A discrete choice experiment to validate the use of areal wombling for detecting social boundaries

Data, code and materials from a discrete experiment to test the validity of an Bayesian areal wombling algorithm for predicting social boundaries. The experiment was conducted as a part of project ‘Life at the Frontier: Researching the Impact of Social Frontiers on the Social Mobility and Integration of Migrants’ (2020-2023; NordForsk/ESRC, project no 95193), and experiment data was collected in Rotherham (UK).
About the experiment
Each border on a map is assigned a boundary value based on how dissimilar the adjacent neighbourhoods are (higher = more dissimilar = more likely to be a social boundary).
The experiment was carried out as follows:
- We created three maps of the same area with different boundaries using the Bayesian areal wombling approach.
- Map A contained the boundaries with the highest boundary values, whilst map C had the lowest boundary values. Map B contained boundaries that were in between.
- During an interview, participants were then shown pairs of maps and asked which map in each pair best corresponds to local community boundaries.
- The sequence and order of the maps shown were randomised.
- Assuming that residents and experts can recognise (but not necessarily recall) social boundaries, we conjecture that participants would choose the map containing borders with higher boundary values.
Hypothesis: We hypothesise that participants will agree with the predictions of the areal wombling algorithm and choose boundaries with higher boundary values.
Null hypothesis: Participants are not more or less likely to choose boundaries with higher boundary values.
Aside from testing a hypothesis, another motivation behind the study is to explore the feasibility of the method for future replications and follow-up research.
More information
This study was approved by the University of Sheffield ethics committee (application number 042378).
Please read the README file for a more detailed description of the content of this repository.