The Department for Transport (DfT) publishes annual and quarterly estimates of traffic on Great Britain's roads here . DfT traffic estimates are based on observations of traffic on samples of road links, defined as a stretch of road between two junctions. The traffic observations are collected by a panel sample of Automatic Traffic Counters and by an annual sample of around 7,000 manual counts. A fundamental component of the traffic estimation process is the stratification of the sample of Automatic Traffic Counters. The aim of stratification is to minimise sampling error by grouping together roads with similar traffic patterns. The main aspect of this project was to explore whether the current stratification groups are the best categorisation to group together road links with similar traffic flow patterns. The approach taken by the project was to look at the effects of certain road attributes on fluctuations in traffic flow. Those attributes found to affect traffic flow could be used as divisors for the stratification categories. This document provides a detailed overview of the stratification project, including statistical methods and results . he current stratification was introduced in the early 2000s . It comprises of a large number of categories, some of which have quite a small sample of Automatic Traffic Counters (ATCs) within them meaning that estimates for these categories can be quite noisy . Analyses The project used data from the DfT's network of 200 ATCs, allowing comparisons across different times of the day, different times of year and between different years to be made. Exploratory analyses aimed to determine what road characteristics are important with regards to variations in traffic flow. The analyses are split into two main sections : the first looked at variations within the year, which relates to expansion factors ; and the second at variations between years, which related to growth factors. The analysis investigated the level of variation in traffic observed at each ATC, as well as variations between day and night, weekday and weekend, seasonal variations and variations in flow between years. All analyses were peer reviewed by an external methodological expert from the Office for National Statistics. The analyses found that different road attributes were significant for within year variation to between year variations. Therefore , two different stratifications are necessary. The recommendation from the project is one set of strata that can be aggregated up in two different ways for the calculation of expansion factors and growth factors. 1 Trunk Motorway 2 Principal Motorway 3 Trunk Urban A roads <20,000 4Trunk Urban A roads >20,000 5Trunk Rural A roads <20,000 6Trunk Rural A roads >20,000 7Principal Urban A roads <20,000 8Principal Urban A roads >20,000 9Principal Rural A roads <20,000 10Principal Rural A roads >20,000 11 London A roads 12 London B roads 13 London CU roads 14 Urban B roads 15 Urban CU roads 16 Rural B roads 17 Rural CU roads. The next step is to allocate each of the ATCs to a stratum and calculate expansion factors an d growth factors and assess the effect that these have on overall traffic estimates. (Author/publisher)
Samenvatting