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:Cambridge Psychometrics Centre Seminars
SUMMARY:Confidence intervals for IRT models: three new exa
ct approaches meeting different criteria of optima
lity - Anna Doebler - WWU Münster
DTSTART;TZID=Europe/London:20100527T100000
DTEND;TZID=Europe/London:20100527T110000
UID:TALK24987AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/24987
DESCRIPTION:A common way to calculate confidence intervals in
a testing situation is to assume that the standard
ised maximum likelihood estimator for the person p
arameter is normally distributed. Although maximum
likelihood estimators are approximately normally
distributed\, this assumption is often questionabl
e for short and medium test length. For example\,
when working with the Rasch model and a test lengt
h of n\, there are only n + 1 different possible o
utcomes of the estimation and so the distribution
of the standardized estimator will be quite discre
te. Another disadvantage of the standard approach
is that even when the normality assumption holds\,
it is not clear if the obtained confidence interv
als meet any criterion of optimality\, for example
regarding some notion of "smallness”. A third und
esirable property is that they will always depend
on the specific choice of the estimator. Therefore
\, we will try to define confidence intervals in a
more "exact" way by making use of the relationshi
p between confidence intervals and hypothesis test
ing. We will outline three different approaches fo
r confidence interval constructions and discuss ho
w far these meet different optimality criteria.
LOCATION:Seminar Room\, The Mond Building\, New Museums Sit
e
CONTACT:Michal Kosinski
END:VEVENT
END:VCALENDAR