BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Measurement error modelling through SEM: Applications in epidemiol
 ogy and health - Tim Croudace (Dept of Psychiatry)
DTSTART:20110124T160000Z
DTEND:20110124T173000Z
UID:TALK28724@talks.cam.ac.uk
CONTACT:Mandy Carter
DESCRIPTION:Abstract Structural Equation Modelling (SEM) is a statistical 
 framework for modelling with sets of simultaneous regression equations (pa
 th analysis) and latent variables (constructs that can only be defined ind
 irectly\, from indicators). Within SEM is it possible\, and extraordinaril
 y useful to understand and account for measurement error in "observed" var
 iables\, and to study the reliability and validity of target constructs th
 at can only be measured as "latent" variables. SEM provides a framework fo
 r estimating scores on construct that otherwise cannot be measured\, and i
 n doing so corrects for error (through a model-based approach). SEM also d
 eals with multiple response problems (multivariate outcomes and longitudin
 al data) and can be extended to account for missing and multilevel data an
 d most recently to accomodate selection mechanisms that might be essential
  in dealing with confounding effects.SEM examples will be discussed.
LOCATION:Seminar Room\, MRC Cognition and Brain Sciences\, 15 Chaucer Road
 \, Cambridge
END:VEVENT
END:VCALENDAR