Sorry, it has just occurred to me, that if I have recoded them all into
z-scores then I could just multiply those scores which I need to reverse
by -1?
On 12/07/2016 14:00, Helen Foster-Collins wrote:
> Hi there,
>
> I am wanting to carry out an exploratory factor analysis on a number
> of outcome variables from ten cognitive tasks.
>
> However, the outcomes from these tasks are measured using various
> different scales (e.g. number of errors, number of items recalled,
> response times, etc.)
>
> I thought that if I turn all these outcomes in z-scores then this
> would overcome the problem of different scales.
> However, I am also thinking that it would be good to recode the
> variables in the same 'direction' for my output to be interpretable.
> So, for instance, I could recode Errors as negative, so that high
> errors would indicate poorer performance on that task.
>
> Does that sound correct?
> And how would I approach something like response times, where
> smaller/quicker response times are better than longer response times?
> Could I just work out what my maximum is and take each score away from
> that?
>
> Many thanks,
> Helen
>
|