Fri Nov 17 08:36:45 2006
========================

Judy - John L

concerned about sample bias in alum survey (only email surveys, we
have only 1/2 the alums survey)

can we create a statistically valid random sample, enable us to make
inferences about the whole population 

"how do you figure out what the right strata are"

we think we know but how do we know if we're right?

at least 3 poss's:

1. cavalier's delib democ project
   they did a delib poll in jan 05 on reunion...
   is this 

   * need a reliable not hard way to select strata 

2. create an alum panel (subj to 2) like gallup or harris or nielson,
   they are in for 2 yrs, periodically go to them with q's
   (representative?) 

3. alum attitudinal survey [probably going to a consultant firm]

-----------

is the data in existing perception and interest surveys right to
determine the right strata?

1. tripp-umbach survey (fall 02)...  [avail]

2. fall04 rear group survey [not so avail - trust]
 
3. spr06 rear group survey  [not so avail - trust]

"constructing an appropriate sampling frame, strata and so forth..."


42,640 rec'd the emailed survey in 06
 8,922 resp'd to at least one q...

how does email possession affect response?

response skews too young... can alum relats be more powerful to bring
more alums in?

1967 merger with mellon inst is a big line between alums... prior was
touchy feely, posterior was "dismal"...