CCSR Cathie Marsh Centre for Census and Survey Research Social Statistics Jose Maria Pina Sanchez Title: Measurement Error in Event History Analysis: Detection, Implication and Adjustment in a Validation Study Funding: Economic & Social Research Council PhD summary: The topic of my thesis is measurement error in survey data, and I am studying it using a three step structure. First I assess the level of measurement error found in retrospective questions asking about work histories. Retrospective questions are subject to recall errors which seem to generate non-conventional patterns of measurement error (i.e. non-classical measurement error). In order to determine the type and magnitude of measurement error I compare survey responses to administrative data from a register of unemployment linked at the individual level. In a second part I assess what are the implications of using durations of unemployment derived from these retrospective questions as the response variable of different event history models. Specifically an accelerated life Weibull model, a proportional hazards Cox model, and a proportional odds logit model are compared in terms of changes in the standard errors and bias in the regression coefficients when data prone to measurement error is used. The last part of my thesis involves the implementation and comparison of methods for the adjustment of measurement error. These methods are regression calibration, multiple imputation for measurement error (MIME), and simulation-extrapolation (SIMEX). Supervisors: Ian Plewis and Johan Koskinen Research interests: I am interested in any topic regarding data quality. Specifically, I am interested in all the aspects of measurement error. How and where does it happen? What are its consequences? And which are the best methods for its adjustment? Another research area of interest is longitudinal data analysis, regarding both the case of data capturing change and event-occurrence. In addition, and although not directly related with my PhD topic I am interested in the fields of fiscal federalism and political decentralization. Biography: I studied Economics as an undergraduate at the Universidad Autonoma de Madrid (Spain) from 2001 to 2005, the following year I did a Ma in Political Economics at the Universite Pierre Mendes (Grenoble, France). From 2007 to 2009 I worked in Expansion Exterior and Quasar Consultores, both of them consulting firms based in Madrid and specialized respectively in management of foreign aid and public policy assessment. In 2009 I joined CCSR as a student of the MSc in Social Research Methods and Statistics, and thanks to an ESRC studentship I continued my studies in CCSR as a PhD student. Contact Details: jose.pinasanchez@postgrad.manchester.ac.uk Publications: Pina-Sánchez, J., Koskinen, J., and Plewis, I. (2012), Measurement Error in Retrospective Reports of Unemployment, CCSR Working Paper. tel +44(0)161 275 4721 http://www.ccsr.ac.uk CCSR Cathie Marsh Centre for Census and Survey Research Social Statistics Conference Papers and Presentations: Royal Statistical Society International Conference, September 2012, Measurement Error in Retrospective Reports of Unemployment. CCSR Tuesday’s Seminar Series, May 2012, Methods for the Adjustment of Measurement Error. CCSR Tuesday’s Seminar Series, February 2012, Measurement Error, an Introduction to the Topic and to its Implications. Teaching: Since 2010 I have been the teaching assistant for Statistical Foundations, and Longitudinal Data Analysis, both compulsory courses from the MSc Social Research Methods and Statistics. I have also been a teaching assistant for 12 short courses at CCSR, including: Introduction to McMC, Bayesian Statistics, and Multiple Imputation in Longitudinal Data. tel +44(0)161 275 4721 http://www.ccsr.ac.uk