BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Effects of ignoring clustered data structures in factor analysis a
 nd item response theory - Dr. Jan Stochl\, University of Cambridge
DTSTART:20120503T130000Z
DTEND:20120503T140000Z
UID:TALK37023@talks.cam.ac.uk
CONTACT:Luning Sun
DESCRIPTION:In research\, data for analysis come principally from two sour
 ces: directly from the respondents themselves and from interviewers/raters
 . In the latter case\, clustering by interviewer/rater needs to be conside
 red when performing analyses such as factor analysis or item response theo
 ry modelling (IRT)\, although it is usually ignored. We use simulated data
  to study the consequences of aggregated analysis (i.e.\, analysis ignorin
 g clustering) on factor analytic estimates (both exploratory factor analys
 is (EFA) and confirmatory factor analysis (CFA)) and fit indices when the 
 data are clustered.\n\nOccasionally\, certain aspects of the hierarchical 
 information on clustering displayed by data are partly known (as in the ca
 se of clustering by patient service or treatment site\, for example). Howe
 ver\, information about the interviewers within each service is likely to 
 be missing. In such cases\, it might be better to consider using the avail
 able information to improve the quality of factor analytic estimates rathe
 r than completely ignoring the hierarchical structure of the data. We stud
 y the usefulness of this approach using simulated datasets. We also study 
 the performance of different estimators - maximum likelihood\, weighted le
 ast squares and Markov chain Monte Carlo - on factor analytic estimates wh
 en hierarchical clustering is ignored.\n\nThe results show that ignoring c
 lustering in the data leads to serious underestimation of the factor loadi
 ngs and item thresholds in ordinal IRT treatment of rating data. In additi
 on\, fit indices tend to show a poor fit for the candidate structural mode
 l. The Markov chain Monte Carlo (MCMC) estimator shows better robustness t
 han the maximum likelihood and weighted least squares approaches. Partial 
 information on clustering helps to correct (and may overcorrect) fit indic
 es\, but unfortunately\, it does not improve the factor analytic model est
 imates themselves.
LOCATION:Seminar Room\, The Mond Building\, New Museums Site
END:VEVENT
END:VCALENDAR