Archive through October 03, 2000

** This is an archived, static copy of the Casebook messages boards dating from 1998 to 2003. These threads cannot be replied to here. If you want to participate in our current forums please go to https://forum.casebook.org **

Archive through October 03, 2000

Casebook Message Boards: Ripper Suspects: General Discussion : Simplicity - Occams Razor: Archive through October 03, 2000

Author: Keith Rogan Thursday, 28 September 2000 - 03:49 am
Jill, Looking forward to it! Keith
Author: Jill De Schrijver Thursday, 28 September 2000 - 07:50 am
Hi Keith, I'm sorry for you but I think I'll be having some bad news: I made two epicentres based on 3 undiscussable victims (Nichols, Chapman adn Eddowes)-the pink one- and then on 4 (Nichols, Chapman, Eddowes, MJK) - the orange one. Neither Stride, nor Tabram would have any effect on the last epicentre (see the 4 points as on a circle). So we have the pink epicentre (PE) and the orange epicentre (OE). Thus follows that we have two environment circles (PC and OC) with a radius from their epicentre to the farthest point of where the victims lived, Dorset Street. The area's encompassed by the largest one, that is OC (not visible, imagine it), have in it almost all the poor buildings of Spitalfields and Whitechapel, a tiny poor area in St-George and only a part area of the district between Whitechapel Road and Commercial Road. Note that of the last district almost all poor blocks are to be found in that area. Graphically you could say that OC would exist for 60% of the poor buildings of Whitechapel and Spitalfields, and 35% of the district between Whitechapel and Commercial Rd. If we take in mind that although the area at Spitalfields maybe wasn't that big, more people were stacked with each other there, than in other poorest area's. Just a simple calculation makes that in OC at the most 700 prostitutes live in another area, against the 1200 of Spitalfields and Whitechapel. Thereby the number of Prostitutes solliciting on Whitechapel Road would by majority be from the area where all the victims came from. Therefor I don't see any point in making the significance calculation. Greetings, Jill
Author: Jill De Schrijver Thursday, 28 September 2000 - 08:24 am
On the other hand I'd better. Because I think you have been calculating already, and thinking he still had 37% of encountering a prostitute of another district, and thus of say 5 victims at least one should have been from the other district, not Sitalfields and Whitechapel. This would only be true if this chance was a total linear one and a thing like rolling a dice. The victims were not picked as a dice, and they didn't walk the streets with the same chance as if rolling them. No, instead this is about 'correlated statistics', and here the question is not how much chance did JtR have to encounter a Whitechapel prostitute or not. But is it odd (significance) that although he had 37% chance to take prostitutes from other area's, he took 5 from the Spitalfields and Whitechapel area? So I will for safety check it, but by experience my guess is that even 7 is not odd enough. Now if there had been more than 10 ... First have to find the formula this week. To be continued, Jill
Author: Keith Rogan Thursday, 28 September 2000 - 01:47 pm
Jill, As I've confessed, statistics are not in my area of expertise. But, to round these numbers off - lets agree that Spitalfields makes up 2/3 and "other" makes up 1/3 of our prostitutes. (In reality, I think I would draw the circle much larger since two of the killings actually fall outside it...?). But anyway, even if we go with this OC it means (to me,) that if I take 3 playing cards with two of them Aces representing Spitalfields, and one of them a Deuce representing "other" I'd have to draw four times without hitting the deuce to duplicate The Ripper killings, no? Pretty long odds. If we drop Millwood, Tabram or Stride into the equation, I have to draw 5, 6 or 7 times without hitting the deuce. I would not care to wager against such odds even at the more conservative figure of four draws.
Author: Jill De Schrijver Friday, 29 September 2000 - 04:30 am
Hi Keith, The difference between cards and deuces, against a matter based on behaviour of humans is that with cards the chance of picking one card against another is the same (1/52) so any statistical calculation is based on each card having the same chance to be picked. With reality, say like an accident this means something else. You can have a statistical figure of like 5 accidents per person in his life (which in itself is an accurate statement, just as saying JtR had 37% chance to meet a prostitute of the non-Spitalfields area). On the other hand there will be people who never will have an accident, and others will have 8 in their lifetime. That is why the previous corect statement is not applyable. Because having an accident is not based on everybody having the equal chance of having an accident, but on the driver's driving mentality, skill, eyesight, weather, speed, motoric reactions, alcohol percentage, ... Now both these men are with the same insurance company, and they want to check if the last one is an exceptional dangerous driver: they will find that him having 8 accidents is within the range to be expected, and that he statistically can't be called an exceptional dangerous driver. Another illustration would be the statistic that 10% of the children are abused. This still doesn't mean that if I have 21 colleague pupils in my class that 2 of them at least are abused. Still another example is voting for the goverment. You could vote for say 5 parties (as in my country) and one of them is racist-nationalistic. The racist party got 25% votes. Since my parents, me and my boyfriend together makes 4, one of us should have voted for the racist party? Of course not one of us has. I'm sure you can come up with a lot of reasons for it, a lot of them demographic and social. Hope this helps to explain the difference between playing poker and killing East-End prostitutes. We are not calculating the 'odds', but the 'oddities'. About the pink circle: it was based on the epicentre of the 3 murder-places with a radius of the living quarters, which are closer to the epicentre than the killing area's like Mitre Square or Buck's Row. The orange circle (4 or more murder places) I haven't physically drawn, but the blue (poor) blocks that would fall in this circle I have put on the map. If I would have drawn a circle only related to the murder places (thus a bigger one), not that many poor blocks would have been added (check the two maps of the area on the link you gave me). One large block as big as the one between Whitechapel High Street and Old Montague, in the Spitalfields area beween Quaker and Commercial Street, one little block extra from St-George, nothing extra from the district between Whitechapel and Commercial Rd. (because all the poor blocks of it all already counted within), and one block as big as Dorset Street and White's row together in Bethnal Green. I haven't even put all of the boor blocks of Whitechapel that fall within the orange circle now, because they are spread around and are more or less just a building. Greetings, Jill
Author: Jill De Schrijver Friday, 29 September 2000 - 09:50 am
Hi Keith and all, I've got the results. Luckily I found a site to do the calculations for me, otherwise you'd probably had to wait some time longer. The calculation link is: http://faculty.vassar.edu/~lowry/ For theoretical interest: http://www.statsoft.com/textbook/stathome.html I have used the Kolmogorov-Smirnov One-Sample Test for Frequence Data (we are talking about serial murder, aren' we?). An alternative in other circumstances would have been the chi-square test. Both are fit-tests, and they assess to what degree the observed pattern differs from the pattern that would be expected on the null hypothesis. In human language in our case: There is the hypothesis that JtR killed only prostitutes from the Spitalfields+Whitechapel area for a reason, and the null hypothesis means that there is no relation between the victims and their living quarters. The expected pattern of the null hypothesis would then be the chance JtR had either picking a prostitute from Spitalfields or from another area. The test then checks if the pattern we have seen of maybe up to 7 victims living in Spitalfields is just unsignificant deviation from what should be expected, or that more is going on. The result of the test is the probabilty that the null hypotheses is still true, and that there is nothing out of the ordinary going on -> the lower the probability the higher the possibility of a relational pattern (NO YOU ARE SEEING TO MUCH IN IT) ->The higher the result, the less we can believe that the supposed observed relation exists (YES, YOU HAVE A BASIS TO BE SUSPICIOUS). The difference between the one I used and the chi-square one is that the first is for really small samples. I had to fill in several data: First there are two categories: A)Spitalfields+Whitechapel B)another district It was agreed that in A lived 1200 prostitutes and in B 400 (one third of 1200). Thus follows that for the null-hypothesis it can be expected that JtR should have in total killed 2/3 of the victims from A and 1/3 from B. Expected proportion for A: 2/3 Expected proportion for B: 1/3 What we in fact observed was that from B there were 0 victims, and all were from A no mather how many we count. Thus I have calculated the significance for 3, 4, 5, 6 even up to 7 victims. The there are also two degrees of freedom we can give the test: .05 or .01 1)Here are the result if we only take 3 victims (Nichols, Chapman, Eddowes) into consideration: -> between 71% (for 0.05) and 83% (for 0.01) probability that this is just a meaningless deviation 2)The result for 4 victims (Nichols, Chapman, Eddowes, MJK): -> between 62% (for 0.05) and 73% (for 0.01) probability that this is just a menaingless deviation 3) The result for 5 victims: -> between 56% and 67% probability that this is just a meaningless deviation. 4) The result for 6 victims: ->between 52% and 62% probability that this is just a meaningless deviation. 5) The result for 7 victims: ->between 48% and 58% probability that this is just a meaningless deviation. Greetings, Jill I'll leave further personal interpretation for you for a while.
Author: Keith Rogan Friday, 29 September 2000 - 12:51 pm
Jill, I'm afraid you've totally lost me in the smoke here.
Author: Diana Friday, 29 September 2000 - 10:53 pm
Jill is very smart and I am very tired and not mathematically inclined!
Author: Warwick Parminter Saturday, 30 September 2000 - 05:47 am
I really hate to show my ignorance but, will someone please explain &74 to me ?, thank you. Rick.
Author: George Sotiriou Saturday, 30 September 2000 - 03:20 pm
What? Pardon? Can someone explain this to me in english. Have to go now, have a headache. Ouch! G
Author: Jill De Schrijver Sunday, 01 October 2000 - 05:07 pm
Hi All, Explenation in plain English: Keith had the notion that Jack the Ripper especially only picked prostitutes who lived in the area of Spitalfields, and ignored prostitutes from other area's. If we say agree that 5 of the murdered women were murdered by JtR, normally it would have been ecpected that 1 or 2 of them should not be living in Spitalfields. Thus the observation differs with what we expect. I have calculated how probable it is, that the difference of what we observe and what we expect is due to mere chance or because JtR only wanted prostitutes from Spitalfields. The conclusion of these calculations are that:If JtR has murdered 5 women from the Spitalfields area that there is between 56% and 67% chance that the fact of the victims living purely in Spitalfields is pure luck. I even think this would be higher, because the basic figures of the prostitutes from the other area are overrated. Only when we consider he killed at least 7 victims only from Spitalfields can we say that something odd is going on and that this is up to the most 58% by chance. Keith's theory only will become really convincing if he can proof 8 victims, living in Spitalfields area, of the hand of JtR. Keith can start discussing the numbers by turning them around, but at the moment there is no statistical foundation for his theory, although it should not be totally discarded, depending on the number of victims. Hope this relieved some headaches and greetings, Jill PS I wrote the previous post in a more statistical approach so I won't be blamed for keeping the foundation of this conclusion behind, and that anyone else can recalculate it on his own with the link.
Author: Keith Rogan Sunday, 01 October 2000 - 06:30 pm
Thanks Jill, I think I'm following you now. Your earlier note was in "statisticalese", a language I don't speak. The only fault I can find with the numbers are that you are basing your findings from a circle drawn so small that two of the murder sites fall outside of it. I realize that it was my suggestion to use a 1 kilometer circle, but I didn't realize then that some of the murder sites would fall outside that zone. Knowing that, I would draw the circle much larger - perhaps a mile or even 1.5 miles (there is a point where the circle would get so large as to be meaningless - perhaps 1 mile would be most appropriate). Anyway, if the circle is large enough to at least contain the murder sites it would also include more potential places for the victims to live and alter your results appreciably. Do you see what I mean? I can't see how your "OC" can be valid when it excludes the murder sites.
Author: David M. Radka Sunday, 01 October 2000 - 07:52 pm
The problem with the statistcal approach above, as I see it, is that JtR only killed 5 people. That's not a big enough population to get statistically valid results. It probably just worked out that he got a 100% particular population with five; if he'd had killed 27, surely the statistical spread would be more nominal, as asserted above. Humbly, David
Author: Jill De Schrijver Monday, 02 October 2000 - 03:35 am
Hi Keith, Hi David, Keith: In my post of Friday, September 29, 2000 - 07:30 pm, I've explained that having a bigger circle including all the murder sites would not change things very much. It even would actually would make the odds of prostitutes living in an other area, other than Whitechapel-Spitalfields to 1/4. Look up the blue blocks in the map-link you gave me. David: the test I used for my calculations is designed for low numbers and samples, and to search for the significance in such a small deviation. It is used when the expected frequenties are lower than 5 (in this case up to the most, for prostitutes to other area's, 2), while the chi-square requires bigger samples. The null-hypotheses presumes a nominal spread. The expectations figures are not based on 5 murdered victims, but the presumed number of prostitutes living in the different area's. And if any of the base material used is shaky, I think it is the last. It is the most inaccurate, and the rest is derived from it. But as I comprehend any of the living conditions of the surroundings, any precision would dispell a relation even more. Greetings, Jill
Author: Jill De Schrijver Monday, 02 October 2000 - 10:19 am
Keith, For your pleasure I've drawn the extra poor blocks for a RC (red circle that includes murder sites). By the way wasn't it you who said that the murder sites didn't matter, but the area they lived: so that's the only thing that mattered when I drew the first pink circle and orange one. Yellow- spots are living quarters from victims Purple-spots are murder sites. Pink circle (PC): epicentre of 3 murder sites, radius farthest living quarter (Miller's court) Orange circle (OC): epicentre of 4 murder sites, radius farthest living quarter (Miller's court) Red circle (RC): epicentre same as OC, radius farthest murder sites (Mitre Square & Buck's Row) Blue blocks: poor buildings calculated within PC, and OC Red blocks: extra poor buildings within RC, other than with PC and OC. Green borders: borders of Whitechapel, and other area's. Greetings, Jill
Author: Keith Rogan Monday, 02 October 2000 - 12:58 pm
Jill, I'm confused again Basically, you now have a circle that covers the bad neigborhoods south of the Whitechapel Road as well as our original neighborhood and I think, a much more valid sample. If our dead prostitutes all lived in one neighborhood comprising 25% (20%?) of the poor neighborhoods, that seems (to me) to be pretty significant. It lends credence to the theory that our man was a stalking type killer rather than a killer of opportunity. Remember, at least for the last 3 killings, all of Whitechapel was blanketed with police within hailing distance of each other. People were being observed and questioned, notes were being taken, and yet our man walked through all of this, killed and got back to his quarters. Much easier to explain if: a. He lived in the neighborhood and looked as if he belonged - or was even recognized and regarded by the police as trustworthy. b. He planned his attacks around police patrols (requiring observation). c. He knew the prostitutes and was liked/trusted by them - the fact that they all lived so close lends credence to the fact that they knew him from THAT neighborhood. All of this points to the fact that he was a local and somebody whose presence on the streets did not raise either police or prostitutes eyebrows. He was not a muttering maniac, skulking through the streets. He was probably not a foreign Jew (who were under suspicion at the time). He could have been a cartman who went back and forth from the docks to Spitalfields Market at all hours - he could have been a street vendor. Whatever he was, he was likely seen and questioned many times and drew no suspicion whatsoever. He was wallpaper. Keith
Author: The Viper Monday, 02 October 2000 - 01:17 pm
Hmm, it looks as if this board has finished up as anything but "Simplicity"! Is anybody else out there completely unable to follow this statistical approach? I'm still failing to understand how you can produce all this with so little data, Jill. Unless you know the population of the circle you are sweeping; the population of the coloured blocks; the percentage of the prostitute population that lived in them and probably several other things besides, how can you calculate anything accurately? Regards, V.
Author: Keith Rogan Monday, 02 October 2000 - 06:21 pm
Viper, I'm also confused on the statistical language above. The way I look at it is like this: If the Dorset/Flower & Dean neighborhood makes up 25% of the doss houses in our zone, then the killer has a 25% chance of any particular prostitute being from that neighborhood. If he kills 4 prostitutes and all are from there, then he has beat the odds four times in a row - fantastic odds! Lay down four cards, 3 deuces and an Ace. Now blindly draw the Ace four times in a row - how many hundreds of repetitions before you hit the ace 4 times? That's the odds Jack was dealing with. Conclusion: It wasn't chance. He lived or worked in the immediate neighborhood and CHOSE those women. Keith
Author: Warwick Parminter Monday, 02 October 2000 - 07:56 pm
Hey Keith, a cartman, you said, a street vendor, what about an ex fishmarket porter turned orange stall vendor or work at anything guy in Spittlefield market!.A quiet man, respectable, dressed well in his leisure time, a pretty young wife---they look good together, but she does bring him down a bit with her excess drinking now and again. He doesn't care for her friends much but he's civil with them and they with him, he stammers and seems uncomfortable in the presence of more than one woman-- they smile, they think he's shy, but they also think,-- he would look after you!. Rick
Author: Warwick Parminter Monday, 02 October 2000 - 09:09 pm
Hello Jon, thanks for the post over the weekend. But I'm sorry you have me wrong, I was being a bit contrary in my words. There's nothing I don't think the human being,(certain ones) wouldn't stoop to, especialy the male. There is always more than a touch of cowardice concerned too, isn't there?, it's nearly always women or animals they vent their rage or revenge on. I don't think (for myself) that the reason was to take organs from the bodies-- look at the way he ripped them open,-- Eddowes and Kelly had both been turned into a dirty mess, so had Chapman by what I have read of her death scene. No one will ever convince me that the being who did that was qualified in anything except slaughterhouse procedure, and maybe not even that! What I don't understand Jon, is you saying, there need be no suicide, accident, or incarceration to account for the killings coming to an end if the reason was the aquirement of organs for money. Yet you won't accept Barnet doing it for love, then getting over it and coming to terms with the rest of his life. As you quoted Jon,"---what was that thing you quoted from Shakespeares Hamlet? J My Regards Rick.