There’s a kind of voodoo associated with geographic profiling. An analyst pours all sorts of seemingly random data into their computer and produces a pretty map with color-coded mountains where no mountains exist. They tell you you’ll find your bad guy on the tops of the highest mountains. It’s not all that far off from looking for water with a divining rod, but it works.
There was geographic profiling before there were computers to do it. A long time ago, I was testifying at the trial of a man I arrested for drunk driving. He had been stopped on a street that was home to a dozen small working-class bars, a target-rich environment for me when I was working deuces. I could spend my uncommitted time driving up and down the two miles or so where the bars were, and it wouldn’t be long before I found myself behind a driver who was weaving, speeding, driving 15 miles per hour or exhibiting some other behavior that led to the two of us having a chat.
[sc name=”Article Mobile Ad” ]
Defense counsel was trying to get me to say I stopped every car I saw on that street, reasonable suspicion or otherwise. I hadn’t done that, and refused to say I had. He was getting quite frustrated, having asked essentially the same question nine different ways. He finally got to a question I could answer: “Why did you spend so much time driving up and down Wells Avenue?”
“When you go duck hunting,” I said, “you have to be where the ducks are. They won’t line up on your front porch and wait to get shot.”
Geographic profiling tells you where the ducks are.
The Electronic Pin Map
Another remnant of policing’s Bronze Age are pin maps. Pin maps were posted in the squad room and maintained by someone who didn’t like to go outside. Colored map pins would be placed on the map where burglaries, robberies, auto thefts and other reported crimes had occurred. Diligent patrol officers were supposed to check the pin maps for their beat and give greater attention to the areas where the pins were accumulating. More commonly, officers would just wait until the room was empty and then move some pins to someone else’s beat, thus making themselves appear to be doing a sterling job of crime fighting.
Geographic information systems (GIS) have given us a kind of electronic pin map, with data linked to each point identified on the map. We see these maps used all the time, for all sorts of applications that have little to do with crime. Bring up Google Maps in your web browser and enter “pizza” followed by your zip code. You’ll be presented with a map with a “pin” stuck in at the location of every pizza joint in town. Zoom in and click on a pin, and the address, phone number and maybe even a review of the pizza will pop up.
Crime maps do the same thing, with each pin linked to a crime report or other occurrence of interest to the police. The beauty of the electronic map is that it can be layered with an infinite number of plots of other data, such as the location of schools, sex offender residences, bus routes, methadone clinics or anything else that may cause an analyst to see a pattern where none was previously apparent.
Local government GIS systems, which are now commonplace, maintain precise plots of everything from the location of fire hydrants to properties where business licenses have been issued. Global positioning system (GPS) hardware has made the placement of these landmarks highly accurate without the need to use land surveyors to deliver the same data. Several devices are available that will record a digital image from hundreds of yards away, at the same time logging the latitude, longitude and elevation of the point in the crosshairs with precision of about four inches.
Crime mapping is just the start of geographic profiling. Criminal profiling is usually seen as a last resort for investigators when they have too few leads and too many potential suspects. The typical scenario involves the case officer compiling a summary of the crime or crimes for analysis by the Behavioral Science Unit at the FBI’s Quantico facility. The analysts there study the available data and possibly dance around a fire while sacrificing small animals, and come up with a description of the person they believe is responsible. They have an excellent track record with this, although a profile may be only 60 80 percent accurate, and until the bad guy is identified, no one knows which 60 80 percent applies. But the profile may allow investigators to de-prioritize suspects who are outside of the profile and concentrate on those who are the best fit.
Geographic profiling works similarly, using complex algorithms to predict where a suspect might live, work or frequent. The foundation of the data is usually the crime map of the locations where linked incidents have taken place. Most criminals have a style or preference for their work. A burglar might use the same method of entry and search, or prefer targets with specific attributes, such as proximity to freeways or money stored in a floor safe. Serial rapists and murderers may prefer a certain age, gender, race or occupation, and use a similar weapon or restraining method in their crimes. Analysis of these factors will produce the sample of incidents most likely committed by a single person.
Other factors entered into a geographic profile may include transportation routes and time-and-distance calculations. A suspect with access to a motor vehicle has a greater choice of routes than one who gets around on a bicycle or by public transportation. The economic and ethnic makeup of neighborhoods can create a crook-friendly environment or serve as a barrier or obstacle to be avoided. A rapist with a broken-down van will attract attention in an upscale neighborhood, and a white crook will avoid black neighborhoods for the same reason.
Vehicles believed to be used in the crimes provide an additional data set. Even if the best description is “a dark-colored, late model SUV,” that information alone, if correct, can eliminate thousands of other cars and trucks. State motor vehicle records will provide the addresses of registered owners of cars with those characteristics, and even if your state doesn’t record the color of cars, the auto manufacturers may be able to provide the paint color applied to vehicles bearing the relevant VINs when they left the factory. If there’s a distinguishing feature such as a sports team logo decal on the vehicle, patrol officers can be asked to roll by the registered addresses located on their beats and determine if any of the cars bear the sticker.
The Key Is in the Relationships
Each bit of relevant data refines the solution. The key is not so much in each data point as in comparing the relationships between data points. Given the amount of data to digest, a human analyst couldn’t possibly compare every item with every other, giving the appropriate weight to each factor (which may change as the analysis progresses).
This is where computing power comes into play. A geographic profile analysis can involve millions of calculations to produce a map illustrating the calculated probabilities of the offender living or frequenting a specific area or areas.
Troy Graves would probably have continued his career as a rapist and murderer without the scale of analysis now available to investigators. Graves raped five women and strangled another in the Rittenhouse Square area of Philadelphia from 1997 1999. In 2001, there was a series of seven rapes of women in their apartments in Fort Collins, Colo., and a DNA sample obtained from a victim showed a match to the Philadelphia series. Unfortunately, neither the Philadelphia nor the Fort Collins detectives had identified a suspect.
Detectives used searches of public and business records databases from companies like LexisNexis and ChoicePoint to identify people who had lived in the areas defined by zip codes where the crimes had occurred, then cross-referenced those data points for people who had lived in Philadelphia from 1997 1999 and then in Fort Collins in 2001. This produced a list of 700 names, a manageable number when you can eliminate people who don’t fit the suspect’s profile. Graves confessed to both series of crimes and is serving a life sentence in prison.
Peaks & Valleys
The output of a geographic profile is a three-dimensional map of the area most likely to include the suspect’s residence or other frequented location, with mountain-like shapes overlaid. The “mountains” are of varying colors, corresponding to their “elevation.” The highest peaks are the areas where the suspect will most likely be found. There may be multiple peaks, depending on how accurate and complete the data sets were. Multiple peaks may be false positives or show more than one base of operations, such as a residence and a workplace. In every case, the analysis will assign a probability to every point on the map, indicating where the data show the suspect will most likely be.
The Royal Canadian Mounted Police used geographic profiling in the investigation of a series of carport arsons in Burnaby, British Columbia. Psychological and geographic profiles allowed the investigator to identify a single suspect, who was subjected to surveillance and then interviewed, resulting in a confession. His residence was across the street from the location at the peak of the geographic profile, which covered 0.6 percent two hundredths of a square mile of the hunting area.
Creation of a geographic profile requires a lot of data gathering, refining and sorting before the case can be sent to an appropriate analyst who will generate the profile. It’s not likely that will become a routine practice because of its labor-intensive nature. In fact, it compares well with psychological profiling for the same reasons. But when you have a major case and you’re out of investigative leads, the effectiveness of these two tools can be the difference between capturing the bad guy and having him continue to prey on the citizens you serve.
Application & Training
Several companies provide geographic profiling software and analysis, some of which overlap. There is training available to gain expertise with this technology, and several professional organizations and conferences for practitioners (and people who want to be practitioners) in the field:
- Environmental Criminology Research, Inc. (ECRI) is a Canadian-based organization that developed the Rigel professional geographic profiling system, used throughout the world. Contact ECRI for information on law enforcement training in geographic profiling. www.geographicprofiling.com
- The Center for Geospatial Intelligence and Investigation (GII) at Texas State University conducts research on applications of geographic profiling for the police, intelligence and military communities.www.txstate.edu/gii
- CrimeStat was developed through a grant awarded by the National Institute of Justice, Office of Justice Programs at the U.S. Department of Justice. The software is available as a free download, and technical assistance can be had through the National Archive of Criminal Justice Data at the Inter-University Consortium for Politics and Social Research.www.icpsr.umich.edu/crimestat/mapping.html
- The Centre for Investigative Psychology is a UK-based organization that administers publications, conferences, training courses and communities for both psychological and geographic profiling.www.i-psy.com
- The Environmental Systems Research Institute, Inc. (ESRI) is probably the largest and best-known provider of GIS solutions. It offers many products, publications, conferences and other products many of them free for the public safety community.www.esri.com
1. Rossmo, D. K. (2006). Geographic profiling in cold case investigations. In R. Walton (Ed.), Cold case homicides: Practical investigative techniques (pp. 537-560). Boca Raton, FL: CRC Press.
The author acknowledges the assistance and kindness of D. Kim Rossmo, Ph.D., at Texas State University for his help in preparing this article.
Tim Dees was a law enforcement officer in Nevada for the Reno Police Department, Reno Municipal Court, and the Pyramid Lake Paiute Tribal Police Department. Following his police career, he was a full-time criminal justice instructor at colleges in Wisconsin, West Virginia, Georgia, and Oregon. He worked as a regional training coordinator for the Oregon Department of Public Safety Standards & Training, and was the first editor-in-chief at Officer.com before joining Elsevier Public Safety and LawOfficer.com in January 2008. His work has been published in nearly every major law enforcement magazine, and he is the author of a book published by Prentice-Hall. He holds a bachelor’s degree in biological science from San José State University, a master of science degree in criminal justice from The University of Alabama, and the Certified Protection Professional credential from ASIS International.