Re: methodology enquiry on "best-friend" controls

["Andre Engels" <engels@xxxxxxxxxx>]

bill higgins wrote:

> Although others are perhaps more skilled in research
> methodology, the best friend approach seems
> problematic.  Yes, it is pragmatic.  Unfortunately,
> kids select friends for many reasons.  Although they
> may match on SES, they are likely to differ
> (significantly) on many other dimensions.  We often
> select friends to complement (rather than mirror) our
> own personalities.  Therefore, it might not prove to
> be the best control.  

I would personal tend to fear the opposite, that 'best friends' have too
much in common. The reason for this is that there will likely also be a
greater equality in the properties being measured, as well as the background
properties we want to get rid of. Best friends will influence each other
quite a bit, and be chosen among people with similar habits and a similar
look on life. Thus a child who is not interested in learning and because of
that gets low grades, will tend to get a best friend who also prefers other
things and is not the brightest in schoolwork. A child who is being depressed
or otherwise not feeling well will probably have an effect on his/her best
friend who will also feel slightly worse.


-- 
Andre Engels, engels@xxxxxxxxxx
Telephone: +31(0)6-11181398 (GSM)
http://www.win.tue.nl/~engels/index_en.html
PGP Public key: see http://www.win.tue.nl/~engels/pgp.asc

If we don't believe in freedom of expression for people we despise,
we don't believe in it at all			-- Noam Chomsky