Granular Time Series Crime Data

Status: 
Closed
Publisher: 
Outcome: 
Data already available
Data request description: 
For our project we need to know the type of individual crimes as far back into the past as possible, and when they happened. Monthly crime records are not granular enough. What is essential for what we are doing is: - The type of crime committed - The time of the crime (to the second ideally, but just as accurate as we can get) What would be really useful in addition is any of these: - The GPS Position ideally, or something to indicate the geographical location of the crime - The outcome - The police force(s) dealing with the crime - Any details about the crime (more detailed than just the type of crime committed, a description of what happend for instance)
Reason: 
A version of the data is published but I need it in a different version
There is crime data published, but it the time granularity is monthly, even though it has GPS positions. We need an accurate time for each crime, but some of the other data that is provided in the current version is not as important. I mention this in case there are privacy issues around the data.
Suggested use: 
Personal Use
Research
Benefits overview: 
The data will be studied for patterns. To see whether weather has an effect on types of crimes, what time of day different kinds of crimes tend to take place, whether different types of crimes happen with a random frequency or if there are clusters, how all these factors have changed over time and many other analysis’s. As a result we can learn about how civilian may be able to adjust their behaviour in certain situations to reduce the chance of being affected by crime, to reduce the amount of crime, and effect the likelihood a criminal may be caught. It could also enlighten us about the nature of crime, and could give councils, businesses, and government the information which aids the prevention of crime, and more effective use of resources to combat crime.

Comments

Granular Time Series Crime Data

There are likely to be quite a few problems with this request. The first few I can think of are:

  1. The data you're asking for don't all exist at the national level. For example, each force describes the method of each offence on their own crime-recording system, but this information is not necessarily collected nationally (as distinct from statistical information, which is collected nationally). Instead it is necessary to merge data files obtained from different police forces: see http://adrn.ac.uk/catalogue/cataloguepage?sn=888053 for more information on this.
  2. In many cases, releasing the GPS co-ordinates and exact times for offences would be likely to disclose the identity of the victim of the crime (and sometimes also the offender), which is personal data protected under the Data Protection Act. If you were a woman whose partner abused her, would you really want details of when, where and what happened released on a government website? See http://data.police.uk/about/#anonymisation for details of the process used to anonymise police.uk so it can be released without the risk of identifying victims.
  3. You want to know the times of offences "to the second really" but the police almost never know when a crime occurred to the minute, let alone the second. Think of a residential burglary, for example: the victim only knows that the crime happened at some time between their leaving the house in the morning and returning in the evening, which might be a span of 10 hours. Even for a street robbery, it is likely that the victim will only be able to say 'it happened at about four thirty' rather than being able to give a specific time. See http://www.crimesciencejournal.com/content/2/1/1 for more details on these problems and how you can deal with them.
  4. You seem to be attempting to replicate large amounts of existing criminological research. For example, you want to "see whether weather has an effect on types of crimes" but a search on Google Scholar brings back 166,000 results of a search for 'weather' and 'crime'. You also want to know "whether different types of crimes happen with a random frequency or if there are clusters", but it is already known that crime always clusters in space and time.

It may be more useful for you to formulate a more-specific research question and then request the data that you specifically need to answer that question.

 Flag as offensive 

Many thanks for your detailed comments

Hello Ralph and Matt,

Thank you Matt. Your advice for Ralph is exceptionally helpful. Alison (the Home Office open data lead) has consulted with her interally statistitions. They have confirmed that advice is spot on.

Kind regards, 
James Arthur Cattell
https://twitter.com/jaCattell

+447500607815
Head of Engagement
Cabinet Office Transparency Team
https://www.gov.uk/government/organisations/cabinet-offic

 Flag as offensive