Get access

Bayesian Adaptive Approach to Estimating Sample Sizes for Seizures of Illicit Drugs

Authors

  • Rossana Moroni Ph.D.,

    1. National Bureau of Investigation Forensic Laboratory, Jokiniemenkuja 4, 01370 Vantaa, Finland.
    2. Åbo Akademi, Department of Mathematics, Biskopsgatan 8 FIN-20500, Turku, Finland.
    Search for more papers by this author
  • Laura Aalberg Ph.D.,

    1. National Bureau of Investigation Forensic Laboratory, Jokiniemenkuja 4, 01370 Vantaa, Finland.
    Search for more papers by this author
  • Tapani Reinikainen Ph.D.,

    1. National Bureau of Investigation Forensic Laboratory, Jokiniemenkuja 4, 01370 Vantaa, Finland.
    Search for more papers by this author
  • Jukka Corander Ph.D.

    1. Åbo Akademi, Department of Mathematics, Biskopsgatan 8 FIN-20500, Turku, Finland.
    2. Department of Mathematics and Statistics, Helsinki University, P.O. Box 68, FIN-00014, Helsinki, Finland.
    Search for more papers by this author

  • Presented at the 20th International Symposium on the Forensic Science, September 5–9, 2010, in Sydney, Australia.

  • Supported by Grant No. 121301 from the Academy of Finland.

Additional information and reprint requests:
Jukka Corander, Prof., Ph.D.
Department of Mathematics and Statistics
P.O. Box 68
University of Helsinki
00014 Helsinki
Finland
E-mail: jukka.corander@helsinki.fi

Abstract

Abstract:  A considerable amount of discussion can be found in the forensics literature about the issue of using statistical sampling to obtain for chemical analyses an appropriate subset of units from a police seizure suspected to contain illicit material. Use of the Bayesian paradigm has been suggested as the most suitable statistical approach to solving the question of how large a sample needs to be to ensure legally and practically acceptable purposes. Here, we introduce a hypergeometric sampling model combined with a specific prior distribution for the homogeneity of the seizure, where a parameter for the analyst’s expectation of homogeneity (α) is included. Our results show how an adaptive approach to sampling can minimize the practical efforts needed in the laboratory analyses, as the model allows the scientist to decide sequentially how to proceed, while maintaining a sufficiently high confidence in the conclusions.

Get access to the full text of this article

Ancillary