Towards a Place-based Measure of Fear of Crime: A Systematic Review of App-based and Crowdsourcing Approaches

Study characteristics

Before going into strengths and limitations it’s worth noting that we found some very interesting patters in our descriptive exploration of these studies. To me what sticks out the most is the incredibly wide range of ways to measure ‘fear of crime’.

This should be familiar to anyone interested in ‘fear of crime’ literature. Already in the 1980s scholars lamented that “the phrase ‘fear of crime’ has acquired so many divergent meanings that its current utility is negligible.”¹

But even on this new frontied of “apps” and “crowdsourcing” we fail to arrive on a consistent way of measurement and instead we have things like:

• Evaluate the momentary sense of security on a scale from 1 to 5
• Green or red tag to indicate locations where they feel comfortable or uncomfortable
• Choose ‘which place looks safer?’ from two images
• Choose between two scared/safe emoticons
• Answer: ‘In this moment, how worried are you about becoming a victim of crime?’

…and many many more!

Strengths

We arrived on the following themes for strengths identified by studies presenting these approaches:

• Capture the spatial-temporal specific nature of attitudes and emotions towards crime.
• Record data on individual variables and specific types of fear/disorder.
• Record data on architectural features.
• Reduced cost of data collection.
• Oriented to evidence-based policy making/urban planning.

The paper details these and highlights examples, but one really exciting one is the ability to produce point-level maps of people’s experiences, and then associate these with the environmental features of these environments in day and night times.

For example, my paper in London, and the study of David Buil-Gil in Elche² both resulted in such maps:

Then David went further to explore through Google Street View the environmental context in which these reports were made:

Limitations

We also wanted to highlight what were the weaknesses of the approach identified by the studies we selected, and what sort of future work these might insire to address them. Limitations emerged in the following themes:

• Sample issues Participation inequality. Specifically:
  • No screening questions.
  • Participation decrease
  • Small sample sizes and low response rates.
• Under-representation of certain areas and times.
• Difficult to interpret results.
• Limitations to generalise results.
• Repeatedly asking about fear might increase/cause fear.
• Lack of temporal variability in some web-based measures.

Again we detail each one of these in the paper in deatail, but I want to highlight one. Sampling is of course a huge issue here, which I know there is work being done to address, but one issue that is not to do with sampling of people, but with sampling of places, presents an interesting problem. One response to this bias in sampling of places (rather than sampling of people) is a study design implemented by Laura Vozmediano and colleagues³ where instead of asking people to self report areas they are in, they tracked people’s movements across the entire study period, resulting in all of the routes. Then these routes were randomly sampled to be shown back to people to ask ‘when you walked this route earlier today, how safe did you feel?’. This is one way to create more representative samples of areas people have been to, rather than asking people to choose where they report.