Les systèmes de transport intelligents, École Polytechnique de Montréal

Articles du groupe MADITUC

Measuring the internal quality of a CATI travel household survey

Référence:

CHAPLEAU, Robert (2003). Measuring the internal quality of a CATI travel household survey, in Stopher, Peter, Jones, Peter (2003). Transport Survey Quality and Innovation, Kruger, South Africa, Pergamon, pages 69-87.

Type:
Publication (divers)

Organisme:
Transport Survey Quality and Innovation

Retour

Nouvelle recherche

Résumé

Literature is relatively abundant to address several issues about the quality of transportation survey data. Effects of nonresponse and sampling biases are well-documented issues and concern much more the domain of expansion and interpretation of final data.

If someone accepts to lessen the comprehensiveness of the quality concept of a travel survey and reduces his perspective to the only concerns of the process itself, it becomes possible to look at the survey’s own behavior.

In other terms, a travel survey constitutes an experiment on its own, or a relation between transportation planners and the observed population. As illustrated in the following figure, the survey itself (interviewers, interview software, interviewee’ answers) acts as an Intelligent Questioner interfacing the chaotic proper knowledge of a population and the planners’ world (organized and structured way of observing transportation problems, and specifically transport networks).

In the context of a telephone Origin-Destination household travel survey, information technology techniques, planning methods and instruments as well as reference databases constitute structured and coordinated knowledge to orient and support the gathering of coherent and quality data.

Formally, the own process of a travel survey may appear as a dynamic sequence of almost fixed interviews (in terms of questions asked) conducted by interviewers subjected to a learning process (by themselves and by feedback from the geocoding and validation processes) within a context continuously monitored. Every step of each interview (data entry and editing) is timed and maintained in a structured database.

The information obtained by the monitoring of the interviews permits the derivation of several performance measures and quality indicators.

The proposed method in this paper consists of the following measures:

oEstimating household interview duration according to residence location;

oModeling the learning process of the interviewers and its consequence on nonresponse, call duration and mobility rates;

oMeasuring the mobility rates differences between the respondent and other members of the household (interviewed through the respondent);

oMeasuring the effect of interview moment (period of day), number of calls required and appointed interviews on mobility rates and household characteristics (size and car ownership).

An application of this method will be demonstrated with data derived from the latest Origin-Destination household travel survey held in 1998 in the Greater Montreal Area. Almost 400 000 phone calls were conducted from August 24 to December 17 (96 days of interviews), 78 000 of which resulting in complete household interviews.

gbisaillon@polymtl.ca 2024-11-05 17:24:16