Crime series of 8....

[mic_ads]

Throughout the numerous publications that have concluded on the 'true' number of Ripper victims, one observed feature of criticism has been the methodology and number of incidents examined before producing a final figure.

Of recent time, I find difficulty in accepting the conclusions of Trevor Marriot, due to the methodology undertaken and conclusions presented based on drawn opinion alone. There was no set approach as to how the individual crimes were compared to each other, or what criteria were used to differentiate between the individual crimes. Phrases such as possibility, suggest and perhaps were used in the conclusions, an indication of speculation.

I submit that, by using modern policing techniques to compare and analyse itemised behaviours between the 15 cases (canonical and suggested victims) in conjunction with a valid statistical approach, a complete Whitechapel crime series can be established. Degrees of similarity between individual cases, focusing on grouped behaviours (Control, Style & Violence) provided the results; for all behaviours there were 16 linked pairs and for the groups of Control and Style behaviours, there were 27 linked pairs (Please view the attached document to see the linkage charts). The statistical results were supported by further evidential examination of the individual cases that considered in the main, wound pattern analysis.

It is this combination of statistical and evidential analysis that has produced a group of eight linked offences, that consists of Martha Tabram, Mary Ann Nichols, Annie Chapman, Elizabeth Stride, Catherine Eddowes, Mary Jane Kelly, Alice McKenzie and Frances Coles.

I am happy to share these results with the forum. I appreciate not everyone will agree but that is the whole point of research - to explore, share and discuss.

[Cogidubnus]

I'm sorry mate but, not being that clever, could you advise me what the diagrams actually show?

Regards
Dave

[Wickerman]

It is this combination of statistical and evidential analysis that has produced a group of eight linked offences, that consists of Martha Tabram, Mary Ann Nichols, Annie Chapman, Elizabeth Stride, Catherine Eddowes, Mary Jane Kelly, Alice McKenzie and Frances Coles.

Some might say that you just decided to include all the victims?
On the other hand, are we to understand that you specifically ruled out Annie Millwood and Ada Wilson, or were they even considered?

I think one of the biggest problems theorists face is that they do not have an agreed, common set of criteria to follow, which raises the question of your own criteria. I think many will find your data and method of interest.
So yes, please present your offering, thankyou. I just hope you're wearing kevlar..

Regards, Jon S.

[K-453]

Hmm, a problem could be that murders resemble each other anyway, although committed by different killers. They are variations of a few methods of inflicting death: Suffocation, loss of blood, poison ...

[lynn cates]

bingo

Hello Jon.

"I think one of the biggest problems theorists face is that they do not have an agreed, common set of criteria to follow"

Bingo.

Cheers.
LC

[K-453]

mic_ads, which behaviours were compared, and which matched best?

[Sally]

Hi mic_ads,

Thanks very much for posting your results.

I'm sure everybody here would be interested to hear about your methodology.

Another member of this forum recently posted a link to an article written in 2005 which used profiling in an attempt to identify which of the Whitechapel Murders were Ripper victims.

http://forum.casebook.org/showthread.php?t=6422&page=29

(post 286)

It would be interesting to compare the results of this study with your own.

Do you intend to publish your study? It would be interesting to see your full analysis.

Thanks again.

[mic_ads]

I'm sorry mate but, not being that clever, could you advise me what the diagrams actually show?

Regards
Dave

Dave.
There are two charts on the attached document. The charts aim to show the paired cases that have a degree of similarity over .60 or 60%.

The 15 cases that were considered were Annie Millwood, Ada Wilson, Emma Smith, Martha Tabram, Mary Ann Nichols, Annie Chapman, Elizabeth Stride, Catherine Eddowes, Mary Jane Kelly, Annie Farmer, Rose Mylett, Elizabeth Jackson, Alice McKenzie, an unknown female referred to as ‘Pinchin Street Torso’ and Frances Coles. Fairy Fay was alluded to, however no details of her attack were located in records so the details were omitted from any further consideration.

The 15 cases were broken down into indiviudal behaviour elements that were grouped into Control, Style and Violence. Each case was compared against the others and paired cases > 60% were retained. Other combination of grouped behaviours could also be calculated.

The first chart shows the results of comparing all offence behaviours - 16 linked pairs. If you look at the icons for AMc (Alice McKenzie) and AC (Annie Chapman), the line that links them has a value of .71 or indicating 71% similarity between the two cases.

The second chart shows the results of comparing only the offence behaviours of Control and Style. The attacks on Elizabeth Stride and Frances Coles were documented as interrupted by a third party so they would not contain the full range of offence behaviours as in an uninterrupted attack. Comparing them against completed attacks would not produce clear results. The results was 27 linked pairs. Of note: MAN (Mary Ann Nichols) & AC (Annie Chapman) has 1.00 or 100% similarity, and FC & ESt had .93 or 93% similarity.

I hope this helps..

Mike

[mic_ads]

Some might say that you just decided to include all the victims?
On the other hand, are we to understand that you specifically ruled out Annie Millwood and Ada Wilson, or were they even considered?

I think one of the biggest problems theorists face is that they do not have an agreed, common set of criteria to follow, which raises the question of your own criteria. I think many will find your data and method of interest.
So yes, please present your offering, thankyou. I just hope you're wearing kevlar..

Regards, Jon S.

Jon

The 15 victims considered were Annie Millwood, Ada Wilson, Emma Smith, Martha Tabram, Mary Ann Nichols, Annie Chapman, Elizabeth Stride, Catherine Eddowes, Mary Jane Kelly, Annie Farmer, Rose Mylett, Elizabeth Jackson, Alice McKenzie, an unknown female referred to as ‘Pinchin Street Torso’ and Frances Coles. Fairy Fay has been alluded to previously, however no details of her attack were located in records so the details were omitted from any further consideration.

Of note, in my results examining just offence behaviours that fell into the Control and Style grouping, Annie Millwood and Ada Wilson linked to each other, but no other cases. There was 70% similarity indicated.

I agree that finding common ground to start researching such matters does raise questions. What I can say is that having reviewed other approched, I took a unique route. I took each of the 15 cases, dismantled them into individual offence behaviours, that were grouped as Control, Style or Violence, that were defined as below:

Control: Behaviours displayed gaining control of the victim, including how the victim was approached.
Style: Behaviours displayed reflecting the offender's personality or personal offence style including those not necessary for the completion of the offence.
Violence: Behaviours displayed in the infliction of violence.

Once all the offence data was broken down and coded, I used a statistical program and test called Jaccard's coefficient to compare paired cases. Any similarity >.60 were retained.

Mike

[Cogidubnus]

Dave.
There are two charts on the attached document. The charts aim to show the paired cases that have a degree of similarity over .60 or 60%.

Thanks for your detailed response. Much appreciated...can I just ask why 60% was chosen as a cut-off?

Cheers
Dave

[curious]

Control: Behaviours displayed gaining control of the victim, including how the victim was approached.
Style: Behaviours displayed reflecting the offender's personality or personal offence style including those not necessary for the completion of the offence.
Violence: Behaviours displayed in the infliction of violence.

Once all the offence data was broken down and coded, I used a statistical program and test called Jaccard's coefficient to compare paired cases. Any similarity >.60 were retained.

Mike

Hi, Mike,
Very interesting information. And if you were in show business you would have already succeeded. I want to know more.

I'd like more details about all the categories. For instance, what behaviors were displayed to gain control of the victim.

When you mention approach, I'm guessing you mean as displayed on the body since there is no way to know whether the victim and killer had a conversation or not. Or maybe you looked at something else.

Anyway, very interesting and any more detail you are inclined to share, you'd have an appreciative audience.

[mic_ads]

Thanks for your detailed response. Much appreciated...can I just ask why 60% was chosen as a cut-off?

Cheers
Dave

Dave

When I was deciding the best way to analyse the cases on a pairwise basis, the best statistical test for this purpose was Jaccard's coefficient. As the aim was to follow the process of crime linkage as used in lmodern law enforcement, it was appropriate. When I was looking for an interpretation of the results, I read a paper by Dr Christian Hennig that used .60 as the lower limit.

The closer the value was to 1, the higher the degree of similarity. The scales were:
Values > .85 = a “highly stable” degree of similarity.
Values > .75 and < .85 = a “valid and stable”degree of similarity and,
Values >.60 and < .75 = an indicative degree of similarity.
Hence the cut off point of .60 or 60%.

Mike

[mic_ads]

mic_ads, which behaviours were compared, and which matched best?

Whilst I say that behaviours were compared, they were, but not on a like for like basis.
To dismantle the individual crimes, I initially created a chart that had the victims as columns and the offence behaviours as rows. It was a case of reading through all material compiled for each case and filling in the boxes.

In terms of the behaviours, it was broken down into areas such type of force (blunt force trauma, sharp point injury) and the degree of force (nil to extreme). There were areas to cover victimology such as clothing, activity at time, intoxication, occupation, etc. I also covered areas such such as wound pattern and weapons. There were something in the region of 50 individual behavioural characteristics. They were recorded as present or not present. So I was looking at the finer detail of the crimes to a point where it was Yes or No, was this feature present or not. I had to bear in mind that there were restrictions in the data as I was dealing with historic data and the recording of such was limited by the police at the time. However, this was a factor that anyone who examines the Whitechapel crimes has to contend with.

The offence behaviours were grouped into the three categories of Control, Style and Violence as previous papers had highlighted this as the best approach.

I used SPSS software to crunch the numbers. I did not look for which behaviour(s) matched best, as I was able to select the combination of grouped behaviours as required.

It is fair to say that whichever combination of behaviours were chosen, when the results for all combinations were compared, it highlighted a core group - the 8 victims. That was evident across the grouped behaviours. The degrees of similarity varied, which was expected as different behaviours were being compared in each analysis.

Mike

[mic_ads]

Hi mic_ads,

Thanks very much for posting your results.

I'm sure everybody here would be interested to hear about your methodology.

Another member of this forum recently posted a link to an article written in 2005 which used profiling in an attempt to identify which of the Whitechapel Murders were Ripper victims.

http://forum.casebook.org/showthread.php?t=6422&page=29

(post 286)

It would be interesting to compare the results of this study with your own.

Do you intend to publish your study? It would be interesting to see your full analysis.

Thanks again.

Sally

I would like to publish my study. I have submitted it to the Journal of Investigative Psychology and Offender Profiling - it received a mixed review. There were positive remarks about the research and approach, but they questioned the topic as they thought JtR had been overdone. So I am happy with my methodology, my analysis and presentation of results. The study also branched into geographic profiling from the results but will probably be for another forum. There are other jounals so maybe one of these will consider it appropriate for their publication.

The study originally started out as an academic piece, but I have drafted another version, a book version that I have submitted to publishing agents. I have not received any offers yet, mind you, I have only contacted six agents.

In terms of the profiling article you mention, I have not read that one. I intentionally stayed away from profiling due to the mixed opinions on its value and results achieved. The process I chose was crime linkage which does share some characteristics with profiling but is a distinct and seperate process.

I hope that in answering other questions, I have outlined my methodology. However, if there are further questions, I'd be happy to answer them.

Mike

[bolo]

Hi mic_ads,

thanks for sharing your interesting work with us.

What are your sources for the data of the control group? As far as I know, the official papers (inquest reports, etc.) do not go into detail on the way the perpetrator gained control over or approached his victims. I may be wrong here, though.

Regards,

Boris