PhD Studentship Calibration studies for the analysis of multiple datasets School of Medicine, Division of Epidemiology and Health Sciences, Faculty of Medical and Human Sciences, University of Manchester. There are an increasing number of research questions that can only be answered by the use of multiple datasets. Sometimes this is for simple reasons of increased power, for example in meta-analysis, but sometimes it is because the effect of interest is not identified within any one of the contributing datasets alone. A good example of the latter is in the analysis of time trends from repeated cross-sections of data. A common problem in performing such synthetic analyses is the lack of a uniform measurement protocol across studies. Calibration samples that allow the mapping of one measurement protocol to another provide one way forward. We have already used latent variable modelling and multiple imputation as alternative techniques to undertake the synthetic analysis of study and calibration samples. Their application has led to high profile findings on trends in adolescent mental health. The proposed programme of work would review this work, particularly in relation to alternative synthetic estimation models and the design and relative efficiency of calibration samples. Obvious target studies would be key measures from the UK's national birth cohorts (1958, 1970 & 2001) (AP is a member of the steering ctee for these studies). A calibration sample would be designed and collected for a range of target variables and exploited in the carrying out of novel analyses. For further information, interested applicants should contact Marlene Middleton marlene.middleton@manchester.ac.uk or Christine Burns christine.burns-2@manchester.ac.uk The deadline for applications is Friday 21st October with interviews to be held Monday 7th November.