overview
teaching
research
consultancy
other
contact
about this site


Koning A.J. (1999).
Goodness of fit for the constancy of a
classical statistical model over time.
Econometric Institute Report EI9959/A,
Erasmus University Rotterdam,
The Netherlands.
[PostScript Version]
The classical statistical model relates to n
independent random variables having a common distribution.
In this paper we consider the situation where the common
distribution involves an unknown parameter, and where at
time 0<t<1 only the first [nt] random
variables are observed. The innovation approach is used to
derive goodness of fit processes which especially detect
alternatives under which the unknown parameter does not
remain constant, but varies over time.
The behaviour of these processes is investigated under the
null hypothesis as well as under alternative hypotheses.
Limiting Pitman efficacies of supremum type tests based on
these processes are evaluated. Fixed change point
alternative hypotheses and smooth alternative hypotheses
receive additional treatment.
The methods are exemplified using covariance structure
models, especially Gaussian graphical models.

Koning A.J. (1999).
Model based control charts in stage 1 quality control.
Econometric Institute Report EI9958/A,
Erasmus University Rotterdam,
The Netherlands.
[PDF Version]
In this paper a general method of constructing control
charts for preliminary analysis of individual observations
is presented, which is based on recursive score residuals.
A simulation study shows that certain implementations of
these charts are highly effective in detecting assignable
causes.

Hjort, N.L., Koning, A.J. (1999).
Tests for constancy of model parameters over time.
Statistical Research Report 0399,
University of Oslo, Norway.
[PostScript Version]
Suppose that a sequence of data points follows a
distribution of a certain parametric form, but that one or
more of the underlying parameters may change over time.
This paper addresses various natural questions in such a
framework. We construct canonical monitoring processes
which under the hypothesis of no change converge in
distribution to independent Brownian bridges, and use these
to construct natural goodnessoffit statistics. Weighted
versions of these are also studied, and optimal weight
functions are derived to give maximum local power against
alternatives of interest. We also discuss how our results
can be used to pinpoint where and what type of changes have
occurred, in the event that initial screening tests
indicate that such exist. Our unified largesample
methodology is quite general and applies to all regular
parametric models, including regression, Markov chain and
time series situations.
