You need to do a mediation analysis. To do that, you need some sort of
regression extension that can give you parameters, and you calculate
the indirect effect using thiese parameters.
Trouble is there are no parameters from a K-W test.
If you used a K-W test because of non-normality, you can bootstrap, if
you did it because of ordinal data, then you can use an ordinal
logistic regression. (If I were going to do something like that, I'd
use Mplus, 'cos it's much easier).
None of these solutions are especially straightforward (in case that
wasn't obvious).
Jeremy
On 15 July 2012 22:23, Kennedy, Mark 2004 (PGR) <[log in to unmask]> wrote:
> Hi all,
>
>
>
> Just a quick stats question. I have three groups of participants and have found a difference in behaviour between them (using Kruskal-Wallis - the data are non-parametric).
>
>
>
> I want to know whether this was a consequence of their group (e.g. ADHD/non-ADHD) or whether this behavioural difference is mediated by another variable (e.g. mother-child interactions).
>
>
>
> Mark
>
>
>
>
>
> DISCLAIMER: This email is intended solely for the addressee. It may contain private and confidential information. If you are not the intended addressee, please take no action based on it nor show a copy to anyone. In this case, please reply to this email to highlight the error. Opinions and information in this email that do not relate to the official business of Nottingham Trent University shall be understood as neither given nor endorsed by the University. Nottingham Trent University has taken steps to ensure that this email and any attachments are virus-free, but we do advise that the recipient should check that the email and its attachments are actually virus free. This is in keeping with good computing practice.
|