All time references are in CEST
Current Developments in Mobility Survey Methods 1
|Session Organisers|| Dr Johannes Eggs (infas)
Mrs Dana Gruschwitz (infas)
Dr Stefan Hubrich (TU Dresden)
|Time||Tuesday 18 July, 16:00 - 17:00|
Travel surveys collect data on the mobility of populations. Large-scale national household travel surveys are used to estimate key mobility figures on national or sub-national levels, like out of home rates, trips and transport mode rates, and to predict the demand of the population regarding the use of transportation modes. But high respondent burden and sinking response rates are pressing on the established travel survey designs. On the other hand new technologies are available to ease survey participation.
The focus of this session is explicitly wide and aims to gather an overview about the cur-rent developments, challenges and ideas for mobility surveys. Papers matching one of the following aspects are invited to be part of this session (but are not restricted to):
– New technologies for data collection and mixes mode approaches
– Integrating GPS data in trip and survey data
– Gamification in travel surveys
– Survey mode effects in travel survey
– Improving trip reporting in different modes
– Improving household completion rates
– Effects of proxy interviews
– Methods to collect mobility data for specific sociodemographic groups (e.g. children)
– New sample frames and weighting procedures
– Incorporation of big data
Keywords: Travel survey, mobility survey, passive data collection
Dr Rico Wittwer (Technische Universität Dresden) - Presenting Author
Mobility surveys, i.e. household travel surveys, are an important instrument for transport planning and modeling. There are different approaches for capturing daily travel patterns of individuals across countries. In many cases, travel diaries are used to report and record trips on individual level. Trips are strongly linked to activities. The need to perform an activity at another spatial location results to travel. Different travel modes can be used to on a certain trip. Accurate trip capturing is paramount for the estimation of travel demand and mode choice by calculating trip-based or distance-based modal split values. Underreporting trips is a concern, due to its systematical occurrence and an expected result bias. However, the ground truth of peoples travel patterns is unknown. GPS-based survey methods help to better understand travel patterns and trip coverage but still have weaknesses by itself. Underreporting is different for transport modes. Particularly short walking trips and travel by car, with short activities in between, might be affected of systematically underreporting. Trips by bicycle and public transport seems to be less affected. The quality of capturing these trips also may differ by survey mode. The two cross-sectional household travel surveys “Mobility in Cities – SrV 2018” (CAWI, CATI) and “Mobility in Germany – MiD 2017” (CATI, CAWI, PAPI), as two different and independent sources for collecting data of the German population, allow to examine those phenomenons. As both surveys have the same trip definition, the differences of trip recording by mode can be accessed and compared. Furthermore, survey specific elements to ensure a high quality of trip capturing are discussed. Conclusions can be drawn, and recommendations will be derived towards improving trip reporting in different survey modes.
Dr Stefan Hubrich (TU Dresden) - Presenting Author
The household travel survey (HTS) “Mobility in Cities – SrV” is conducted in 2023 for the 12th time since 1972. More than 270,000 individuals will be surveyed regarding their travel behavior in more than 500 German municipalities. SrV 2023 is a year-long, cross-sectional HTS, based on municipality-specific stratified random samples that have been selected from the municipalities’ population register. All selected individuals are asked to provide information about their household and household members as well as all trips for one complete day. Data is collected via telephone interview or online questionnaire. This contribution addresses the challenging circumstances for travel behavior surveys in Germany, provides an overview of the organizational and methodological concept of the survey, followed by first insights into experiences of the still ongoing fieldwork (i.e. response rates, response modes).
Mr Dennis Oliver Kubitza (Federal Institute for Vocational Education and Training (bibb)) - Presenting Author
In this project, we investigate the influence of the local transport infrastructure on regional markets for vocational education and training (VET) in Germany. Our research is based on spatial regression models by means of regional data on the supply and demand of VET positions (bibb), administrative data (BBSR/INKAR) and to commuting times across borders of administrative regions (Google Distances API).
Many school graduates that just finished school cannot realise to find a position for their aspired training. Simultaneously, companies struggle to fill their VET positions. A skill-mismatch and unrealistic aspirations might partially explain this phenomenon, but also spatial mismatches are relevant causes. It is problematic to test the spatial-mismatch hypothesis against the other two hypotheses because it is difficult to quantify and operationalise spatial influence; are they based on proximity, distances, car driving times, or public transport times? From a theoretical point of view commuting times with public transport should be the main driver for adolescents.
Following this, we propose a spatial regression of the number of students that applied for VET positions but did not find a position yet on regional labour market characteristics (unemployment, availability of VET positions, availability of universities). We argue that neighbouring labour market regions should influence each other and spatial spill-over effects for demand and supply should be visible if the regions are well interconnected with public transport.
Modelling the strength of the regional connections by different proximity measures, like commuting times (queried from the Google Distances API), as a spatial regression with lagged independent variables enables us to correctly capture these effects. By varying the underlying weighting matrixes for spatial influence and nesting models, we can test and compare spatial interactions with each other and determine if commuting times by public transport are a reasonable driver of spatial mismatch.
Ms Danielle McCool (Utrecht University/Statistics Netherlands) - Presenting Author
Mr Barry Schouten (Statistics Netherlands)
Mr Peter Lugtig (Utrecht University)
Human mobility can be measured using the sensors on a participant’s personal mobile device, alleviating many concerns of traditional surveys with new smart surveys. One primary issue with collecting this data (semi-)passively is the high percentage of missingness, much of which is an unavoidable consequence of device restrictions. The temporal nature of human mobility limits the avenues that researchers may take in aggregating individual data to generate statistics without proper consideration of the missing data. This paper compares multiple imputation, both with and without the use of Dynamic Time Warping for candidate selection, to interpolation, mean imputation and complete case analysis. The methods are applied to data generated by the 2018 Statistics Netherlands mobility study, and the impact of each method is investigated on various travel statistics. The choice of mechanism has a meaningful impact on the generation of mobility statistics, including distance traveled, time traveled, radius of gyration and average trip length. Mean imputation without respect to the time of day overestimates travel behavior when compared to household survey results. Complete case analysis provides theoretically plausible results, but the reduction in the total number of available cases severely limits breakouts along person or journey characteristics. Multiple imputation leads to increased trip length relative to linear interpolation and has the additional benefit of providing confidence intervals.