Shedding more light on the Dark Figure



This is yet another Venn diagram. I have been a little addicted  recently. This is taken from my 15 minute presentation entitled "Shedding more Light on the Dark Figure" which I gave on 7th December 2010 to the Crime Survey Uuser Group meeting. They have published it on the web along with the other presentations. They can be found here. Mine is the 2pm time slot.

Perceptions of Crime

I have been studying my Venn Diagram in the post one before the last decided that I can improve on it. This is not perfect but it is better. I actually like the diagram without the labels. Perhaps I should enter it for the Turner Prize entitled "Perceptions of Crime". OK.................. here it is with labels.

Linking Crime, Incident, BCS and Police Recorded Crime data sets

Posts to do with the the Metropolitan Police Service CAD and CRIS databases have been quite popular so I thought I would share a flow diagram that I have just created of how the data arrive in them. The other two data sets I am interested in - the British Crime Survey and Police Recorded Crime are also shown.

Overlapping Crime Domains

This is my latest Venn diagram trying to explain Crime: how it is measured/recorded and how it is perceived. I think it is reasonably self-explanatory. The interesting bits are the overlaps of the various domains. I am approaching this diagram from three different angles;

• Shedding more light on the "dark figure" a talk I am giving to the Crime Survey Users Group on Tuesday 7th December.
• Explaining the utility of police incident data in augmenting the other crime measures (BCS and police recorded crime) particularly regarding violent crime for a paper that I am writing, and
• My ongoing interest in describing the fear of crime.

The early history of crime maps with modern visualisation

I would like to draw your attention to a high quality paper (I have just discovered) that should be of interest to everyone interested in crime mapping. Its got it all, it has the history, the maps, the visualisation and a statistician who can write good English without feeling the necessity to resort to Ancient Greek lettering to make his point except where it is truly necessary.

This is an example of one of many maps within the paper. You will have to read it to find out what it is all about. The free .pdf can found here.

Temperature and violence



This blog entry is dedicated to my colleagues in CASA, especially Andy Hudson-Smith, Ollie O'Brien and Fabian Nuehaus.

Yesterday I collected data from CASA's weather station for the 2009 calendar year. I have collated it for analysis and these are my first results.

The reason why I have written a blog entry for three weeks is that I have been writing a paper entitled `A Geographical and Temporal Analysis of Violent Crime in London in 2009' which hope to submit to a journal for publication. This has switched me on to the characteristics of violent crime that are different to other crimes. One of things I noticed was the affect of events such as New Year's Day had on violent incidents but also the weather.

I thought I would look into the weather aspect a bit further and I am able to thanks Andy and Ollie. The graph above does not look to promising. It shows the maximum temperature for each day in 2009 on the y-axis and the number of incidents that were allocated a code 1 class 'violence against person` and graded as an 'I` or emergency call on the x-axis. The result shows a very weak positive correlation. Nothing to get too excited about there then.

This where I thank Fabian for helping me to regard everything in police data as monitoring a machine (or a body) that has rhythms and cycles. Police incident data has rhythms and cycles that you can set your watch and calendar by.

You can see that number of calls is affected by the day of the week. Working on the hypothesis that the rhythms and cycles of London that are reflection of people's work/leisure routines have a greater influence than weather I negated much of this variable by just using Saturday data. I tried various combinations of weather data including average daily temperature and various different CAD codes in combinations or divisions. The best correlation I found was with CAD code 1 emergency calls and maximum temperature.

An R squared of  0.4417 means that temperature is not the only influence on violent incidents, holidays, celebrations, special events all play their part. These in some cases may be a contributory mechanism by which temperature is an influence, (eg hot May Bank Holiday weekends) and in others may be a negative influence (eg colder New Year's day( not on a Saturday in 2009) or Halloween). In any case the R squared value represents a correlation of  0.6646 which is generally regarded as significant in social sciences.

First Global Crime Map?

 



To go to the interactive map click here

I stumbled across this website. I like it for all sorts of reasons. The main one is because I think, in its way, this may be the first interactive global crime map. OK prove me wrong.

Venn Diagram with the example of Burglary in Dwelling

Venn Diagram with the example of Burglary Dwelling in London in 2009



 In the last posting I showed a Venn diagram of the relationship betwee CAD and CRIS. Today I have amended it slightly and populated it with numbers relating to Burglaries in Dwellings. You will notice that from the figures almost all the burglaries are repoted via a CAD incident. This is what I would expect as burglaries in people's home are a priority.

Drugs - Harm or Danger, or both?

I would just like to briefly comment on the current headline story "Alcohol 'more harmful than heroin' says Prof David Nutt" This appears to be an update on a previous Lancet Article in 2007 "Alcohol 'nearly as harmful as heroin'"

I know Prof. Nutt due to working on a Foresight Project Brain Science, Addiction and Drugs back in 2005 and the co-author Dr King who advised on a some drug trafficking research I was part of.
 
The aspect I would like comment on is the assessment of harms that is the basis of the research. From my understanding the same methodology could be used to come up with the headline "Sugar more harmful than Guns" or "Potatoes more harmful than Arsenic" or "Toasters more harmful than Nuclear Submarines".

In my opinion an assessment of harm should not be isolated from an assessment of risk. Prior to the Misuse of Drugs Act 1971 there were various Dangerous Drugs Acts.
 
We should be assessing what is dangerous as well as what is harmful.

The Crime Record Information System CRIS discussed

Today I am discussing the Metropolitan Police Service Crime System, CRIS. I have just done a number of Google searches allow me to refer you to MPS or other sites that give background information about CRIS but there are none. So I will have to try my best to supply the relevant information.

My major first hand experience of CRIS was running a Crime Desk as Sergeant at Chiswick Police Station and then Hounslow Police Station when the two Divisions amalgamated as part of Borough Policing at the beginning of 1998. I went from a rather poor user of CRIS before the job to an expert in under 6 months due to working on it 7-10 hours a day. CRIS  has changed a bit over the years but not significantly from what I am able to tell from my searches of the system.

Unlike CAD, which is a system that is simple, elegant and clever due to it being designed from the start to maximise the strengths of computerisation, CRIS is none of those things for one basic reason, it totally copied the paper based crime reporting system that proceeded  it. Let me briefly take you through the history.

When I joined the police and for over ten years the way I would record and investigate a crime was by completing a paper crime sheet with a pen. These crime sheets had been developed over the previous 150 years of the Metropolitan Police history and were very good in ensuring that there were places to complete all the necessary details of the crime, show the investigation, suspects, people arrested, supervision, Home Office classification etc, etc. These were in a series of boxes and new sheets could be added as necessary. Different crime books were maintained. Major Crime, Beat Crime, Vehicle Crime, Burglary, Drugs and sometimes Robbery.

This is another very important point. The purpose of the paper record was to record crime and  its investigation; and be counted for official crime statistic purposes.

All that happened with computerisation was that the paper system with it series of boxes were turned into a computerised system with the same boxes which every police officer had to update like they were completing a written record.  No thought was given to how computerisation could actually make policing more efficient and effective. An opportunity was lost to free up police officers time, for instance, by having  specialist civilian operators taking reports from police officers at the scene of the crime . No thought was given to making the investigation details not free text but a series of searchable structured fields, or how the information could be used for intelligence and mapping purposes. The Metropolitan Police is still suffering from these short sighted decisions.

CRIS then is the definitive record of crime recorded by the police. Some of those crimes are recorded as a result of police incidents recorded on CAD. Others are reported over the telephone, via the Internet, at the front counter of police stations or  discovered by police, for instance after arresting someone, investigating another crime, etc.

Now this is where things get slightly complicated. There is a difference between that which is recorded on the CRIS - given a CRIS number - and that which is counted towards the official crime statistics. There are three main reasons for this;

1. The CRIS record is initually of an allegation, after investigation and supervision the correct classification is given to the record. This is where good "housekeeping" can "legitimately" reduce the number classified as burglary, robbery etc. - the priority crimes.
2. In the MPS all through my crime recording and investigation days were very straight forward with their crime recording - all allegations went in one system and were classified as crimes unless there was good evidence to show it was not a crime. I believe it was the practice in certain forces to keep the crime rate down and clear-up rate up for a dual system to operate where only crimes that were supported by good evidence went into the official crime book, the others were recorded separately and were only transferred across if more evidence was forthcoming. It seems that the spirit of this has been adopted by the MPS by creating something called the crime incident. This is to ensure that nothing that turns out no to be crime can possibly be counted in the official crime statistics. For instance, where the occurrence of a crime depends on the outcome of forensic tests; a CRIS record is created not as a crime but as crime incident so that a 7 day (it may have changed) classification rule does not apply.
3. CRIS has filled the gap in the recording of non-crime incidents, most notably Domestic Incidents, known as non-crime domestic incidents. This means that nearly all CAD domestic incidents have a CRIS number; this only sometimes means a crime has been recorded.

Simple Venn Diagram Showing the relationship between CRIS and CAD

Police Incident Data Set Discussed

Police Crime Incidents involving a firearm in London in 2009 at Borough Scale

Legend - Number of Incidents

Police Crime Incidents involving a firearm in London in 2009 at Ward Scale



Legend - Number of Incidents



In this post I am going to start to tackle a difficult question but one which is very important to crime mapping and my research. You may find some aspects of the answer surprising, I do.

The question is "How do my two data sets, the police incident data set - Computer Aided Dispatch CAD and my crime data set  - Crime Recording Information System  CRIS relate to each other?"

This is a very important question as internationally the United States "crime" mapping systems for the public appear  to be almost exclusively based on their incident or call for service databases whereas in the UK the crime database is exclusively used.

The first thing to say is that the two systems I am taking my data from - CRIS and CAD were introduced into the Metropolitan Police long before computerised crime mapping became common place. The two systems where introduced for different purposes. So a bit of history.

When I joined the Metropolitan Police in the late 1970s computers where just being introduced but the police incident recording system was still 90% non-computerised. There was a dual system of central and local. Emergency calls (999) were routed to Information Room at New Scotland Yard who would assign local units with mainset radios to calls. The local police station communications room would be notified through a Message Switching System (MSS) which printed out the call onto a printer. This call would be placed on the A5 pad that had been started at midnight that day and given the next number in order. When the call was dealt with the result would been written on the sheet. The MSS system would also be used to send and receive messages from other police stations to be dealt with - bail enquiries, Please Allow An Officer To Call On (PAAOTCO), court warnings etc. In addition the local police station would have the non-emergency telephone calls routed to it (which were quite often emergencies). These would be manually written onto a message pad with the nature of the call the location, details of the caller, time and unit(s) assigned, and when the call was dealt with write the result. This was often  in local code abbreviations the actual meaning of which usually had a correct cover story but often a (politically) incorrect everyday usage. The Comms officer would use the local police station radio system to communicate with police units and be responsible for monitoring incoming calls. Back then as now there was a problem with capacity so the telephone operators would divert everything that was not needed to be dealt with by the uniform police response team on duty to other police offices.

This is the first very important point.  The police incident system does not and never has recorded all calls from the public.

In the early 1980s the CAD system was introduced,  first in New Scotland Yard Information Room and gradually over the next few years to the Police Stations control rooms around London to replace the paper system. Numbers replaced the words relating to the nature of the incident and the result. A graded response system was also introduced. This is how I knew the system when I last used it operationally in late 2004.

So for the next bit of history, bringing it up to the present day, I rely on second hand information from what I have learnt through conversation, MPS documentation and these two website sources. The first is from the MPS website here and the Wikipedia site here, which is accurate (in the main) but needs updating as the information is now at least two years old.

It appears that capacity has again been the problem which has meant that CAD system being supplemented by the Call Handling System and the introduction of three Information Rooms around the MPS to replace the Police Station Control Rooms and the New Scotland Yard Information Room.

Call numbers are confusing. The MPS claims there are about 6,000 to 10,000 emergency calls a day and about 15,000 non emergency calls, a total of 21,000 to 25,000. The confusing bit is that within the emergency calls there are non-emergency incidents and visa versa. The result is that in 2009 there were about average of about 10,000 CAD incidents a day with a low of about 7,000 and a high of about 15,000. Of these about a third were dealt with as true emergencies - attend  as quickly as possible, a third not as emergencies at all - attend when police can or sort it out by other means, and a third as semi-emergencies, attend within the hour, if possible. This means that about another 10,000 calls are dealt with by the Call Handling System (CHS).

So in my analysis I have to be aware that if I confine myself to CAD data, which I have, and not delve into the into the CHS data, which I have not, then I am dealing with a subset of the total demand on police. I am reasonably confident, and I am working on this basis, that I am analysing a data set that has been selected by rules that are consistently applied throughout London over time. This I am assuming has produced a complete data set of incidents it is necessary to be referred to officers on duty to deal with. The filtering by call grading is a secondary part of this process.  


In the next post I will discuss CRIS and explain the maps at the beginning of this post. I think I has written enough for now.



The deficiencies of Home Office Crime Type 1 exposed again

Home Office Crime Type 1 -Violence Against the Person in London in 2009 shown at ward level

Legend - number of crimes



In this recent post I discussed Home Office Crime Type 1 -violence against the person and the fact that only about 40% of crimes involving firearms in London in 2009 were included in that category. The above maps shown the geographic distribution of such crimes at ward level. I obtained the data from the Metropolitan Police website here

The correlation between locations at ward level of my data set of crimes involving firearms and HO type 1 is 0.53. This is a low figure given that the average member of the public would expect that in most cases that crime involving firearms should be included in Home Office Crime Type 1.

"Do the recorded victims and offenders of police recorded crimes involving firearms live in the same places?"

The question I am addressing in this post is "Do the recorded victims and offenders of police recorded crimes involving firearms live in the same places?" Then I go on tho answer similar questions relating to deprivation and crime venue.

First a few details about my data set. I have been using the police recorded crime data set in my recent posts about firearm related crime.  "police recorded" means that it is only a subset of the total occurrence of crimes involving firearms. I can make no guess, comment or observation on whether this subset is typical or representative of the total. Thus "police recorded crime" part of the question. Police are good at recording victims and their addresses. There are only 8% of victims without a grid reference address. The details of a suspect or a person who has been arrested for an offence, an accused is far less complete due to police only detecting a percentage of crimes. Thus the question having "recorded" twice in it. I am using offender as shorthand for suspect or accused.

Next thing is; I am not looking at individual crimes, I am looking at all the crimes together. So if offenders and victims do live near each other it does not mean that they have been involved in the same crime, though it is possible. What I have done is correlate the location counts of victim and offender addresses at the different  scales I have been mapping. Borough, Ward and Grid Square. The results are as follows;

• Borough  -  0.95 correlation meaning that the boroughs with high number of victims have high number of offenders - an almost perfect correlation
• Ward - 0.68 correlation meaning that wards with high number of victims have high number of offenders - a significant correlation
• Grid - 0.30  correlation meaning that grid squares with high number of victims have high number of offenders - but this is not a strong correlation meaning there are many exceptions to this trend.

What can be inferred from this? It can said to a high degree of confidence that victims and offenders live in the same general areas as it each but not necessarily in exactly the same places. In London millionaires can live close to those claiming benefits so there is diversity in housing even at a grid square scale.

The table above gives the correlation results when victim and offenders address locations are compared with the Index of Multiple Deprivations (IMD) scores. I have discussed the IMD in detail on this blog so I will not explain further here. Just to give a word of caution, the spatial scales I am using are not those that the IMD was calculated at so there is a degree of inaccuracy that could be present at Ward and Borough level, though not much as they are adding LSOA scores together and taking the mean; the grid square IMD score is very much an approximation as I have discussed previously. This probably explained the Grid Square correlations show a higher correlation for victims where the other two show a higher correlation for offenders

Now this table showing correlations between the victim's and offender's addresses (in bulk) and the venue of the crimes (in bulk). It show that on a local level - grid square and ward the location of crimes are nearer victim's homes but at a slightly more general level - borough the trend is reverse. All these correlation are high though.

Locations of Firearm Crime Perpetrators' homes

Recorded Crime in London in 2009 involving firearms showing recorded accused/suspect address at borough scale

 


Legend - Number of Suspect/accused addresses

  

Recorded Crime in London in 2009 involving firearms showing recorded accused/suspect address at ward scale



Legend - Number of Suspect/accused addresses



Recorded Crime in London in 2009 involving firearms showing recorded accused/suspect address at grid square scale



Legend - Number of Suspect/accused addresses 

Recorded Crime in London in 2009 involving firearms showing recorded accused/suspect address at  zoomed grid square scale



Firearm Crime London - more witness and victim maps

Recorded Crime in London in 2009 involving firearms showing recorded victim and witness address at borough scale

Legend - number of witnesses and victim addresses

Recorded Crime in London in 2009 involving firearms showing recorded victim and witness address at ward scale

Legend - number of witnesses and victim addresses

Following on from the last posts where I mentioned the influence of firearms crime goes beyond the recorded victims to at least the witnesses and probably further. See my discussions about fear of crime and signal crimes/events. The maps above showing both witness and victim addresses in London give slightly different maps.

Firearm Victims in London Maps

Recorded Crime in London in 2009 involving firearms showing recorded victim address at borough scale



Legend - number of person victims

Recorded Crime in London in 2009 involving firearms showing recorded victim address at ward scale

Legend - number of person victims



Recorded Crime in London in 2009 involving firearms showing recorded victim address at grid square scale

Legend - number of person victims

I think these are self-explanatory. Just two things to say. Some crimes have businesses as victims, these are not shown, just the people victims. In my experience if the victim has beeen acting as an employee the work address is often given otherwise the home address is recorded, sometimes both. The second piont is I have only mapped victims of London crimes who live in London.

Firearms,victims, witnesses and splitting hairs

For a crime to have occurred there has have been an offender or offenders. Offenders are usually human beings though a corporate bodies can be offenders in rare cases.  Because people occupy space and time  (and so must corporate bodies because of laws regarding jurisdiction) it can be said that a crime always has a location and a time, that is, where the perpetrators are when the crime are committed. As I have discussed in a recent post there can be more than one location and time if an offence is a continuing one or where there is a victim or victims who are at different locations or where the crime, though "victimless" is one which the perpetrator can affect remotely.

Now why am I splitting hairs like this? Well the crime commission process is extremely important in the analysis of the occurrence of crime. The more precisely the factors that are involved in the occurrence of crime are understood and defined the more precise the resulting analysis will be.

For a crime to occur therefore there has to be three factors offender(s), location(s) and time(s).  This is a slightly different angle on the Problem Analysis Triangle from Problem Oriented Policing (POP) shown below

 This has been taken from the Centre for Problem Oriented Policing website that can be found here

Above is my graffiti emendations. Firstly offenders commit offences or crimes not problems, secondly as I have shown throughout this blog time is extremely important and is always present (forgive the pun) with place.  Now the third factor which is not always present (eg driving offences) are; a victim (eg theft, assault, etc), a target (eg goods to steal or damage) or a commodity (eg illegal drugs, firearms etc).

So yesterday I showed location. Today in the graph above I show the gender and age characteristics of the victims and witnesses of crimes involving firearms. This brings me on to another concept. Who is affected by crime? The victim obviously, but probably those that witnessed it as well if it was a violent or frightening event as many crimes involving firearms are. That is the reason why I have shown both graphs. The reason for the jagged peaks is because these are often estimated ages and these at often estimated at 5 year periods.

Zooming in on Firearm crime locations in London

Recorded Crime in London in 2009 involving firearms showing recorded Ward location



Legend - Number of crimes

Following on from the previous post I am showing the same data as the second at council ward and grid level. What I forgot to mention is my classification of quantities uses equal intervals with 0 excluded for all the maps; I have also tried to use the same clours so that they can be compared.



Recorded Crime in London in 2009 involving firearms showing recorded grid square location

Legend - Number of crimes



Recorded Crime in London in 2009 involving firearms showing recorded grid square location - Zoom detail

The interesting aspect of these maps to me is the borough of Haringey (top centre of zoom grid map). The borough map published yesterday showed that Haringey is not in the top four boroughs ranked by number of crimes but it has a ward that is ranked number 2 and a grid squre that is ranked number 2 which is not located within the second ranked ward! 

Firearm Crime, the problem of counting and location discussed



Recorded crimes involving firearms in London in 2009 showing Investigating Borough



Legend - Number of Crimes

Following on from the previous post I am answering the question "Where do these crimes occur?" I am in danger of immediately confusing you when answering this question. There is an assumption with crime mapping that a crime happens in a specific place at a specific time. This of course is not always the case. The commission of a crime can happen over a long period of time in many different places.  Good examples are possession of illegal drugs or driving a stolen car. The location and time of a crime are usually recorded where they are discovered. What about internet related crimes, where are they committed, in cyberspace or the location of the victim, if there is one, or the location of the perpertrator, if that can be discovered? So one way of answering the question is which police "own" the crime and are responsible for investigating it. This is usually to do with physical location of the crime but when this is not easy to establish or where there are many locations then the Home Ofiice rules apply - which defaults back to where it has been discovered or first reported to police if no other rule applies.

So the first map above answers the question by saying which borough police are responsible for investigating and recording the crime. This has been worked out based on a data set of 5587 crime incidents. You notice that in the previous post the total of offence is 5648 - a difference of 61. This is to confuse you further and to further illustrate that numbers are rarely simple with crime counting, this is because on rare occassions the same incident is recorded as more than one offence. This is unusual at the initial reporting stage (which my data set is obtained from) but happens quite often when a crime has been solved and further offences can be added, thus helping the clear-up figures, but that is a different story.



Recorded crimes involving firearms in London in 2009 showing recorded location 



Legend - number of crimes

Whichever way the crimes are counted London is fortunate in having a low level of crime which involve firearms. The most dangerous boroughs appear to be Lambeth, Southwark, Lewisham and Hackney.