Here is a proff report that probably can learn us somthing:
https://docs.google.com/viewer?a=v&...YdoelC&sig=AHIEtbRZge4vEbSC5XvG87t5tVcErRi5qQ
Behavioral Science and the Law
Behav. Sci. Law 19: 595-610 (2001)
DOI: 10.1002/bsl.431
RESEARCH REPORT
Spatial Patterns of Serial Murder:
An Analysis of Disposal Site Location Choice
Samantha Lundrigan, Ph.D., and David Canter, Ph.D
I love it, it shows, there is still hope when psychologists use math. Lets go in the facts, won't we?
1.) This study uses for the distance beelines, or was they word it "as the crow flies". Problem is, the most serial killers can't use beelines. They have to use roads. For example, Corll had to his boat shed beeline about 9 miles. But the real driving distance was about 14 1/2 miles. Means, for Corll at least, the basic value of distance used in the study has a failure of 50% from the start. And we can be sure, the differences between beelines and real driving distances are also existent for all other serial killers. So, any study based on value pairs with about 50% failure would normally be rubbish in mathematical sense, unless the authors of the study could prove mathematically correct, that their failure is in relation to the base value almost constant. In this case, the study would be mathematical rubbish, but the tendency it shows would be still usable because only the relations changed. And that is not an opinion, but mathematical fact.
2.) The study could, for obvious reasons, only use serial killers who were caught and had at the time of body drops an exactly known address. This brings two major mathematical problems with it:
a.) The base set of the statistics is very limited, in this case to 126. For comparison, Hickey used 399 in his database in 1994. The variance of failure increases to the second of the reciprocally proportional of the relation in size between two base sets. Means, 1/2 the size - four times the failure variance, 1/3 the size - 9 times the variance in failure. Failure means, the impact of extreme and incorrect data in the base set. So by mathematical rules, we have 9 times the impact of extremes and mistakes in the data collection than we had in Hickey (which wasn't too stable either).
b.) The study also has to rely on caught serial killers for obvious reasons. That means, it implicitly excludes those who were smarter or luckier than the average and didn't get caught. The Zodiac for example couldn't be in the study in the first place, because nobody knows where he lived, he got never caught. He left his victims where he killed them and we know, he didn't live where he killed, because the next places people could leave were a little away. Thus, the study implicitly excludes those, who did things different and probably therefore, weren't caught in the first place. So the base set is necessarily incomplete and all absolute results are considered wrong in the sense of mathematics. However, since Statistics is often forgiving and most serial killers are anyway not caught by investigative measures but rather by parking in forn of hydrants (Berkowitz) or speeding (Bundy) or similar stupid things (Rifkin transported a body in a truck with expired plates), there is still a good chance, that in the set of caught serial killers is a sufficient number of the bright ones caught by accident. That would technically compensate for the first failure, the implicit exclusion. But since we don't know the degree of compensation, this only brings the study (always in the sense of mathematics) from utterly wrong to highly doubtful.
3.) As all studies of this kind, the base set is defined solely by a definition of serial killers by number of victims and the existence of a cool down phase. There is no detailed look on organizational degree of the offenders nor on typology in general. So a low IQ offender (like Toole), a disorganized psychopath (like Berdella or Sowell or Franklin), a mission-driven delusional SK (like Herb Mullin) and so on are all equal parts of the same base set.
This is entirely okay for a study determining a rough tendency or is on the search for the statistical methods for the subject, but not sufficient to conclude from it with certainty in the first place. Not even, if the study wouldn't have the first two problems.
4.) The authors described extensive cleansing of the sample for the sake of the statistics. So they cut out extreme long distance cases because they would raise the average too much. However, later on, they describe the use of the median instead the average. Using the median in such a case is surely the better criteria, but to make it work, they would have needed to add the earlier excluded long distance cases back into the study. The way, they did it, they cut out cases and used then the mathematical function which would have dempened bot not eliminated the impact of those cases. In other words, technically, they cut out those cases two times.
5.) This study, since the base set is caught SKs with defined addresses, doesn't deal with the subject of the highly mobile ones in the first place. A long-haul trucker SK wouldn't be included and if, would cause an over-proportional impact. Imageine the guy having a home in Ohio but because he is on the road all the time with his truck, kills in Texas, New York and wherever his tours lead him. The distances would be enormous. A mathematical statement is proven wrong if there can be even only one case brought up for which the statement is incorrect. Now, statistics and it's results are not entirely statements in the sense of mathematics, neither should they be treated as those. However, we should keep in mind, that there are SK types who deviate from the tendencies in this study for reasons of their type. Which proves, this study isn't really covering all serial killers but only certain types.
Now, after pointing out the five biggest mathematical glitches in this study, of which each would be enough to rip it to pieces if we were mathematicians, lets do a little comparison to real life experience.
1.) The home as base of operation can only be valid for offenders, who have a home in the first place. Any kind of drifter would show an entirely different picture. Since there is under the serial killers a significant number of homeless, even those cases rarely catch media attention, we would have to determine this part of the typology BEFORE we can even use of any conclusion from this study.
2.) The home as base of operation can only be valid for offenders if the offender spends a significant time at home in the first place. I mentioned the truckers above, so need to elaborate a second time.
3.) The home as base for a distance calculation is only valid for the time, the offender has that home. If he moves, the game starts again. He has to become familiar with the area, he has to scout the hunting grounds, he has to find new dump sites. Therefore, the distances change after an SK moved. This part is not included in the story, the comparisons were all drawn as if moving wouldn't mean any change of the outside circumstances, which we all know (hopefully), is wrong.
4.) A number of SKs have defined dump sites. The GB4 were basically all near to each other. So there is near to no deviation in distance to his home, wherever he lives. And there is also no variation in traveling angle, which can anyway only apply if the killer would be able to travel beelines, so that part of the study is entirely useless at all.
In general, for all cases with SKs who build trophy gardens, there can only be variation in the distances as far as they have several such dump sites at the same time. Bundy for example had three in Washington State, Corll had three at his boat shed, at High Island Beach and at Lake Rayburn. All those dump sites fulfilled one or more of the following criteria, of which is none implemented in the study:
- accessibility
The killer needs a possibility to reach that place
- security
The killer needs at least some safety not to be caught during the drop
- comfort
The killer has to be comfortable in that zone. Bundy used places, he knew from hiking, Hanson places, he knew from hunting. Means places connected to other activities in the SKs life.
- emotional bond to the place
A lot of SKs who build their own little burial grounds, use areas they feel as calming and nice. In my opinion, that is the reason, why so many of them go to sea shores, lake shores or forests.
- possibility of revisiting
A number of SKs revisit the dump sites.
5.) Experience tells me, SKs use several ways of body disposal. Some dismember the victims and put them to the trash, others keep them in their houses and again others transport them to dump sites, which may or may not established trophy gardens. However, it is highly unlikely, an SK takes a body and transports it just a mile away into an area, that tells everybody in the first place, there is only one human settlement near. And even, if in that study, the median distances, they calculated are much too short because they used beelines instead of real traveling distance, they come to much bigger distances than Oak Beach to Gilgo. Which now poses the supporters of a "killer from OB" theory for a problem because either they have to dismiss that study or their theory.
What we obviously can learn from all of those studies is, never trust any study before you did the math for yourself. Still, I think, this one has something. The idea is right, only the mathematics they use spoils it.
