All time references are in CEST
Pitfalls and challenges of combining survey and georeferenced data |
|
| Session Organisers | Dr Philip Adebahr-Maskow (Martin Luther University Halle-Wittenberg) Professor Oliver Arránz Becker (Martin Luther University Halle-Wittenberg) Mr Torvid Kreisler (Martin Luther University Halle-Wittenberg) |
| Time | Tuesday 18 July, 09:00 - 10:30 |
| Room | |
The growing use of geo-referenced data, such as geospatial, satellite, GPS tracking and social media data, offers promising opportunities to enrich survey data and provide regionally sensitive insights. However, as their importance grows, both old and new challenges are emerging. This session aims to bring together researchers to discuss the methodological, ethical and practical complexities of incorporating geo-referenced data into survey research.
Key topics include:
1) Data quality: Crowdsourced data (e.g. OpenStreetMap) often lack standardisation, leading to quality concerns. There are also technical issues in interpreting data, such as satellite imagery, and dealing with incomplete data due to volunteerism. Because of the latter, we know little or nothing about what data is missing. There is also a need to rethink response bias and even common concepts of (unit/item) non-response, as well as common solutions such as weighting and imputation.
2) Composition of spatial aggregates: Regional analysis can also be challenging as spatial units vary in size, quality and even change over time.
3) In addition, the use of approximate locations raises ethical concerns about privacy and data protection. Methods to meet data protection requirements need to be discussed in order to open up data sources.
4) In addition to the challenges mentioned above, the integration and harmonisation of geo-referenced and survey data poses its own challenges, which can also be address in this session.
We welcome contributions from all disciplines and encourage papers that present novel tools, techniques and ideas, methodological solutions, critical reflections on current practices, and studies that highlight the challenges and innovations in combining survey and georeferenced data.
Keywords: survey data, geospatial data, data integration, regional analytics, spatial analysis
Mr Torvid Kreisler (Martin Luther University Halle-Wittenberg)
Mr Daniel Meyer (Federal Institute for Research on Building, Urban Affairs and Spatial Development) - Presenting Author
Mrs Elisabeth Stürmer (Federal Institute for Research on Building, Urban Affairs and Spatial Development)
The relationship between climate extremes, environmental attitudes, and political action is an increasingly studied, albeit contested, topic in the social and environmental sciences. Its complexity stems from variations in the operationalization of those concepts, the data sources used, and the interplay of objective and subjective perspectives, which together have contributed to inconclusive findings in the literature. With our study, we advance the debate by focusing on discrepancies between objective exposure to extreme weather events, subjective concerns about climate change, and pro-environmental action and how those discrepancies might be explained by media attention, sociodemographic characteristics, and regional context. To explore those interconnections, we combine geocoded survey data from the German Socio-Economic Panel, weather data from the German Meteorological Service, protest data from Fridays for Future, climate-related media coverage data from German broadcasters ARD and ZDF, and regional indicators from official statistics. With the data linked at the individual level, we calculate deviations between objective and subjective indicators and aggregate them to the district level. From a methodological perspective, our presentation addresses two additional aspects: first, the challenges and opportunities of linking objective indicators (e.g., heatwaves, droughts, and floods) with subjective measures (e.g., concerns about climate change) and behavioral outcomes (e.g., voting for Green parties and climate protests); and second, the strengths and limitations of various approaches for visualizing those discrepancies, ranging from conventional plots and tables to innovative formats including opinion maps and gap maps. Our work contributes to a broader monitoring system of subjective and objective indicators being developed at the German Federal Institute for Research on Building, Urban Affairs and Spatial Development. The system aims to monitor the socioecological transformation of Germany’s lignite mining regions and provide a hub of information and data for researchers, policymakers, and the broader public.
Mr Torvid Kreisler (Martin Luther University Halle-Wittenberg) - Presenting Author
Dr Philip Adebahr-Maskow (Martin Luther University Halle-Wittenberg)
Professor Oliver Arránz Becker (Martin Luther University Halle-Wittenberg)
Background: As one of the most comprehensive sources of spatial infrastructure
data, OpenStreetMaps (OSM) is an important data infrastructure project for
enhancing survey data. OSM is based on voluntary contributions from individuals and
organisations, which has enabled the enormous growth of the database, but also
presents researchers with the challenge of having to assess the completeness of
certain features in each region in order to determine the reliability of OSM data for a
particular project.
Aim: As part of a research project on cycling infrastructure and mode choice in 4
German cities, we determine the degree of saturation of the OSM cycle network in
these cities in order to assess the reliability of the database.
Methods: We use the method of Barrington-Leigh and Millard-Ball (2019) to estimate
the completeness of specific OSM features in each region from the history of OSM
data, without relying on external data sources. The method assumes that the growth
of records for certain features is initially characterised by increasing interest and later
slows down due to increasing saturation, and can be modelled as a sigmoid shape
that approximates the actual extent of the feature. We extracted the historical OSM
data for our 4 cities and prepared it for the estimation of parametric non-linear models
to determine the degree of saturation of bike lane mapping in the 4 cities.
Results: Our results allow us not only to make statements about the reliability of OSM
bike lane data, but also to detect missing data and identify spatial determinants. As a
use case for assessing the reliability of OSM data, our presentation will provide an
important impetus for other research projects.