University of Eastern Finland, School of Computing, Statistics

Since 1996 I have been working in the National Drug Monitoring Centre of Finland at Stakes 
(National Research and Development Centre for Welfare and Health) applying the capture-recapture 
technique in estimating the prevalence of amphetamine and opiate use in Finland. I have also been 
nominated as national expert for the harmonisation work on the prevalence estimations co-ordinated 
by the European Monitoring Centre for Drugs and Drug Addiction (EMCDDA), an EU agency based in 
Lisbon. An international connection with drug researchers is possible by the medium of my 
membership in The International Harm Reduction Association (IHRA). 



Dissertation Research: 
Title:   Statistical estimation of illicit use of heavy drugs
Supervisor:  Prof. Juha Alho


The object of the research is to improve the current estimation procedure where the accuracy of 
the estimate is affected by diverse errors arising from the registration and statistical modelling. 
A method known as total error analysis is used for a conceptual analysis of the registration 
mechanism. The bias caused by heterogeneous capture probabilities is corrected by a logistic 
regression analysis and expanding the group of explanatory variables. Beside the frequentist 
technique, the logistic regression method is formulated in terms of a hierarchical Bayesian 
approach. A statistical method of adding random noise to the data is used to maintain the 
individual anonymity. Furthermore, simulation tests are carried out to define the minimum 
amount of transferred background information, which still guarantees valid estimates.

The empirical analyses are based on Finnish data but the approach of the problems 
is such that the statistical results are expected to be applicable elsewhere, as well.

Key words: capture-recapture, logistic regression, heterogeneity, correlation bias,
                    total error, masking

• University of Eastern Finland, Statistics