In message <[log in to unmask]>,
[log in to unmask] writes
>I would appreciate some advice from the list on handling paired data. I have
>had unequal amounts of data from subjects. In a small pilot study of a course
>for teaching communication skills, subjects were taped and scored for their
>use of taught (micro)skills. Some managed to record the requested number of
>consultations but some did not. I realise it is an error to treat each set of
>data as an individual data point, should I average each subject's change
>scores,
>but then this seems to overvalue the data for those who contributed fewer
>consultations?
>Is it valid to average whole group scores to look for a global shift?
>
A quick response as off to work.
You don't have paired data here, nor can you treat the scores as
continuous (ie like height or weight). They are really ordered
categorical (ie you know a change of 4 points is larger than a change of
2 points, but you can't say it's twice as large). You should therefore
be using a non-parametric method of analysis (I am assuming you have 2
groups?).
Must rush
Toby
--
Toby Lipman
General practitioner, Newcastle upon Tyne
Northern and Yorkshire research training fellow
Tel 0191-2811060 (home), 0191-2437000 (surgery)
Northern and Yorkshire Evidence-Based Practice Workshops
http://www.eb-practice.fsnet.co.uk/
|