Abstract
In this talk I will present recent examples of statistical analysis from the
Research Department at Division of Mental Health and Addiction, Oslo
University Hospital.
There is a large potential for use of more sophisticated statistical tools
in psychiatric research. Methodological aspects of design, measurement error
and causal inference can probably contribute to improved analysis in a field
where most of the uncertainty cannot be avoided.
The examples in this presentation is two ongoing projects where the
statistical methodology has played an important role. The first project is
based on an innovative design. Most patients (like everybody else) carry a
cell phone with them at all time. With an "APP" specifically designed for
the purpose of data - collection, they can be prompted regularly to report
for example on different emotions during the day. I will show results from a
comparison of such emotion time series between two diagnostic groups
(submitted paper).
The second project is an application of a repeated - measures marginal
structural model to answer the causal question: does therapeutic
interventions on changeable aspects of personality problems improve
psychosocial functioning (work in progress)?