The Project Bureau for Integrated Transport Studies (Pb IVVS) commissioned the Department of Empirical and Theoretical Sociology - as from January 1992 this department became the Department of Sociology - of the University of Utrecht to conduct an evaluation of the Dutch National Mobility Panel (LVO). The LVO panel survey started in 1984 and was concluded in 1989 after ten surveys. Each survey consisted of approximately 1700 households. The main purpose of the evaluation of the LVO panel survey was to look into the methodological and statistical complexities which accompanied this panel survey and how these problems have hindered the attainment of the research goals of the LVO panel survey. Therefore, in light of the above mentioned purpose, the panel history was described as well as some methodological problems which stemmed from the planning and data collecting phase. These problems are: (1) the complexity of the sample, or, in other words, the generalization problem; (2) a massive and selective initial non-response (of 62%); (3) selective drop-out (attrition), from one survey to the next; (4) within and between measurement biases; and (5) the sample size, which was sometimes too limited. Next, a description is given of how the attainment of the research goals was hindered by these methodological problems and their resulting biases. A consideration is then given to the correction procedures which were employed to correct the biases, and how this complicated the different types of analyses and affected the validity of the substantive research findings. The following substantive conclusions which were generated by the LVO and which are known and used in policy-making circles are likely not to be affected seriously by the biases: (i) Travel behaviour tends to change when changes in the life style occur; (ii) Changes in mobility are thus related to changes in personal and household conditions. These changes, however, are asymmetrical; (iii) It is important to distinguish between short and long term effects on mobility; (iv) To some degree, travel behaviour is stable over time; and (v) there is no fixed group of transit users. Some other conclusions or insights are employed, not as concepts or ideas, but in a more instrumental way. These conclusions are: a) The LVO has generated a methodological-technical expertise that has been employed for designing and executing other panel studies; and b) cross elasticities were estimated on the LVO-data, and accompanied in the simulation model MIDAS, and were compared with similar estimates as used in simulation programs such as the "National Model System". Furthermore, it is indicated how the methodological problems can be avoided, or reduced, in the field during the data collection phase and be handled during the analysis phase. Finally, a few directions are indicated in which future developments of longitudinal policy research can be expected to be optimal.
Samenvatting