New forms of data for public safety
Dr Réka Solymosi
University of Manchester
04/02/2021
Crowdsourcing
Jeff Howe The Rise of Crowdsourcing
Harnessing information and skills from large crowds into one collaborative project.
Crowdsourcing
Jeff Howe The Rise of Crowdsourcing
Harnessing information and skills from large crowds into one collaborative project.
Constant contribution
Constant contribution
Constant contribution
Insight
Insight
Insight
How is this relevant for public safety?
3 key approaches
3 key approaches
1: found data
3 key approaches
1: found data
2: collected data
3 key approaches
1: found data
2: collected data
3: meta data
1: Found data
Web scraping to gather reports
Identify risky situations
Web scraping to gather reports
Map experiences with disorder
Dynamic with time
2: Collected data
Experience Sampling Method
In-context survey
Spatial data on experiences
Additional dimensions (eg time)
3: Meta data
Study characteristics of the data
Limitations
3 key issues
3 key issues
1: bias
3 key issues
1: bias
2: assumptions
3 key issues
1: bias
2: assumptions
3: ethics
Bias
Bias
- Self-selection
Bias
Self-selection
Unequal contribution (participation inequality)
Participation inequality
FixMyStreet data
Assumptions
Assumptions
- time of report = time of experience?
Assumptions
time of report = time of experience?
inferred variables (e.g. demographics)
Assumptions
time of report = time of experience?
inferred variables (e.g. demographics)
shared understanding
Ethics
Ethics
- data ownership
Ethics
data ownership
data linkage
Ethics
data ownership
data linkage
privacy/ surveillance
Ethics
data ownership
data linkage
privacy/ surveillance
implications of findings (service slanting)
Summary
Summary
- many benefits of new forms of data
Summary
many benefits of new forms of data
- insight into experiences and behaviours
Summary
many benefits of new forms of data
insight into experiences and behaviours
spatial and temporal data
Summary
many benefits of new forms of data
insight into experiences and behaviours
spatial and temporal data
large volumes/ low cost
Summary
many benefits of new forms of data
insight into experiences and behaviours
spatial and temporal data
large volumes/ low cost
key issues
Summary
many benefits of new forms of data
insight into experiences and behaviours
spatial and temporal data
large volumes/ low cost
key issues
- representativeness / generalisability
Summary
many benefits of new forms of data
insight into experiences and behaviours
spatial and temporal data
large volumes/ low cost
key issues
representativeness / generalisability
assumptions
Summary
many benefits of new forms of data
insight into experiences and behaviours
spatial and temporal data
large volumes/ low cost
key issues
representativeness / generalisability
assumptions
ethical issues
Thank you
These slides: https://rekadata.net/talks/newformsofdataforsafety.html
Questions? Contact by email: (reka.solymosi@manchester.ac.uk) or Twitter (@r_solymosi)
Further reading:
- Solymosi, R., Buil-Gil, D., Vozmediano, L., & Guedes, I. S. (2020). Towards a place-based measure of fear of crime: A systematic review of app-based and crowdsourcing approaches. Environment and Behavior.
- Buil-Gil, D, R Solymosi, and A Moretti (2020). Non-parametric bootstrap and small area estimation to mitigate bias in crowdsourced data. Simulation study and application to perceived safety. Big Data Meets Survey Science. Hill, C. A., Biemer, P. P., Buskirk, T., Japec, L., Kirchner, A. and Lyberg, L. E. (eds.). Wiley.
- Langton, S., & Solymosi, R. (2020). Open Street Map for crime and place
- Buil-Gil, D., & Solymosi, R. (2020) Using Crowdsourced Data to Study Crime and Place
- Solymosi, R., Petcu, O., & Wilkinson, J. (2020). Exploring public engagement with missing person appeals on Twitter. Policing and Society, 1-24.
- Slides created via the R package xaringan. The chakra comes from remark.js, knitr, and R Markdown.