The latest crime data was published on Wednesday by the Office for National Statistics, covering everything from homicides to sexual offences to vehicle theft.
Crime statistics cut across so many different policy areas, and can be very useful in highlighting the seriousness of an issue to new MPs.
So we took a detailed look at how this data could be converted to constituency level to form part of MP briefings.
The result was the heatmap and interactive constituency profiles you can visit here, but to find out how we did it, read on.
Sourcing localised crime data
We started by looking for the raw data, which covers the year to March 2024 in England and Wales, on the Office for National Statistics website.
For our purposes, rather than national data, we’re looking for data broken down into smaller geographical areas.
Helpfully, the ONS publish this data both by Police Force Area and by Community Safety Partnership (which generally map to local authorities).
Although on paper CSPs, as smaller areas, would provide a tighter fit to constituencies than PFAs, the data incorporates figures that are “unassigned” and would be lost with any attempt to convert to constituency.
So using Police Force Areas as a starting point seems like a much better bet.
Closer inspection of the Police Force Areas data sets shows that as well as recording the actual number of offences in a particular area, it also includes a rate of offences per 1,000 population.
This is particularly useful as it flattens out the differences between Police Force Area populations. Northumbria, for example, has a population of nearly 1.5 million whereas neighbouring Cleveland and Durham have populations of closer to 600K.
This data point gives us a more useful like-for-like comparison, and a useful starting point for creating constituency level estimates.
But first we need to map the new parliamentary constituencies to police force areas.
Calculating police force to constituency overlap
For this we need GIS software. We use QGIS but other options will do the same job.
We use the GIS software to calculate the overlaps between shapefiles for police force areas and parliamentary constituencies.
We want to find out the largest police force area in each constituency because this will give us our starting point for converting data.
The outcome is that of the 575 constituencies that make up England & Wales, for only 20 does the largest police force area cover less than 99% of the constituency. We will come back to these 20 later on.
Mapping theft statistics to constituencies
For the purposes of this exercise, we decided to focus on the “theft” statistics.
Our next step is therefore to download the crime data we need per 1,000 population by Police Force Area. This comprises one tab of our Excel spreadsheet.
In a new tab, we paste a column of parliamentary constituencies and a second column of the police force areas they are mapped to. We have also included a column detailing the total population of each constituency.
Using a lookup we are able to obtain a figure for each different type of theft offence (vehicle theft, theft from the person, bicycle theft etc) per 1,000 population.
We arrive at our ‘estimated’ number of offences in a constituency, by dividing the constituency population by 1,000 and then multiplying it by the rate of offence per 1,000 population.
For the constituency of Bath for example, which is part of the Avon and Somerset police force area, where the rate of vehicles offences per 1,000 population is 5.5, we are able to estimate that there were 563 vehicle offences.
Using the constituency population provides a small degree of variation inside the police force area, as we have only 487 estimated offences in the comparatively smaller Wells and Mendip Hills constituency, compared to 605 in the larger Bristol East constituency.
Now we have our data sheet ready, the final step is to visualise it and we can do this easily using Polimapper.
Check out the visualisation below of estimated “theft statistics” by parliamentary constituency.
Constituencies with more than one police force
And finally back to our 20 parliamentary constituencies where the main police force area is less clear cut.
Our original spreadsheet includes details of all overlapping police force areas. Let’s take the most extreme example of Morecambe and Lunesdale where Cumbria accounts for just over half of the constituency, with the remainder the domain of Lancashire police.
For this one, we will add a note saying that as the split is very even, we have included data for both police forces, and publish them side-by-side in a table underneath the main stats.
