Statistical analysis of solid waste composition data: Arithmetic mean, standard deviation and correlation...

Area of Science:

• Environmental Science
• Statistics

Background:

• Solid waste composition data are inherently 'closed' due to the sum constraint (percentages always sum to 100%).
• Classical statistical methods are often inappropriately applied to this compositional data, ignoring its unique characteristics.
• Ignoring the closed nature of waste data can lead to biased interpretations and spurious findings, such as negative proportions.

Purpose of the Study:

• To highlight the statistical challenges posed by compositional waste data.
• To demonstrate the potential for misleading results when closed data characteristics are ignored.
• To advocate for appropriate data transformation techniques before statistical analysis of waste composition.

Main Methods:

• Analysis of fractional solid waste composition data.
• Application of Pearson's correlation test to both absolute mass and percentage values of waste fractions.
• Identification of statistical biases arising from the analysis of closed data.

Main Results:

• Analysis of waste composition data revealed issues like biased negative proportions for certain waste fractions.
• Correlation tests on absolute waste mass indicated positive associations, while tests on percentage composition showed negative associations for the same variables.
• These discrepancies underscore the potential for generating spurious or misleading results when compositional data is not adequately transformed.

Conclusions:

• Compositional waste data requires specific statistical treatment due to its inherent 'closed' nature.
• Standard statistical methods applied without accounting for the sum constraint can produce erroneous conclusions.
• Adequate transformation of compositional data is crucial before performing statistical analyses like calculating means, standard deviations, or correlation coefficients.