Over the period July 1977 to June 1978, the Bureau of Transport Economics (BTE) carried out a general survey of non-urban travel in Australia. This survey was known as the national travel survey (NTS). The nts was a postal survey eliciting voluntary responses from households throughout Australia. Such surveys have three primary sources of error- sampling bias, response errors and non-response bias. This paper examines all of these sources of error in the context of the nts. Some of the more important and interesting results derived from the nts are presented in this paper. The estimates given in the tables presented are estimates for the whole population, after adjustment for non-response bias. Some discussion and interpretation of the tables in this paper is included. During the course of the nts it became apparent that, despite careful planning, there were several areas in the design of the nts which resulted in some problems in processing and analysing the final results. These deficiencies in the nts design are outlined in the paper to assist the design of any future survey along the same lines. Although it is by no means clear that all of these deficiencies could have been overcome, perhaps the main outcome of this review of the survey design was desirability of adequate pilot- testing of the complete operational approach to a survey, including not only the questionnaire format but the coding and administrative process as well.
Samenvatting