in brain injury), underpowered studies are
a real problem for researchers and clinicians to deal with. My main concern
has been when researchers have not even considered Type II error problems.
Conlcusions are then stated that treatments are ineffective and the less
'critical' reader may well accept that conclusion. I'd like to think that
fewer of these articles see the light of day, but they still seem to get
published, complete with overstated conclusions!
Rather than relying on metanalysis to extract statistical significance from
a number of small studies (risky I think for a whole range of reasons), it
is quite challenging for researchers to consider the range of methodologies
and choose (and justify) the most appropriate one for their population in
order that both statistical and clinical significance can be demonstrated.
regards to the members of the list
_________________________________________
Kath McPherson PhD
Senior Lecturer - Rehabilitation (non-medical)
Wellington School of Medicine
ph 04 385-5999 (ext 6122)
fax 04 389-5427
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|