BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//talks.cam.ac.uk//v3//EN
BEGIN:VTIMEZONE
TZID:Europe/London
BEGIN:DAYLIGHT
TZOFFSETFROM:+0000
TZOFFSETTO:+0100
TZNAME:BST
DTSTART:19700329T010000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0100
TZOFFSETTO:+0000
TZNAME:GMT
DTSTART:19701025T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
CATEGORIES:Isaac Newton Institute Seminar Series
SUMMARY:Optimising the allocation of participants in a two
 -stage randomised experiment to estimate selection
 \, preference and treatment effects - Walter\, S (
 McMaster)
DTSTART;TZID=Europe/London:20110902T163000
DTEND;TZID=Europe/London:20110902T170000
UID:TALK32622AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/32622
DESCRIPTION:Experimental outcomes may be affected by the choic
 e of treatment that participants might make (if th
 ey were indeed allowed to choose)\, a so-called se
 lection effect\, and by whether they actually rece
 ive their preferred treatment\, a so-called prefer
 ence effect. Selection and preference effects can 
 be important (possibly even larger than the usual 
 treatment effect)\, but they cannot be estimated i
 n conventional randomised experimental designs. \n
 \nAn alternative approach is the two-stage randomi
 sed design\, in which participants are first rando
 mly divided into two subgroups. In one subgroup\, 
 participants are randomly assigned to treatments\,
  while in the other\, participants are allowed to 
 choose their own treatment. This approach can yiel
 d estimates of the direct treatment effect\, and o
 f the preference and selection effects. The latter
  two provide insight that goes considerably beyond
  what is possible in standard randomised experimen
 ts\, notably the usual parallel group design. \n\n
 In this presentation\, we will consider the optima
 l proportion of participants who should be allocat
 ed to the choice subgroup and allowed to determine
  their own treatment. The precision of the estimat
 ed selection\, preference and treatment effects ar
 e functions of: the total sample size\; the propor
 tion of participants allocated to choose their tre
 atment\; the variances of the response (or outcome
 )\; the proportions of participants who select eac
 h treatment in the choice group\; and the selectio
 n\, preference and treatment effects themselves. W
 e develop general expressions for the optimum prop
 ortion of participants in the choice group\, depen
 ding on the inverses of these variances\, and on w
 hich effects are of primary interest. We illustrat
 e the results with trial data comparing alternativ
 e clinical management strategies for women with ab
 normal results on cervical screening. \n
LOCATION:Seminar Room 1\, Newton Institute
CONTACT:Mustapha Amrani
END:VEVENT
END:VCALENDAR