|Also available as an Acrobat File|
Visualisation of historical events using Lexis pencils
2. The need for new visualisation methods for event histories
Methods for the statistical analysis of event history data have been extended dramatically in the last twenty years. In the example above, models could be developed for the probability of reoffending or reconviction as a function of the age of the offender, current family and work history, and previous criminal history. This in general will lead to survival model methodology, and if multiple reoffending within an individual is taken into account, to survival frailty models where there is an additional unknown factor - the frailty - measuring the propensity of the individual to reoffend.
However, there has been little development of graphical methods to visualise an event history dataset before analysis. A recent book on event history analysis (Blossfeld et al., 1989) devotes less than 5 pages to the graphical presentation of event history data. A state-of-the-art book on statistical graphics also presents only one graphic on event data (Cleveland, 1994). This is partly because of the complexity of such datasets; it is easy to be overwhelmed with the number of variables and different dimensions of a typical study. This work intends to rectify this imbalance by exploring the potential of modern interactive scientific visualisation systems for the initial graphical examination of complex event history data. By this, we mean that we want to provide graphical displays which can display the full complexity of a dataset if this is what is required.
Graphics Multimedia Virtual Environments Visualisation Contents