22 December 2004

We will know where you live. The art of geographic profiling.


I was thumbing through the November/December issues of 34 Magazine (kind of like a better, more mature version of Wallpaper) and came upon an article on geographic profiling. For the layman, geographic profiling is a criminal investigative methodology that uses the locations of a connected series of crimes to determine the most probable area of offender residence. It is generally applied in cases of serial murder, rape, arson, and robbery, though it can be used in single crimes (car theft, burglary bombing etc) that involve multiple scenes or other significant geographic characteristics.

The basis of geographic profiling is the link between geographic crime site information and the known propensities of serial criminals in their selection of a target victim and location. The system produces a map of the most probable location of the criminal’s centre of activity, which in most cases is the offender’s residence. When linked with additional information relating to the crime incidents, and with additional data sources, such as motor vehicle databases and suspect databases, geographic profiling has been proven to have a profound impact on the effectiveness of a police investigation.

Geographic profiling can be used as the basis for several investigative strategies, including suspect and tip prioritization, address-based searches of police record systems, patrol saturation and surveillance, neighbourhood canvasses and searches, DNA screening prioritization, Department of Motor Vehicle searches, postal/zip code prioritization and information request mail-outs.

The thing is that while it is not really a new investigative science, it does have some serious marketing applications, particularly as it is all about behaviours. For example , if you were a book store owner who wanted to target book buyers in the local area, it might be possible, using geographic profiling to nail this information down and sell more books to people who actually wanted to buy them. The important thing is that unlike traditional direct marketing,which relies upon segmenting target audiences against standard residence data; geographic profiling relies on building a clearer understanding of behaviour linked to that place of residence. In the same way micromarketing concentrates on demographic and some psychographic information but is commonly formed around a more stable set of data from buying patterns which are based on the demographic, rather than the more predictable behaviours that radiate from the home.

In some investigative work say for a serial killer, geographic profiling works to establish what the author of Geographic Profiling Dr. Kim Rossmo calls the “distance to K” functions. Rossmo defines this by stating that most individuals will travel fairly limited distances to do certain things.

"There’s a large body of research called journey to crime research that shows that criminal’s will tend to commit their crimes fairly close to where they live," says Rossmo.

According to Rossmo older offenders will travel further than younger offenders, bank robbers will travel further than burglars, whites will travel further than blacks, but it’s a good rule that’s been studied in a number of different contexts over a number of different time periods and with a number of different types of criminals.

"Psychologists give this a fancy name, they call it the “least effort principle”, geographers, to be different, they give it their own name, they call it the “nearness principle”, but the point is that the same patterns of behaviour that McDonalds will study when they’re trying to determine where to place a new restaurant, or that a government may look at in terms of the optimal location for a new hospital or fire hall, also apply to criminals. There is one important difference though, and that’s what we call a “buffer zone”, the existence of an area that if you get too close to the offender’s home suggests the probability of crime interaction goes down. And so at some point where the desire for anonymity and the desire to operate in one’s comfort zone balance, that’s your area of peak probability."

What Rossmo means is that everyone has an activity space, areas we live in, work in, play in, and our movement patterns around the city and suburbs. And the identification of these sites and the probability of an action, an activity, a behaviour, a buying pattern, in the case of the customers for our bookseller, is going to be predictable and there is going to be a high probablity that we are going to be pretty right about that.

Rossmo actually provides names for those the activities and sites that criminals use. For example, in a homicide, there is what he calls an "encounter site between the offender and the victim", an "actual attack location, the point where he first makes his evil intentions known", the murder scene itself, which may be different from the attack location and then the body dump site. Rossmo says these sites may all be clustered together, or they may be spread out.

The bookseller, using geographic profiling, may be able to define the "encounter site" when a bookbuyer first sees a book, the "attack location" may be the place where decisions about the purchase are made known (this may be the home or a cafe, for example). In the same way, the scene of the crime might be the place of purchase, which may not even be the booksellers store, it could be department store or at home online at Amazon, for example. The point being, in the two steps between commitment to buy, the bookbuyer may have been at a number of locations, but ostensibly these are going to be close or clustered.

Rossomo notes that in terms of police use of geographic profiling for the location of crimes, they use fairly sophisticated geosymmetric measurements to produce what he calls a "jeopardy surface", which he defines as the probability surface that shows us where the offender most likely lives. Our bookseller may not have accesss to this kind of data, but other larger organisations and instutions, such a political parties and major retailers, may be able to perform similar analysis to determine the "the jeopardy surface" for their own subjects. And clearly, while geographic profiling is used to help identify a criminal, it can also help identify those consumers or electors, who are likely to be at the "attack" point in their behaviours.

Rossmo uses many other categories to label criminal behaviours. These include someone he calls a “raptor”, that is someone who attacked a victim more or less upon encounter. A second group is called a “stalker”, someone who at some fishing hole usually will find his victims, then follow them back to their homes and that’s where the crimes occur. The immediate purchase and the delayed purchase are common retail equivalents to the raptor and the stalker.

Rossmo says that raptors and stalkers are more complex to analyse, because their activities are not confined to a single area, in the same way, someone who derives satisfaction from finding the best price for an item will travel away from their residence to find prices (similar to hunter/gatherer).

The opportunity and access to data for use in geographic profiling is increasing as the growth of state based surveillance is. The planned reintroduction of an identity card in the UK, the introduction of biomorphic scanning into passports by the United States and increasing camera surveillance, will soon mean that supplying postal information, the use of marketing databases and electoral information will be a thing of the past. Geographic profiling will be able to be used in conjunction with other information gathering methods so you won't have to say where you live, someone will know, because they would have predicted it.

www.geographicprofiling.com