> Hi all,
>
> I have conducted a small scale longitudinal study on co-resident carers of
> people with dementia, looking at the factors (stressors, psychosocial
> resources) related to perceptions of burden and distress over a 6-month
> period. I want to look at whether changes in the DVs (burden and distress)
> are related or predicted by changes in the IVs (stressors and psychosocial
> resources). I was advised to run a series of multiple regression analyses
> whereby the DV at time 2 is predicted by the raw difference score of each
IV
> and the DV at time 1 (used as a covariate) but I am not sure whether this
is
> appropriate given that my sample is fairly small (N= 35) and raw
difference
> scores do not remove the linear effects of initial (time 1) scores on the
> index of change that is analysed. Several studies I have read have used
> residualised change scores to avoid the latter problem but I am not clear
as
> to how this is done and whether it would be appropriate for my analyses.
Any
> advise on this issue would be greatly appreciated!
>
> many thanks,
>
>
> Paul Hutchison
>
>
|