Dear all
I would be grateful for your advice regarding the proper analysis of
some cell fluorescence data. My colleagues are investigating whether
rheumatoid arthritis patients differ from healthy controls in the
numbers of certain receptors on the surface of their blood cells. They
do this by feeding a blood sample (which has been stained with a
fluorescent dye) into a machine, which then tells them how many cells of
a particular type are present, and measures the level of fluorescence of
these cells. The statistics the machine produces detail the mean, cov,
geometric mean, and median fluorescence of the cells. These summary
statistics are produced for each patient. It is clear from these that
the distribution of the fluorescence data is right-skewed - the
geometric mean and median are similar, the normal mean is generally much
higher. If we wish to see whether there are any statistically
significant differences between the levels of fluorescence of cells from
10 RA patients to 10 controls, do we therefore have to compare either 10
medians to 10 medians, or 10 geometric means to 10 geometric means? I
have a feeling this isn't statistically sound. Are there methods of
analysis which are routinely used for this sort of data - and could
anyone please point me in the direction of a concise text which would
detail these? Otherwise - since the geometric mean and median are
similar, would you agree that choosing the median for each patient is
akin to selecting one representative data point for each, and therefore
possibly suitable data for conventional pairwise comparison in, say, a
Mann-Whitney U test?
Thank-you for taking the time to read this. I will summarise the replies
to the list.
Regards
Liz Hensor
Dr Elizabeth M A Hensor PhD
Data Analyst
Academic Unit of Musculoskeletal and Rehabilitation Medicine
36 Clarendon Road
Leeds
West Yorkshire
LS2 9NZ
Tel: +44 (0) 113 3434944
Fax: +44 (0) 113 2430366
[log in to unmask]
|