Bootstrap Estimation of the Properties of Estimators in a Road Freight Transport Survey

Sammanfattning

Road freight transport represents a large part of the economic activity in the EU and simultaneously results in a large environmental impact, making accurate data important for decision-making in line with environmental goals. To this end, surveys are carried out for the collection of such data which is used for estimates and by extension the production of official statistics. After collecting survey data for such use, it is important to evaluate how accurate the estimates made are, which is what this study aims to do using multiple bootstrap methods (doubled half, non-parametric, pseudo-population). The objects of study are three different parameters (goal-quantities) related to goods transport. These parameters are each a total of a large set of products between a measured value and a separately estimated coefficient collected through the survey. To study the properties of the estimators for these parameters, a set of confidence intervals and margins of error are estimated. In this study, the uncertainty of the estimators was estimated as lower than the maximum level of uncertainty allowed by Eurostat for these estimates. The estimates of uncertainty were lower in this study compared to the original survey study it is based on, which used Horvitz-Thompson estimators instead of bootstrap estimators.