Particulate matter, a.k.a. particle pollution, is a complex mixture of small particles and liquid droplets that are present in the air. Once inhaled, these particles can affect the heart and lungs and cause serious health problems. A recent study, based on geographically referenced datasets of pollutant emissions has shown that non-exhaust related pollution is at present dominant and increasing. Emissions from paved roads are poorly estimated due to the lack of knowledge about the resuspension process. Recent literature works have attempted to provide a reliable framework for the estimation of emission factors. Estimations are obtained by linear regression with a single-valued discriminant for the acceptance/rejection of the experimental dataset based on the evaluation of the r-squared coefficient. In this paper, we explore alternative methods to evaluate the "quality" of the data and consequently discriminate whether a given sample can be accepted to provide estimation of the emission factors. Uncertainties are characterised both in the data and in the statistical model. Measurements are expressed with interval-valued datapoints to include the experiment precision directly within the estimation process. Alternative fitting techniques that avoid the use a single-valued discriminant are also explored for an inclusive estimation of the emission factors.