Some Issues in the Statistical Analysis of Vehicle Emissions

This paper presents some of the issues that complicate the statistical analysis of real-world vehicle emissions and the effectiveness of emissions control programs. The following issues are discussed: 1) inter- and intra-vehicle emissions variability, 2) skewness of the distribution of emissions from in-use vehicles, 3) the difficulty of obtaining statistically representative vehicle samples, 4) the influence of repeat testing on only a subset of the vehicle fleet, and 5) differences among common test methods and pollutant measurement devices. The relevance of these issues is discussed in light of three regulatory purposes: testing the compliance of in-use vehicles with certification standards, evaluating the effectiveness of vehicle inspection and maintenance programs, and estimating emissions inventories for air quality modeling and compliance planning. A brief history and description of common vehicle emissions tests is also provided.