Survey and sampling strategies.

Data are an essential component of transport modelling. The collection of data is therefore a significant activity of the transport planner. It is also an expensive activity, so that careful design and planning of survey instruments and procedures is essential, to reduce costs and increase the effectiveness of the data to be collected. Data are needed for three main purposes: description of the present situation; input to development and use of transport models; and monitoring the effects of the implementation of policies, strategies, and investments. Transport-data collection normally involves a number of different data-collection activities, each using different survey methods, instruments, and sampling procedures. The survey methods typically encountered in transport applications are reviewed. All surveys need to be based on the application of strict sampling theory (Kish, 1965; Yates, 1981). This permits quite small samples to be representative of the population from which the samples are drawn. There are a substantial number of sampling methods available. However, the principal ones used in transport surveys are: (1) simple random sampling; (2) stratified random sampling with uniform sampling fraction (proportionate sampling); (3) stratified random sampling with variable sampling fraction (optimal or disproportionate sampling); (4) cluster sampling; (5) systematic sampling; (6) choice-based sampling); (7) multistage sampling; and (8) overlapping samples. In the area of sampling, there appear to be few avenues of future development. Completely new ways of sampling seem unlikely to be developed. However, the potential of relying more on panels and less on large cross-sectional surveys is a possible direction of development. Major work in sampling seems likely to focus on reductions in non-response, and improvement in adherence to sample designs. Sampling is also likely to have to concentrate more and more on locating and including "rare" populations of interest for modelling and policy issues, which may result in transport planners using more sophisticated methods of sampling than has heretofore been the case.