Textruta: Anders Nordgaard 






Ø    Home

Ø    Personal presentation

Ø    CV

Ø    Research

Ø    Teaching


Textruta: I work as forensic specialist in statistics at the Swedish National Laboratory of Forensic Science in the city of Linköping Sweden [in Sweden known as Statens Kriminaltekniska Laboratorium (SKL) ]. I also work part-time as a senior lecturer in Statistics at Linköping University (same city), Department of Computer and Information Science, where I teach theoretical courses and supervise and examine Master’s theses. (I worked full-time as a senior lecturer at the same department (and earlier at the Department of Mathematics) between the years 1996 and 2010.) The subject of statistics in Sweden is from an international point of view to be classified as social science statistics, while the subject mathematical statistics is taught at faculties of natural sciences and technology. I wrote my PhD in mathematical statistics but the majority of my teaching history is at the faculty of arts and sciences.
A statistician at a forensic laboratory is fully occupied with problems and questions covering the whole range from elementary advice about the design of laboratory experiments to philosophical discussions about the applicability of probabilistic reasoning in criminal cases. To be successful it is not only important to have a firm theoretical ground in the subject of statistics/mathematical statistics, but also to be open-minded and willing to learn about other scientific disciplines, in particular chemistry and biology.  In addition, basic knowledge about law and the judicial system of the country is essential. 
My key interests within forensic science are evidence evaluation, quality assessment and sampling from seizures. I fully adopt the Bayesian approach to evidence evaluation which means I work a lot with development of methods for estimating a likelihood ratio in particular case work. At the laboratory I have taken part in the development of a unified ordinal scale of conclusions based on the likelihood ratio with scale levels corresponding with carefully selected intervals of likelihood ratios.
Does this mean I call myself a Bayesian? I wouldn’t say so simply because I do not concur with the opinion that you are either a frequentist or a Bayesian. Modern statisticians have learnt that it is the problem itself that decides whether a Bayesian or a frequentistic approach is the most efficient to use. At a forensic laboratory, the former dominates due to the often very small amount of measurements available, but also because the inference made from forensic findings in a particular case must be possible to chain at the court level no matter if they are of completely different kinds. However, am I asked to estimate the mean salary in a population from a random sample of 5000 individuals and give a standard error of this estimate, I would probably never consider any prior for this mean.
The “community” of Forensic Statisticians in Europe is very small (it is even so in the entire world). Lots of the development I take part in therefore must have support in international co-work. I am a member of the FORSTAT Research Group, which is a group with membership by invitation. One of our main tasks is to organise the annual FORSTAT workshop which gather forensic scientists all across Europe to a couple of days with tutorials on statistical methods in forensic science. 
Besides my duties at the laboratory and at the university I am also involved in some projects conducted by the Joint FAO/IAEA Division on Nuclear Techniques in Food and Agrigulture (sited in Vienna, Austria). My task within these projects is to act as an expert in statistics in the development of models for monitoring Good Agricultural Practice including efficient sampling schemes.

In my leisure time I prefer to be with my family (wife and two grown-up children), practise cross-country-skiing/running and sip a glass of peaty Scotch whisky.