The importance of the time and space of the offence highlighted by Canter in his Investigative Psychology approach, is a theme picked up in the speciality of geographical profiling. So while in both Criminal Investigative Analysis and Investigative Psychology the main practical emphasis of the offender profile is to provide information about the kind of person the offender may be, in geographical profiling the practical emphasis of the profile is to tell you where the offender is likely to be based. So like the more traditional criminal profiles it is a tool that investigators can use to help prioritize their search areas. Geographical profiling is typically used to give estimates of the most likely geographic location of an offender's home, but can also be used to identify work sites or just areas that the offender is highly familiar with. So broadly, geographical profiling works on the principle that most crimes are not random. There is a reason or logic that they take place in the location that they do. The various approaches to geographical profiling are based on a variety of observations of human behaviour and the resulting principles and theories. The first of these is the least effort principle. That is, as its name suggests, given two or more alternatives people will take the option that requires the least effort, or is easiest. So at a basic level, in terms of where a crime is committed, if there are two identically attractive options then the offender will choose the one closest to him or her. Related to this is distance decay. This is the idea that crimes will decrease in frequency the farther away the offender travels from their home. So you may expect that an offender will show a preference for crime scenes closer to their home. However, they are unlikely to be too close because that would increase the chance of recognition. So close to home, but not in the 'buffer zone' where the person is comparatively well known. But both of these principles are a bit of an oversimplification. If I am an offender and I'm thinking that a dark shadowy corner would be ideal in which to commit a crime and I can identify two locations, then it is highly probable that other aspects will impact my choice when I'm considering relative desirability. What if one doesn't have an easy access route, but is closer to my home? What if the other doesn't have a ready stream of people who may pass by, on whom I can commit my crime? What if the other has too many people so I might get caught? All of these aspects come to bear on my choice of crime site and these may be unknown when looking at where I committed my crime and trying to make some sensible inference about where my base is. This is where the importance of mental schemata or mental maps comes into play. Now these are internal representations of an area and reflect how the person typically interacts with their environment. So aspects highlighted as significant in the area will vary depending on how you interact with that particular area. For example, a car driver may have a very different mental map of a city centre than someone who routinely uses public transport. Similarly, as Canter and Youngs in 2002 noted 'experience shapes understanding of places which in turn influences where a person does what.' So in the context of an offender, where they commit their crime is not random. For example, offenders choose areas where familiarity of some sort, through being close to home, related to work or leisure activities, leads them to feeling 'safe' to commit their crime there. But as Canter and Youngs in 2008 pointed out, this may not simply be a passive learning of opportunity through engaging in 'routine activity'. Rather, this feeling of safety might come about through seeing possibilities and, if you like, 'testing' the area or deliberately interacting with the area to build up the necessary mental maps or schemata. That is, if you like, seeing what is possible in an area they are becoming increasingly familiar with and increasingly feel safe in. Now in terms of practical applications the centrality of the base for an offender was used by Canter & Larkin in 1993 to propose two types of offending which was directly linked to the degree that the offender is tied to their home in their offending. So a marauder is someone who uses his or her home location as a base. So the hypothesis here is, if you can identify the two offences that are furthest away from each other and that you are sure have been committed by the same person, you can draw a cycle around them and then the perpetrator's home will be somewhere in that circle. The alternative type of offender is the commuter. Someone who leaves their home and commutes to a different region in which to offend. So this is useful if you can determine whether you're dealing with a marauder or commuter, but that of course raises the question of how could you know this at the outset? While Canter and Larkin in a 1993 analysis said that 87% of offenders were marauders; subsequent studies have found some variability by crime and by country. So for example, Meaney (2004) found that only 35% of burglars in a state in Australia were marauders, whereas 93% of sex offenders in the same state were marauders. Drawing on research by Warren and colleagues and Lundrigan and Canter, Canter and Youngs noted that commuters tend to commit their crimes geographically closer together while marauders will, in contrast, spread out across the area in which they commit their crime. And so looking at the distance over which crimes are being committed may give some practical insight into whether your offender is a commuter or marauder. However, just like the least effort and distance decay principles, there are some complications in this differentiation between marauders and commuters. First, it may not apply if there are other contextual features of the crime at play. For example, the need to have a particular type of victim who can only be found in discrete locations. A more general complication is that you need to be fairly certain that all the crimes you're looking at have been committed by the same person. And this can be tricky, particularly if the perpetrator's mode of offending is evolving as they commit their crimes. In addition, you also need to be sure that you have captured in your analysis the majority of crimes committed by that particular offender. If you have not, then there is the potential that you will have some spatial bias in your sampling and this will distort the conclusions that you can draw from your analysis. Now despite the potential problems with some of the logic in geographical profiling, computer programs to assist with the process have been developed. The various programs differ somewhat in terms of calculations, cost, interface and output, and some, to a degree, can be customised to the area in which they are applied. Based on the geographical distribution of crimes, these programs show which geographical areas to prioritize in investigations. Further, these programs allow investigators to explore the impact of any given case on the highlighted prioritization map. So for example, as an investigating officer if you're not sure whether case X has actually been committed by the same individual as cases A, B and C, you can explore the impact on the geographical map by omitting case X from your analysis. However, a general limitation that applies to these programs is that they can only be as good as the input. So, for example, Rossmo in 2000 suggested that you needed five distinct but accurate locations and you assume the offender hasn't moved residence in the time and you're fairly confident that it is the same offender. And this may be problematic if you are dealing with an offender whose offence pattern is escalating as their fantasies that may be associated with the offending evolve. So in practice, geographical profiling, like offender profiling more generally, may not be seen as adding anything to investigations if investigators feel that their existing methods are at least as good.