methodology enquiry on "best-friend" controls

["JA Dennis, School Policy Studies" <J.Dennis@xxxxxxxxxxxxx>]

While this is not child-maltreatment as such the methodological aspects
of the following enquiry can perhaps best be addressed by asking this 
group.  Would appreciate any advice/views.

A colleague is looking at the psychosocial impact of narcolepsy on
children and adolescents. He has a battery of questionnaires which he 
has administered to children with the diagnosis. As part of the q'aire 
he has got permission to contact their teachers in order to get
  1 - a set of information about their schoolwork;
  2 - the name and address of the next age/sex matched child on the
school register so that he can approach them to be a control.
 
The problem is this. Teachers are somewhat overloaded and responses 
from them have been poor. When he has got a control name and address 
responses have been similarly poor. His U.S-based collaborators are
getting round this problem by asking the index chilren to get their 
best friend to complete the q'aires which the USA people say is better 
because you tend to get the same socio-economic group etc.
 
However, it may be that narcoleptic children might perhaps attract
'unusual' friends and this may make it more difficult for him to detect
a difference between them and these controls on some measures e.g.
self-esteem, depression, health, functioning. 

Does anyone have a view on this issue of 'best friend controls'? He 
doesn't want referees (assuming something worth publishing emerges) 
complaining that the methodology was flawed because of this. It is 
attractive to use best friends for pragmatic reasons... doesn't want to
find himself short of controls..... he reckons he could always compare 
the narcoleptic kids with norms (where good ones exist) but would 
prefer age-sex matched contemporaneous controls for all the usual 
reasons.
 
Any views please email the list and/or 
paul.montgomery@psych.ox.ac.uk
(Paul Montgomery, Park Hospital, University of Oxford, UK)

*************************************************************