Employing components-of-variance to evaluate forensic breath test instruments

Abstract 

The evaluation of breath alcohol instruments for forensic suitability generally includes the assessment of accuracy, precision, linearity, blood/breath comparisons, etc. Although relevant and important, these methods fail to evaluate other important analytical and biological components related to measurement variability. An experimental design comparing different instruments measuring replicate breath samples from several subjects is presented here. Three volunteers provided n=10 breath samples into each of six different instruments within an 18 minute time period. Two-way analysis of variance was employed which quantified the between-instrument effect and the subject/instrument interaction. Variance contributions were also determined for the analytical and biological components. Significant between-instrument and subject/instrument interaction were observed. The biological component of total variance ranged from 56% to 98% among all subject instrument combinations. Such a design can help quantify the influence of and optimize breath sampling parameters that will reduce total measurement variability and enhance overall forensic confidence.

Keywords: Breath alcohol, Components-of-variance, Analysis-of-variance, Measurement variability