The Ministry of Transport first commissioned the National Travel Survey (NTS) in 1965/1966 to provide up-to-date and regular information about personal travel within Great Britain and to monitor trends in travel behaviour. The NTS gathers information about several different aspects of travel: purpose of travel, method of travel (public or private transport), origin and destination, amount of time spent travelling and distance. In 1988, the NTS became a continuous survey with an annual issued sample size of 5,040 addresses. This was increased to 15,048 addresses from January 2002. At present the NTS is not weighted when analysed. Therefore no correction is made for any biases resulting from unequal probability of selection and/or differential non-response. The National Statistics Quality Assurance review of the NTS recommended that weighting should be introduced for the NTS 2002 and so the National Centre for Social Research was commissioned by the Department for Transport to explore options for weighting the NTS. The project involves several stages of work during which a weighting strategy for the NTS will be developed. This report contains the details and conclusions of the second stage of that work: the recommended weighting strategy. The following points summarise the key aspects of weighting the National Travel Survey (NTS). • For analyses that do not involve the travel diary data, the analysis sample would consist of all participating households with completed individual interview for all household members (either in person or by proxy), regardless of the amount of travel diary information collected. We refer to this sample as the ‘interview sample’. • For analyses that involve the travel diary data, the analysis sample would consist of all fully responding households. Households that are fully responding have complete individual interviews and travel diaries for all householders and information on at least one vehicle is collected (if applicable). We refer to this sample as the ‘diary sample’ • Two sets of weights were generated – one set for analyses of the interview sample, the other for analyses of the diary sample. • The interview sample weights allow for the following stages: selection; household non-participation; removal of households with missing individual interviews. • The diary sample weights allow for the following stages: selection; household non-participation; removal of households with missing individual interviews; and removal of interview sample households that were not fully responding. • Calibration weighting adjusted the weights for the interview sample so that the age/sex and GOR distributions of the respondents matched population estimates. • Two-phase calibration weighting adjusted the weights for the diary sample so that the age/sex and GOR distributions of the respondents matched population estimates and the weighted estimates for key survey measures matched the weighted estimates in the interview sample. • Weights were also produced to adjust for drop-off during the week of the travel diary and for the short-fall in reporting of long distance journeys. (Author/publisher)
Samenvatting