Re: checking for differences between independent ICCs

Angela Digout wrote:
> Sorry...I should have been more clear. :-)
>
> *snip*
>
> > If this is what you mean, how do you determine the placement
> > of friends in columns?
> >
> > Cheers,
> > Bruce
>
>
> My data file looks like this:
>
> ID/case   mygender      friend's gender  my score  friend's score
> 1               f       f               10              11
> 2               f       f               12              13
> 3               m       m               9               10
> 4               m       m               5               8
> ...
>
> Essentially, I have pairs of same-sex friendship dyads, and I am looking
> at the association between individual scores and their best friend's
> scores. I can't use Pearson correlations, because individuals could
> initially be in either column (as an individual or a friend), hence the
> intra class correlations. I have already removed doubles from the data,
> and split my data file by gender, so I end up with intra-class
> correlations for female dyads (say, .27) and male dyads (say, .15). I want
> to test if there is a significant difference between these "correlations".
> Should Fisher's transformation work for ICC as it would for Pearson's r?
>
> Any help would be appreciated!
> Ange


Try a Google Groups search on "test for differences between 
ICCs".  A couple of the posts you'll find give references.

--
Bruce Weaver
[EMAIL PROTECTED]
www.angelfire.com/wv/bwhomedir/