Likelihood inference for pollutant loading under type I censoring

Area of Science:

• Environmental Science
• Ecotoxicology
• Statistical Modeling

Background:

• Environmental contaminants negatively impact human and animal health.
• Ecosystems face disruptions in integrity and function due to toxic substances.
• Contaminant concentrations below detection limits are often recorded as 'non-detect', complicating analysis.

Purpose of the Study:

• To develop and compare statistical inference methods for environmental toxicant data below detection limits.
• To address the challenge of analyzing 'non-detect' values in environmental monitoring.
• To apply these methods to real-world environmental data.

Main Methods:

• Utilized exact and modified likelihood methods for the location-scale family.
• Focused on the normal distribution as a special case for comparison.
• Applied the developed statistical procedures to environmental samples.

Main Results:

• Demonstrated the application of statistical inference for 'non-detect' environmental data.
• Provided a comparative analysis of different statistical approaches.
• Successfully applied methods to the Niagara River monitoring dataset.

Conclusions:

• Statistical methods can effectively handle environmental toxicant data with values below detection limits.
• The chosen methods offer robust inferences for ecological and health risk assessments.
• Analysis of the Niagara River data highlights the importance of appropriate statistical treatment for accurate environmental monitoring.