Publication detail

Evaluation of contamination data with non-detects using censored distributions

FUSEK, M. MICHÁLEK, J. VÁVROVÁ, M.

Original Title

Evaluation of contamination data with non-detects using censored distributions

Type

journal article in Web of Science

Language

English

Original Abstract

When measuring concentration of chemical compounds, we often have to deal with a~situation when the resulting values are found below the limit of detection or limit of quantification of the determination method. In order to statistically evaluate such data, the method of maximum likelihood considering doubly left-censored samples is applied. As a model distribution of measured concentrations, Weibull distribution is considered. Moreover, considering the asymptotic properties of maximum likelihood estimates, concentrations of chemicals can be compared using Wald's test based on the expected Fisher information matrix. Here we show that the described statistical method allows for a better evaluation of the obtained experimental data than commonly used methods where all values below the detection limits are replaced by a~constant. These methods are used for an analysis of the worldwide commonly used synthetic musk compounds (nitro and polycyclic) which were extracted from the fish samples caught upstream (Group 1) and downstream (Group 2) from a high-capacity wastewater treatment plant.

Keywords

Doubly left-censored sample, maximum likelihood, musk compound, Wald's test, Weibull distribution

Authors

FUSEK, M.; MICHÁLEK, J.; VÁVROVÁ, M.

RIV year

2015

Released

3. 12. 2015

ISBN

1018-4619

Periodical

Fresenius Environmental Bulletin

Year of study

24

Number

11c

State

Federal Republic of Germany

Pages from

4165

Pages to

4172

Pages count

8

BibTex

@article{BUT115693,
  author="Michal {Fusek} and Jaroslav {Michálek} and Milada {Vávrová}",
  title="Evaluation of contamination data with non-detects using censored distributions",
  journal="Fresenius Environmental Bulletin",
  year="2015",
  volume="24",
  number="11c",
  pages="4165--4172",
  issn="1018-4619"
}