Key findings

• Using a method called diffusion maps we find a set of variables that describe the differences in soft properties between areas in the city of Bristol, and the relative importance of each of those properties.

• The most important soft property in Bristol is the presence of a large student population in an area.

• The second most important soft property relates to socio-economic deprivation. This closely, but not exactly, correlates to the locations of social housing.

• In some areas of the city differences in socio-economic deprivation is identified at a building level.

• Using these properties we are able to quickly and accurately identify student areas and social deprivation in other cities.

• Further variables are shown to be important, however, establishing the soft properties they correspond to requires further analysis.

Policy recommendations

This research provides diffusion maps of census data as a new tool for policymakers. Similarities and differences between areas can now be easily identified aiding design, implementation and assessment of interventions. Examples include:

• Establishing which soft properties are most important in a city can guide which interventions are likely to have the biggest impact. For example, in Bristol the biggest driver of difference between areas is the presence of students, so interventions that alter the distribution of students will have a large impact on the social fabric of the city.

• Some interventions may have already been deployed in different areas with varying success. Soft properties can identify which areas are most similar to those where the intervention was most successful, and so where the greatest benefit may be felt.

• By enabling comparison between areas in different cities diffusion maps can be used to guide the importing of interventions used elsewhere. For example, a scheme that is successful in one particular area in London could be implemented in the most similar area to that in Bristol, increasing its chances of success and the scale of its impact.

• Previously qualitative goals for interventions can now be quantified by the use of soft properties to measure impact.

• Targets for intervention can be better identified. This tool measures soft properties directly, reducing the need for the use of hard properties as proxies. The first step is identifying which soft properties diffusion maps have identified as important, beyond student populations and deprivation.

• The census is only conducted every 10 years. Policymakers should seek the capability to analyse more frequently available data with diffusion maps, and link this to the results from the census.