Hi Louise
2008/12/1 Louise Anthropos <[log in to unmask]>:
> Hello. I am interested in conducting a pre and post analysis of some people
> who are involved in an intervention. Some of the people may be involved in
> the therapy for a couple of months, others for three, four or five months. I
> am particularly interested in looking at the outcome scores (measured on an
> ordinal scale so parametric analysis seems to be indicated). I would like some
> advice about the best way to go about it, is it okay to look at everyone's
> start and finish scores on the outcome measure?
>
How ordinal are the data? Almost all data in psychology are strictly
ordinal, but most of the time we don't worry about it. You'll have a
much easier time if you can stretch an assumption and call them
interval.
> From a research question point of view, I am really interested in finding out
> about the average change over time, in a global sense to get a 'quick and
> dirty' sense of change. The intervention is necessarily tailored to client need
> so some people will always be involved for a shorter time than others.
> Moreover, I will not know this information ahead of time. I would like to get an
> overall picture and the time in the intervention isn't a significant concern at
> this stage. In the future, when I have more data, I will at that point be
> interested in short term, medium term and long term effects but not for now.
>
> Is it okay to look at pre and post scores in this macro sense using non
> parametric analyses?
>
Yes. Or maybe parametric too. With non-parametric you'll have a very
hard time assessing predictors of change, which might be interesting.
> Thank you in advance for reading. This list serve is a wonderful means to
> learn so much (especially from Jeremy). Much appreciated by us post grads.
>
:) Thanks.
Jeremy
--
Jeremy Miles
Learning statistics blog: www.jeremymiles.co.uk/learningstats
Psychology Research Methods Wiki: www.researchmethodsinpsychology.com
|