Btk

[Trapperologist]

As Trevor said on another thread, which I agree with, if the Ripper was an outsider with a bolt hole, which I also agree with, then the Rigel profile is hardly of any investigatory use.

Dr Watson for BTK is better if you only went by the first 7 crimes even though he’s driving and his car is his bolt hole, only because he was dropping off his wife at times etc. Otherwise you have no anchor but his parking space. But the geographic profile would mirror the Ripper’s and he ended up being a suburban.

Could it be the biggest suburb as a logical conclusion? I don’t know how you’d factor that into the program. Could the program be altered so the first 7 crimes alone would put a yellow dot outside the circle of murders up in Park City where he lived? I have some ideas - one based on proximity to the buffer zones or borders of the city, although I haven’t looked at the Wichita City borders to compare with the Whitechapel borders and murders which hug them at the top and parallel them.

[JeffHamm]

It should be noted that Rader stalked -or as he would say, made “projects” of- many more women than he ended up murdering. Some, rumoredly, up to 150 miles away from Wichita. Their identities and locations have been kept confidential, so without that data it’s difficult to say for certain how much his ‘day to day routine’ influenced who and where he killed. I think with Rader, as with most every serial killer, we’re working with only a portion of the puzzle pieces and aren’t able to truly have a complete picture.

JM

True, but Rader was also deep into his fantasies. I'm sure he reveled in thinking about murdering women all over the place, but he appears to have only acted upon those that fell within a practical range. I believe the woman he was planning on targeting next was a co-worker, so again, someone in his every day life.

- Jeff

[JeffHamm]

As Trevor said on another thread, which I agree with, if the Ripper was an outsider with a bolt hole, which I also agree with, then the Rigel profile is hardly of any investigatory use.

Dr Watson for BTK is better if you only went by the first 7 crimes even though he’s driving and his car is his bolt hole, only because he was dropping off his wife at times etc. Otherwise you have no anchor but his parking space. But the geographic profile would mirror the Ripper’s and he ended up being a suburban.

Could it be the biggest suburb as a logical conclusion? I don’t know how you’d factor that into the program. Could the program be altered so the first 7 crimes alone would put a yellow dot outside the circle of murders up in Park City where he lived? I have some ideas - one based on proximity to the buffer zones or borders of the city, although I haven’t looked at the Wichita City borders to compare with the Whitechapel borders and murders which hug them at the top and parallel them.

Well, yes, dropping his wife off to work would be his anchor point though - it's a location that he is associated with in his daily life that results in becoming familiar with the area (and increasing his chances of spotting a potential victim, etc). If he didn't drop his wife off, and so didn't have that anchor point, he probably would not have spotted Mrs. Otero, and that crime might never have occurred. His later employment at ADT, though provides an additional anchor point in that area - I just don't know which one he worked at and have never been able to track that down (seriously, if anybody knows, please tell me and provide me with a source ).

Basically, we know he actually was associated and familiar with something where the profile indicated a hotspot. Whether or not he would have committed the same series of crimes in the same locations without that is an unanswerable question because he did have it. Just as we could assume he would, we could assume he would not, it's hypothetical and unprovable either way. What we do know, though, is that he did commit those offenses, and the analyses do highlight a location that Rader was known to be familiar with (which is what the analysis is attempting to do - highlight locations that the offender is associated with; it's not only about finding their residence. Their home location, however, often does get highly ranked, but for those like Rader who live outside the crime range (and so are commuters with respect to their home location), the profiles often still pick up on other anchor points, like this one for Rader. Remember, it's not only about locating the home, it's just the home is commonly located in these and is often used as the way to judge success, partly because it makes for good TV and news stories.

When you include all of Rader's crimes, though, all of the ADT locations in the area are highly scored, his church is in the red zones (<20, I think it's 15), his wife's place of work is up there (around zone 4), the University is in the red, his house in zone 1, and his old place of work (Cesna) is also in the red. Basically, all of the locations in Rader's life are showing up as areas with a better than chance probability of being a location associated with the offender.

Also, Rader's crimes in Park City didn't start until after he started working in Park City as a compliance officer, again, his daily routines shifted to be more centred in Park City and where he lived.

- Jeff

[jmenges]

His later employment at ADT, though provides an additional anchor point in that area - I just don't know which one he worked at and have never been able to track that down (seriously, if anybody knows, please tell me and provide me with a source ).

Hi Jeff,

Thanks for the reply.
I believe during this time period there was only one ADT location in Wichita and it was downtown. Possibly the same physical location on Waterman Street where it is now but the current building is newer. If it was not on Waterman, then it was very nearby.
His job at ADT allowed him to travel all over the city, state and country. 90 miles from Wichita, in Salina, Kansas, while working for ADT and driving their van, he attempted to pull of what he called 'Project Iron Mountain' : breaking and entering into a young woman's house and murdering her. He got inside and was laying in wait but her boyfriend entered the home first, taking him by surprise. He pulled a gun and demanded the keys to their car and made his escape back to his ADT van which he had parked a few blocks away.
He was often out of town while working with ADT, even traveling to New York and New Jersey for training seminars. During these out of town trips he claimed to have always been on the hunt, but he never confessed to any murders. He confessed to doing a lot of breaking and enterings and attempted hits. Do we take him at his word? I'm not so sure.

JM

[JeffHamm]

Hi Jeff,

Thanks for the reply.
I believe during this time period there was only one ADT location in Wichita and it was downtown. Possibly the same physical location on Waterman Street where it is now but the current building is newer. If it was not on Waterman, then it was very nearby.
His job at ADT allowed him to travel all over the city, state and country. 90 miles from Wichita, in Salina, Kansas, while working for ADT and driving their van, he attempted to pull of what he called 'Project Iron Mountain' : breaking and entering into a young woman's house and murdering her. He got inside and was laying in wait but her boyfriend entered the home first, taking him by surprise. He pulled a gun and demanded the keys to their car and made his escape back to his ADT van which he had parked a few blocks away.
He was often out of town while working with ADT, even traveling to New York and New Jersey for training seminars. During these out of town trips he claimed to have always been on the hunt, but he never confessed to any murders. He confessed to doing a lot of breaking and enterings and attempted hits. Do we take him at his word? I'm not so sure.

JM

Oh fantastic! Thansk jmenges! That would be one of the two possible locations close together (the 2 westerly ones in zone 9 and 14), creating a 2nd anchor point in that area along with his wife's workplace. The influence of the latter would reduce over time while the influence of this one would increase, and over time the crime locations do shift from east to west. However, you're right, he did patrol the city in his van, though I was unaware (or have forgotten) that he travelled more widely as well. Whether or not he did engage in a lot of B&E's is hard to say based upon his word alone, but I wouldn't be surprised if he did do that, at least on some occasions if not as often as he claims.

I've just put together a map that includes Selina (I just plopped the location in the middle of the town, as the map is so zoomed out that the analysis will be fairly robust with regards to exact placement because the zones themselves are still large areas due to the large crime zone we're dealing with).

Here's what Dr. Watson produced. I admit, it's always unnerving to test a new configuration but it's interesting how this increases the emphasis on Park City, and the main area is also zone 1. While this looks good in theory, once the crime range becomes this large, it's not going to be of much investigative help, other than to say "Someone in Park City and/or Wichita". Generally, very distant offenses, like the Selina one, would suggest a mobile offender (as Rader was), and so one would then zoom in on the Park City/Witchita offenses and leave out those distant ones, in an attempt to find the more local area influences. Similar to how we then focused on the 5 in Witchita as well, and each time we did that, the hot spots continued to give us a more precise idea of a location that is, indeed, associated with Rader.

But again, one case does not mean it would work this well in general. It is nice, though, that this illustrates the ideal of what this sort of analysis is trying to achieve. To the extent that ideal can be reached consistently, well, that requires a lot of empirical evidence and testing.

- Jeff

[JeffHamm]

If it's the one at 800 East Waterman Street, that looks to be the closer of the two branches, which would be in zone 9.

Thanks again! Can I like a post twice?

- Jeff

[Trapperologist]

If you stick with the early series of crimes in Wichita, what I'm saying becomes clear when you look at the aerial view of the city. He's hitting the residential areas in a fan radiating from what looks like the downtown area, a place where there would be several anchor points, that does not have to include the residence. My idea of inputting the proximity to the buffer zone or border, in this case the city core, probably won't work because he's targeting the residential area outside the core. But some simple land use inputs for the main yellow zone might shift the focus in relation to residence or primary anchor point, would it not?

https://www.google.ca/maps/place/Wichita,+KS,+USA/@37.692066,-97.329176,10014m/data=!3m1!1e3!4m5!3m4!1s0x87badb6ad27f182d:0x9396d 5bf74d33d3e!8m2!3d37.6871761!4d-97.330053

[JeffHamm]

If you stick with the early series of crimes in Wichita, what I'm saying becomes clear when you look at the aerial view of the city. He's hitting the residential areas in a fan radiating from what looks like the downtown area, a place where there would be several anchor points, that does not have to include the residence. My idea of inputting the proximity to the buffer zone or border, in this case the city core, probably won't work because he's targeting the residential area outside the core. But some simple land use inputs for the main yellow zone might shift the focus in relation to residence or primary anchor point, would it not?

https://www.google.ca/maps/place/Wichita,+KS,+USA/@37.692066,-97.329176,10014m/data=!3m1!1e3!4m5!3m4!1s0x87badb6ad27f182d:0x9396d 5bf74d33d3e!8m2!3d37.6871761!4d-97.330053

Not quite sure I'm following you, so correct me if I've clearly misunderstood. But the short answer is, I suppose so. If you include new inputs of some sort, they will influence the final output. Whether they do so in a way that helps more often than hurts is the question though. One can't decide "oh this is better/worse" simply by looking at how one specific case changes. Putting in additional information, for example, could result in shifting things away for Rader, but those same inputs might greatly improve 5 other cases, and have no real effect on the rest. If that were the case, then it would probably be considered a benefit overall, even if Rader's case got worse.

There are attempts at building in local area type information. For example, one might want to include a population type overlay, since for Rader his targets were people in their residences, so one might want to mask out areas like "industrial parks", or other, non-residential zones. Some will model the road systems, and rather than just calculate distances in Euclidean or Manhattan distances, they will calculate distances along actual roads. And so forth.

Those can be useful, the latter probably when dealing with a very small crime range, where the fine details start to matter more. I could see that being potentially the case in the group of offenses you're talking about here. Those sorts of approaches are, of course, very time consuming to set up. You need access to the digititzed road maps, and so forth. My focus has been on this more general level at the moment, trying to see if I can improve the performance of what I've got so far, in part because the program is trivial to use (load a map, click on the crime locations, set the scale, and click "do it", and it works out the spatial probability map. One can then just mouse over the image to see zone numbers and so forth, and place markers for suspects, or places of interest, and so forth. If I can continue to improve it, I'm hoping to eventually be able to get a far better, and larger, set of crime series data and make sure what I've ended up with does generalize to other series as well. So far, I have one set of crime series that I build from, and then a second set that I use to test it. So far, it's looking good and performance on the test set is as good as on the data from which the routines are derived (though there have been one or two ideas that looked like they worked, only to be something idiosyncratic to the developmental series - that's why I need the test set too ).

But yes, it's a rich area of research, with lots of potential. The main point, though, is that even fairly simple approaches do produce results that more often than not provide useful information. Again, so far with the BTK series, the anchor point has been falling in zone 1, that's not all that uncommon, but it's not the majority result either (probably around 1/3 I think my last test run found; though that was just looking for the residence as the anchor point; here we're being a bit more complete with BTKs spatial influences).

Regardless, it is just a probability map, it's not proof. And the information is not "someone's back yard", it's neighborhood, or regions. The real investigative work still needs to be done. This just suggests areas that might be worth targeting first.

- Jeff

[Trapperologist]

You understood the inputting part. Further input would have to be based on behaviour that’s based partly on geography and partly on some theoretical psychology. Unless he has a condo in the downtown core, the anchor there is non-residential. It’s also likely not just an anchor but a junction or a pivot point. He doesn’t live there but instead commutes. Until he kills in Park City where his house is his anchor and pivot.

[JeffHamm]

You understood the inputting part. Further input would have to be based on behaviour that’s based partly on geography and partly on some theoretical psychology. Unless he has a condo in the downtown core, the anchor there is non-residential. It’s also likely not just an anchor but a junction or a pivot point. He doesn’t live there but instead commutes. Until he kills in Park City where his house is his anchor and pivot.

Yes, actually, all of the geographical profiling approaches are based upon underlying theoretical psychological constructs. Rigel (Rossmo) and Dragnet (Canter) are based upon the principle of risk reduction. Both work on the princple that as you get further away from your anchor points are you venturing into less and less familiar areas, and committing an offense in an unfamiliar area increases risk. Therefore, they reason, the frequency of offenses should decrease as you get further from the anchor point. Distant clusters of offenses (as with Rader's small cluster of 2 in Park city, and the larger "downtown" cluster) are probabily reflecting different anchor points (which is what it appears to be with Rader's crimes; his home in Park City was the main anchor point at that time, while his wife's workplace and later his, were the main anchor points for that group). There would be others, he was an active member in his church, so that could potentially be an anchor point, and he was doing some courses at the University, and prior to the crimes worked at Cessna, all locations that would result in him regularly travelling to those locations, building up a familiarity of the local area (i.e. acting as an anchor point for his spatial maps). Rosmo also includes a buffer zone around anchor points, which reflects the notion that drawing attnetion to yourself increases risk, so committing lots of offenses right next door is not wise as it draws the police to your area, etc, so Rosmo's routines will start to decrease the expectation for offenses to be located in an area as you get closer to an anchor point. Dragnet does not seem to have this aspect built in. The version I'm working on is based not so much on modelling risk aspects, but rather is based upon aspects of travel and trying to extract things like the offenders general flow of movements, the maximum distance they might travel, the minimum distance, efficient travel, and so forth, and from that try and work out where their journeys appear to be starting from.

I've marked in yellow those anchor points I mentioned above, and adding his home, that makes for 6. And all 6 of them fall in zones 1-20 (actually, 18 or less). The grouping of those points around the downtown cluster being close together, probably get smeared a bit in terms of their influence. His wife's work place (in zone 4) and his work place (ADT, getting more sure that's the one; is in zone 9), are high enough to warrant interest (particularity combined with his residence in zone 1), and the fact all 6 of the known main locations for him are in the red (which would only have about 1.6% chance of happening at random) might make him of interest - this is presuming his name came up on a list, and they entered known information about POIs, which is one way this information can be used. Also interesting is that he made the phone call from a station not far from where he and work mates were having coffee, and that was made from zone 6, showing how he's familiar with that area (he knew there was a phone at the gas station, for example) and it appears from work it was a part of their routine to move towards the "hot spot". Again, that's what geographical profiling is based upon, the notion that normal daily activities are conducted with respect to a set of spatial anchor points, and we journey out from them. So while Rader might have had to go to all parts of the city at various times, all of thoes journeys would include part of it back at his work place, he's going to travel there multiple times, while the more distant parts of the city he will travel to less frequently as they could be in different directions after all, but all of his journeys start and eventually return to, his work place, hence anchor point familiarity is high relatively to more distant locations.

- Jeff

[Trapperologist]

I was disappointed when I read Gordon Kerr's illustrated book on geoprofiling and only one crime map had the known offender's residence mapped, and that was the unorganized Vampire Killer who killed someone outside [a few blocks from] his home (BTK was not included), much less a comparison to the schematic, choropleth map.

Correction: it was the Richard Ramirez map that showed the residence. Although they didn't run the program to see if it would come up as an anchor, I assume it would since the murders ran in a flat arch to the NW of his location. That's very similar to the arch of the BTK around his non-residential downtown core anchors.

Although he's a fairly organized killer, he was flat poor and a loner so he wouldn't have a lot of anchors anyway. In this case, geographic profiling might have helped but, again like the BTK, his anchor is downtown. It's a high density area. If a killer could possibly live in a low density residential area in the suburbs like BTK, (no reason he's a commuter rather than a transient and/or downtown dweller), the program should pick it up before he kills close to home, especially after 7 kills. I don't care if it lights up three suburbs. Rigel highlighted three high density, probably transient, areas in Whitechapel for the Ripper. That might only help the lodging house keepers at the time.

[JeffHamm]

Correction: it was the Richard Ramirez map that showed the residence. Although they didn't run the program to see if it would come up as an anchor, I assume it would since the murders ran in a flat arch to the NW of his location. That's very similar to the arch of the BTK around his non-residential downtown core anchors.

Although he's a fairly organized killer, he was flat poor and a loner so he wouldn't have a lot of anchors anyway. In this case, geographic profiling might have helped but, again like the BTK, his anchor is downtown. It's a high density area. If a killer could possibly live in a low density residential area in the suburbs like BTK, (no reason he's a commuter rather than a transient and/or downtown dweller), the program should pick it up before he kills close to home, especially after 7 kills. I don't care if it lights up three suburbs. Rigel highlighted three high density, probably transient, areas in Whitechapel for the Ripper. That might only help the lodging house keepers at the time.

No, the programs won't pick up commuters' residences. The analysis is trying to pick up the anchor points in the vicinity of the offenses, and a commuters residence is to distant to produce much impact. And, Rigel produced three peaks, but the whole profile basically covers most of Whitechappel; it's just less spectacular when offenders aren't in the peak so those cases aren't publicized. This gives a very skewed impression of what these sorts of analyses can do, and what they are supposed to do. While it's great if they do locate the offender in zone 1, and that's not uncommon (I think my test sets are around 33% in zone 1, and 50% of the cases are zone 3 or less?), we're still dealing with around 50% of cases will be further down the probability list, and some will just be wrong.

Both Rigel and Dr. Watson highlight downtown areas associated with Rader when he was committing his crimes in that region. When he shifted to Park City, the influence of his residence as an anchor point emerges. Would a downtown anchor point identify him? No, of course, not, it's not about identifying the offender, it's about identifying locations where police efforts might be worth being directed to.

With JtR, one of the peaks is around Flower and Dean, and the police of the day were checking that area out quite a bit. There's another down around the graffiti, and one a bit nothwest, but the whole profile spans pretty much the whole area. The crimes are clustered close enough together that Rigel more or less says "start here, although he could be anywhere". But the idea is that Rigel gives a way of prioritizing that "could be anywhere", and generally, following that search pattern finds the offender sooner than randomly searching the area.

- Jeff

[Trapperologist]

Alright, Jeff. As long as it's understood that the program doesn't necessarily have to do with residence and therefore doesn't necessarily support the "unknown local" theory. It's obvious to me that if he had any sense and a family, he wouldn't create "terror" where he and his family lives. I think the bodies of both neighbors BTK killed were driven far afield as well as one's purse and car. Was that to distance the crime from his home residence where he lived with his wife and two kids?

[JeffHamm]

Alright, Jeff. As long as it's understood that the program doesn't necessarily have to do with residence and therefore doesn't necessarily support the "unknown local" theory. It's obvious to me that if he had any sense and a family, he wouldn't create "terror" where he and his family lives. I think the bodies of both neighbors BTK killed were driven far afield as well as one's purse and car. Was that to distance the crime from his home residence where he lived with his wife and two kids?

No, that's correct. These analyses that I've been presented, like Rigel and Dr. Watson, are generally based upon looking for offenders that are marauders meaning, they have an anchor point of some sort inside the crime range, which is not necessarily their residence. That's generally because commuters (true commuters, those who travel to an area that they have no other connection with except for their offending) are pretty rare. Even defined by residence, around 80% of offenders are marauders, but of those 20%, many "commuters" would actually be marauders because they work in the crime zone (or have some other daily connection). Rader is an example of that, at least early on. Later his offenses spread to Park City and that expanded his crime range to now take in his residence as well.

But no, the anchor points need not be their residence. There is some literature looking at trying to separate out marauders from commuters, and I believe by commuter they mean the strict case (no association with the crime range other than their offending). Even then, an offender will develop a spatial map of the area through their stalking of the area, and so develop anchor points which will probably focus in on major intersections and roadways, for example. Those would still be useful locations to search, but it would take a lot of good police work to find the offender as they have to be noted in the area while they are actually there as they can't be placed there due to an association with work, let's say.

Anyway, don't worry, if you're favorite suspect isn't a local, this sort of analysis doesn't rule them out. While commuters are more rare at around 20% (residence based definition), they're not so rare that whomever you favor would be unreasonable to consider.

And yes, the only time Rader moved a body (that we know of) was in the two murders close to home. Probably more to do with "risk reduction", he committed those offenses too close to home, so he tried to distance the focus of the investigation away from his immediate area. I doubt he had any concern for his family, it would all have been his concern for himself and trying not to get caught.

- Jeff