I have been asked to post this question. If anyone out there is interested in
giving some guidance to Mr Harry Belcher, it would be much appreciated.
Please contact Mr Belcher directly.
Harry Belcher [log in to unmask]
Mr H.J.C.R. Belcher, MS, FRCS (Plast)
Consultant Hand & Plastic Surgeon,
Queen Victoria Hospital,
I only wish for guidance as to the correct statistical test(s).
The study/audit is intended to develop a nice simple model and method of
presentation. I have got fed up with presentations that seek to confuse
with numerous inappropriate graphs and tables. I hope the 3D concept can be
applied to other areas of surgery
In hand surgery, there are three basic types of problem; dysfunction, deformity,
Pre-operative patients have a varying clinical presentations; some conditions
being very painful, others deforming, etc.
Having measured pathology, I want to be able to demonstrate normalisation after
I have designed a Likert-style questionnaire to examine these three areas
(dysfunction, deformity, pain/tenderness) from a patient's perspective
I realise that I can individually analyze/compare values for each parameter to
produce statistics for each measure as shown in my preliminary calculations.
That is inelegant and probably loses power. Much better to present one test
Pre-operatively: what test allows me to test the hypothesis that condition A is
different from condition B simultaneously in this three-dimensional model?
Post-operatively, what (presumably paired) test will allow me to assess the
impact of surgery comparing pre-operative and post-operative scores for the
three measures simultaneously?
I understand my data is non-continuous, non-normal and therefore living in a
Is it possible to calculate confidence limits for the median values?
I cannot fathom what technique to use. I have a suspicion it might be in the
realm of spatial statistics which is rather too sophisticated for a jobbing
I can provide the following documents if needed.
2. Excel spreadsheet containing the raw data.
3. Word document with my preliminary results including some illustrative graphs.
You may leave the list at any time by sending the command
to [log in to unmask], leaving the subject line blank.