BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Conflicts between optimality criteria for block designs with low r
 eplication - Rosemary Bailey\, Queen Mary University of London
DTSTART:20130125T160000Z
DTEND:20130125T170000Z
UID:TALK42610@talks.cam.ac.uk
CONTACT:Richard Samworth
DESCRIPTION:In designing experiments\, we usually seek to make variances\n
 of estimators small.  For incomplete-block designs\, various ways of\nmeas
 uring multi-dimensional variance give the A-\, D-\, and E-optimality\ncrit
 eria\, which I shall explain.  If there exists a\nbalanced incomplete-bloc
 k design for the given parameters\, then it is\noptimal on all these crite
 ria.  It is therefore natural to use the\nproxy criteria of (almost) equal
  replication and (almost) equal\nconcurrences when choosing a block design
 .\n\nHowever\, work over the last decade for block size 2 has shown that\n
 when the number of blocks is near the lower limit for estimability of\nall
  treatment contrasts then the D-criterion favours very different\ndesigns 
 from the A- and E-criteria.  In fact\, the A- and E-optimal\ndesigns are f
 ar from equi-replicate and are amongst the worst on the\nD-criterion.\n\nI
  shall report on current work which extends these results to all\nblock si
 zes.  Thus the problem is not blocks of size 2\; it is low\nreplication.\n
LOCATION:MR12\, CMS\, Wilberforce Road\, Cambridge\, CB3 0WB
END:VEVENT
END:VCALENDAR