Tuesday, March 19, 2013

Visualizing the Directional Bias in Property Crime Incidents for Five Canadian Municipalities

Crime mapping and related research can be divided between analysis of places, distances, and directions.  Authors Richard Frank, Martin A. Andresen, and Patricia L. Brantingham argue that directionality is the most underrepresented of the three crime mapping analysis elements.  Places and distances traveled are sometimes not comprehensive enough for crime patterns or for particular directions crime is more likely to target.  The authors argue that people are directionally biased towards certain locations.  Strong directionality is found when a person travels from the home to other locations, such as entertainment, work, and shopping, all within a 45 degree angle from the home.  Another way to put this is that a person with strong directionality is likely to travel to locations that are closer to one another and then go home, rather than circling around or returning home before completion of these activities.  Directionality is also influenced by frequency of trips to particular locations.

The authors analyzed data from five municipalities in British Columbia, Canada: Coquitlam, Maple Ridge, Surrey (all within the Metro Vancouver area), Prince George (just outside of the Metro Vancouver area), and Nanaimo (on Vancouver Island).  They used two different kinds of analysis: arrows to show the direction from the home to the criminal incident (exclusive of distance) and colored dots to show the direction of the crime (green equates to south, red equates to north, west equates to blue, and east equates to yellow).  The authors argue that this type of crime mapping visualization technique is an improvement on previously used techniques which appeared convoluted. 

The authors found that this technique was largely successful for mapping the directional bias of criminal activity for each of the municipalities.  They found that the directional arrows or colored dots accurately depicted where criminal opportunity was likely to occur (shopping centers and downtown areas).  The authors suggest analyzing strength of directionality for future research to show which areas are more likely to have higher amounts of criminal activity stemming from them.

This study does a good job emphasizing the importance of directionality for crime mapping.  It is easily applicable to law enforcement intelligence as a tool to locate origin points in neighborhoods with particularly high quantities of crime emanating from them.  The same idea can be used for other topics such as competitive analysis for product information distribution (ex: examining the magnitude of where people from certain areas like to shop). 

I was somewhat confused how geometry was critical to this analysis as directionality was only shown in one direction instead of the interconnectivity of home, work, entertainment, and shopping.  Additionally, the authors argue the arrows simplify the mapping process.  However, I believe some areas are unreadable due to the quantity of arrows in a particular location.  Lastly, the rainbow dots symbolizing crime direction seem like a good idea at first, but are not easily readable in reality.  I found myself examining the color key on many occasions trying to remember which color symbolized which direction.  This analysis could be improved by changing how the arrows are represented (maybe one big arrow for a higher crime area), remove similar arrows for simplification, or simplifying the colors to four or eight colors for readability. 

Frank, R., Andresen, M.A., & Brantingham, P.L. (2013). Visualizing the directional bias in property crime incidents for five Canadian municipalities. The Canadian Geographer, 57(1), 31-42. Retrieved from: http://onlinelibrary.wiley.com/doi/10.1111/j.1541-0064.2012.00450.x/pdf


  1. Directionality was no something that came up in my research on crime mapping. This supports the argument that this is one of the least used functions of crime mapping.I completely agree with your assessment. With directionality, crime maps could point towards origin points, making the technique that much more useful. Rather than focus on where crime has occurred, it could point to where it originates, which is very useful for law enforcement.

  2. Cori, I also think the directionality should be represented more simplistically. Figure 2 reminds me of a scatter plot where you can draw a line that best fits the data points in the map. In doing so, you can determine the general direction of an incident. For example, it may indicate that larceny is more likely to occur in the North East than South West. I think the author’s argument regarding people with strong directionality traveling at 45o angle is unrealistic. Maybe I’m a bit confused, but not all locations will fall with in 45o angle to one’s home. A criminal fleeing the crime scene may not necessarily follow the shortest route (at a 45o angle) to his or her home. I’m also not sure how this can be applied to crime mapping.

    As Professor Wozneak mentioned in class criminals commit crimes away from their homes, in locations their familiar with. Therefore, I’m not sure if directionality from the criminal’s home is very important when identifying crime patterns. Using criminals’ home address to determine directionality is effective only if the criminal has been apprehended. Hence, authorities know the criminal’s home address. This method is ineffective in determining crime patterns for unsolved crimes. If the criminal is not apprehended, then law enforcement authorities do not have criminal’s home address in order to determine directionality.

  3. I think the author's method in describing the three elements of crime, places, distances and directions, was very interesting as it seems to not be the method I've seen most authors take in my research when discussing crime mapping. He provided a new insight into crime mapping with the concept of directionality. I'm not sure how effective the arrows were however, because the more clustered they became, the less I was able to see the trend. This is the same issue that sometimes occurs using dots with crime mapping. Although I do find this to be an important addition to the technique, restructuring it somehow to clear the image up, much in the same way kernel density smoothing does with dots, would improve it.