“Longitudinal Data Modelling with High Attrition”
The Centre
has developed a strong expertise in longitudinal data collection and analysis
through many retrospective surveys and through its collaboration with various Health
and Demographic Surveillance System (HDSS) in Africa. HDSS make intensive use
of large longitudinal datasets collected over many years. The Centre wishes to
strengthen its capacity in longitudinal data analysis by recruiting a PhD
fellow to work on the specific problem of controlling for non-random attrition (informative
censoring) to produce estimates of morbidity, mortality, fertility and other
socio-demographic indicators that make no or weak assumption of independence between censoring and event.
The PhD
candidate is expected to apply and to test the validity of various existing statistical
models that deal with informative censoring, and possibly to develop an
original model suitable for HDSS data. Some existing models address the issue
of right censoring (attrition) but the new model should also tackle the issue
of left-censoring (mainly through in-migration to the site). Models should be
tested on longitudinal data collected by one or two HDSS, with possible
extension to the member sites of the International
Network for the Demographic Evaluation of Populations and Their Health in
Developing Countries (INDEPTH). The results are expected to contribute
significantly to the understanding of the health conditions prevailing in the
sites, net of migration effect.
The PhD
candidate will participate actively to HDSS data analysis and management. S/he
will be expected to make regular contribution to the improvement of data
collection and quality monitoring. S/he will be expected to contribute to
research capacity building. The candidate should have:
-
A
Masters degree in Statistics, Demography, Epidemiology or related fields;
-
An
interest in, or practice of, maximum likelihood programming;
-
An
experience of longitudinal data analysis; if not, the candidate should commit
to a technical training in event history analysis;
-
A
wish to spend most of his/her time at the Université Catholique de Louvain,
Belgium, with several visits to the HDSS sites, in particular in South Africa,
Kenya, and Burkina Faso;
-
An
ability to work in a team.
The
PhD supervisor, Prof Philippe Bocquier (philippe.bocquier@uclouvain.be),
will receive letters of motivation and CVs up to December 30th 2011.
The selected PhD candidate with the support of her/his supervisor will submit her/his
application to various grant schemes including the Fonds National de la Recherche
Scientifique (FNRS) in Belgium.
