Longitudinal surveys in transport are a logical development in the broadening orientation of data sources. Although single cross-section surveys will continue to predominate, the recent history of forecasting capability of models estimated on a cross-section has suggested that longitudinal surveys are required to account for important influences such as the timing of change, the role of habit and expectations, and the degree of stability and/or growth of key exogenous forces. This paper outlines alternative longitudinal data strategies, discusses the advantages and disadvantages of each strategy, and presents selective methodological and practical issues. The paper draws on recent experience with a continuing four-wave panel of 1434 sydney households, interviewed annually. The often-stated concern for panel sample attrition is clarified. (Author/publisher) for the covering abstract of the conference see IRRD 286978.
Samenvatting