Analysis of the Rippers MO

[Fleetwood Mac]

Here is an analysis of the Rippers MO by Journal of Investigative Psychology and Offender Profiling.

They examine 11 murders and conclude C5+1

In the cases of the five other Whitechapel murder victims, they could not be linked to Jack the Ripper through signature analysis. The killer(s) of Smith, Mylett, Coles, McKenzie, and the unidentified victim did not engage in the same pattern of escalating signature behaviors exhibited by Jack the Ripper, including the careful planning of the murders, picquerism, and the posing and mutilation of victims.

I've copied and pasted the above from the conclusion drawn in your linked article.

From what I can tell there isn't much in the way of empirical evidence supporting that conclusion. Feel free to correct me.

Here's an article from the American Academy of Psychiatry and the Law, Ritual and Signature in Serial Sexual Homicide:

Ritual and Signature in Serial Sexual Homicide | Journal of the American Academy of Psychiatry and the Law (jaapl.org)

The empirical data consists of 38 offenders and their 162 victims.

The article states:

The dearth of scientific studies of serial sexual homicide^13,14,38,39 is striking in comparison with the enormous interest this topic has received in film, print, and television media.

The article concludes:

Our research suggests that the crime scene actions of serial sexual murderers are fairly complex and varied. Specifically, the notion that offenders leave unique signatures at every scene is not supported by the data.

Perceived wisdom claims the WM murdered 5 or 6 women at the most. I'd say that's an error of judgement, as is the idea that the murderer or murderers of Emma Smith, for example, could not have been the same person or people who murdered Mary.

[JeffHamm]

In the cases of the five other Whitechapel murder victims, they could not be linked to Jack the Ripper through signature analysis. The killer(s) of Smith, Mylett, Coles, McKenzie, and the unidentified victim did not engage in the same pattern of escalating signature behaviors exhibited by Jack the Ripper, including the careful planning of the murders, picquerism, and the posing and mutilation of victims.

I've copied and pasted the above from the conclusion drawn in your linked article.

From what I can tell there isn't much in the way of empirical evidence supporting that conclusion. Feel free to correct me.

Here's an article from the American Academy of Psychiatry and the Law, Ritual and Signature in Serial Sexual Homicide:

Ritual and Signature in Serial Sexual Homicide | Journal of the American Academy of Psychiatry and the Law (jaapl.org)

The empirical data consists of 38 offenders and their 162 victims.

The article states:

The dearth of scientific studies of serial sexual homicide^13,14,38,39 is striking in comparison with the enormous interest this topic has received in film, print, and television media.

The article concludes:

Our research suggests that the crime scene actions of serial sexual murderers are fairly complex and varied. Specifically, the notion that offenders leave unique signatures at every scene is not supported by the data.

Perceived wisdom claims the WM murdered 5 or 6 women at the most. I'd say that's an error of judgement, as is the idea that the murderer or murderers of Emma Smith, for example, could not have been the same person or people who murdered Mary.

The above summary in the AAPL article is one of the reasons why I personally find "behavioural profiling" to be rather suspect. Behavioural profiling is of the "Mindhunter" type. There are some underlying ideas that certainly make it an interesting idea to pursue, but what has to be shown is that those ideas actually result in sufficiently strong patterns to be of practical utility.

The underlying big-picture idea is that our personalities and preferences govern or influence our choice of behaviour in situations where we can make a choice. So, if you know somebody's personality and their preferences, etc, you can predict their behaviour. Here's a simple example. If you know someone's personality is to be very shy and introverted, you can probably predict that they won't get up and perform karaoke in public. However, if you know they are also a huge music fan, you might also predict they are likely to sing along to music in other situations, and so forth.

The idea behind behavioural profiling is that one can go the other way as well, that if you know the behaviour you can work backwards to the personality/preferences of the person. So, if you see someone performing karaoke in public, they are probably unlikely to be a shy and introverted individual, etc. Obviously sometimes that inference will be wrong, but the idea is that it will be correct more often than not.

That is often how it gets presented (I think it's even presented in one of the Mindhunter episodes, and that general idea is included in John Douglas's books, and may even appear in some of the research literature).

The problem is, as Fleetwood's article highlights, the empirical work to determine just how well that reverse link "works" is minimal. What I mean is, let's say people divide 50/50 into "introverts" and "extroverts". And we decide that because Joe Bloggs is up there doing karaoke, they are more likely to be an extrovert. But to know if that's a reasonable prediction, we would have to know if the breakdown of karaoke performers is different from 50/50, with more of them being extroverted than introverted.

Now if I just ask the 38 people who performed karaoke at a local pub one night, and found that 34 of them were indeed extroverts (about 90%), then that would suggest that Karaoke performing provides one with a good (not perfect) indication of their introvert/extrovert status. But 38 people is a pretty small sample of the human population. Small samples tend to provide fairly poor estimates of the strength of the effect (meaning, a poor estimate of how good karaoke singing really is in terms of introvert/extrovert classification).

That small study would certainly justify doing a larger study, but then we might end up finding that, after interviewing say 1000 karaoke singers, from multiple cities, etc, it turns out only 55% of them are extroverts, and our initial sample greatly overestimated how well it predicts that personality trait, and it turns out there are lots of reasons why introverts end up on stage.

Now, while there is still some information about introvert/extrovert to be drawn from karaoke singing, in the end, that information doesn't really let you draw a strong conclusion beyond knowing that the population is 50/50 to begin with (you're just going to tilt towards a 55/45 type thing, which is hardly enough to divert an investigation away from both possibilities).

And the lack of such large scale studies, and the complexity of crime scene behaviors, makes behavioural profiling of this sort "unsubstantiated" in terms of its efficacy (by this sort I mean of the type where the profiler "intuits" the characteristics of the offender).

There are some approaches which are more empirically based, and they have shown some promise (police forces that have been trained in these more empirically based forms of profiling (i.e. you enter details of the offence into a computer, which does calculations based upon statistical analyses of a larger sample, and the profile is generated by those statistical regularities and gives a sort of "strength" to the output) have shown higher case resolution rates than forces that do not use these methods (and the comparison was made using two areas with similar starting case resolution rates - so the introduction of the empirical profiling routines resulted in an increase in case resolutions while the other force remained at about the same level). So it's not all bad, but a lot of it is a bit "magical", and it is the magical forms that get the most press and movie times.

The spatial analysis (geographical profiling) is also "overhyped" in the press and movies. But, it does provide some information, which is useful to know, provided it is understood properly. It won't solve a case (that's not what it does; that's what detectives do), but it can provide some useful information to a detective if they understand what that information is and how best to use it. And what is best is not how it is depicted in the movies or press.

Anyway, human behaviour of any sort, including criminal, is extremely complex. And while there are regularities between personality and behaviour, those regularities are contained within such a complicated network of influences that reverse engineering personality by starting at observed behaviours is not likely to change the odds anywhere near as dramatically as it gets shown.

- Jeff

[Aethelwulf]

The above summary in the AAPL article is one of the reasons why I personally find "behavioural profiling" to be rather suspect. Behavioural profiling is of the "Mindhunter" type. There are some underlying ideas that certainly make it an interesting idea to pursue, but what has to be shown is that those ideas actually result in sufficiently strong patterns to be of practical utility.

The underlying big-picture idea is that our personalities and preferences govern or influence our choice of behaviour in situations where we can make a choice. So, if you know somebody's personality and their preferences, etc, you can predict their behaviour. Here's a simple example. If you know someone's personality is to be very shy and introverted, you can probably predict that they won't get up and perform karaoke in public. However, if you know they are also a huge music fan, you might also predict they are likely to sing along to music in other situations, and so forth.

The idea behind behavioural profiling is that one can go the other way as well, that if you know the behaviour you can work backwards to the personality/preferences of the person. So, if you see someone performing karaoke in public, they are probably unlikely to be a shy and introverted individual, etc. Obviously sometimes that inference will be wrong, but the idea is that it will be correct more often than not.

That is often how it gets presented (I think it's even presented in one of the Mindhunter episodes, and that general idea is included in John Douglas's books, and may even appear in some of the research literature).

The problem is, as Fleetwood's article highlights, the empirical work to determine just how well that reverse link "works" is minimal. What I mean is, let's say people divide 50/50 into "introverts" and "extroverts". And we decide that because Joe Bloggs is up there doing karaoke, they are more likely to be an extrovert. But to know if that's a reasonable prediction, we would have to know if the breakdown of karaoke performers is different from 50/50, with more of them being extroverted than introverted.

Now if I just ask the 38 people who performed karaoke at a local pub one night, and found that 34 of them were indeed extroverts (about 90%), then that would suggest that Karaoke performing provides one with a good (not perfect) indication of their introvert/extrovert status. But 38 people is a pretty small sample of the human population. Small samples tend to provide fairly poor estimates of the strength of the effect (meaning, a poor estimate of how good karaoke singing really is in terms of introvert/extrovert classification).

That small study would certainly justify doing a larger study, but then we might end up finding that, after interviewing say 1000 karaoke singers, from multiple cities, etc, it turns out only 55% of them are extroverts, and our initial sample greatly overestimated how well it predicts that personality trait, and it turns out there are lots of reasons why introverts end up on stage.

Now, while there is still some information about introvert/extrovert to be drawn from karaoke singing, in the end, that information doesn't really let you draw a strong conclusion beyond knowing that the population is 50/50 to begin with (you're just going to tilt towards a 55/45 type thing, which is hardly enough to divert an investigation away from both possibilities).

And the lack of such large scale studies, and the complexity of crime scene behaviors, makes behavioural profiling of this sort "unsubstantiated" in terms of its efficacy (by this sort I mean of the type where the profiler "intuits" the characteristics of the offender).

There are some approaches which are more empirically based, and they have shown some promise (police forces that have been trained in these more empirically based forms of profiling (i.e. you enter details of the offence into a computer, which does calculations based upon statistical analyses of a larger sample, and the profile is generated by those statistical regularities and gives a sort of "strength" to the output) have shown higher case resolution rates than forces that do not use these methods (and the comparison was made using two areas with similar starting case resolution rates - so the introduction of the empirical profiling routines resulted in an increase in case resolutions while the other force remained at about the same level). So it's not all bad, but a lot of it is a bit "magical", and it is the magical forms that get the most press and movie times.

The spatial analysis (geographical profiling) is also "overhyped" in the press and movies. But, it does provide some information, which is useful to know, provided it is understood properly. It won't solve a case (that's not what it does; that's what detectives do), but it can provide some useful information to a detective if they understand what that information is and how best to use it. And what is best is not how it is depicted in the movies or press.

Anyway, human behaviour of any sort, including criminal, is extremely complex. And while there are regularities between personality and behaviour, those regularities are contained within such a complicated network of influences that reverse engineering personality by starting at observed behaviours is not likely to change the odds anywhere near as dramatically as it gets shown.

- Jeff

Good post Jeff. Out of interest, on the geoprofile, are there good examples of when it does 'work'? For example, if you put all of Peter Sutcliffe's murders and attacks (or some other offender) into a model, would it give a plausible answer? How often would you expect a decent answer?

[Aethelwulf]

Good post Jeff. Out of interest, on the geoprofile, are there good examples of when it does 'work'? For example, if you put all of Peter Sutcliffe's murders and attacks (or some other offender) into a model, would it give a plausible answer? How often would you expect a decent answer?

In answer to my own question I found this article: https://chalkdustmagazine.com/featur...laria-mammals/ which mentions the YR and JtR. The image below is from the same site - black dots are Sutcliffe's body sites and red squares his two addresses during the murders. Seems a decent enough match apart from the two near Manchester.

.

[JeffHamm]

In answer to my own question I found this article: https://chalkdustmagazine.com/featur...laria-mammals/ which mentions the YR and JtR. The image below is from the same site - black dots are Sutcliffe's body sites and red squares his two addresses during the murders. Seems a decent enough match apart from the two near Manchester.

Hi Aethelwulf,

Don't forget, spatial analysis isn't specifically about locating the offender's residence (though it is often described that way in the press and movies). It's about locating "anchor points", which are locations that the offender is likely to have a strong association with through their every day lives. Obviously one's residence is an anchor point for non-transients, but an offender's crimes may not necessarily be anchored around their residence. It could be more associated with the offender's place of work, for example (and Sutcliff's job as a lorry driver took him to Manchester regularly). Dennis Radar's crimes were more strongly anchored to his places of work, and he didn't commit an offence near his home until he stopped working at ADT Securities and started working as a compliance officer for Park City (where he lived). Collin Ireland's crimes (in London) and Bruce McArthur's crimes (in Toronto) produce search areas that focuses in on the area that includes the bars each of them were known to frequent and find victims.

The example maps you will find will generally be impressive ones, where the offender's residence ends up in the top priority zone (as per your example). If you draw the smallest circle you can that includes all the offense locations (and then expand it by an amount related to how many offenses you have; basically, if you have only a few you make the area a lot bigger, but if you've got 10 a fair number you only make it a little bigger). The area of that circle defines our "territory size". You divide that area by 40 to get 40 zones, each zone is equal to 2.5% of the territory size. You then do the spatial analysis, and each location gets a score, and you group all the locations with the highest scores until they total 2.5 of the territory size, and that is "zone 1". Get the next highest set of locations for zone 2, and so forth.

If the series is generated randomly within some area, and the main anchor point of the offender is also random (but from the same area as the random offenses), then you have to search about 21 or 22 zones to get a 50/50 chance of detection (it's not 20/40, because of course, sometimes the smallest circle will underestimate just how big of an area you have to search, so even if you searched all 40 you still wouldn't locate them).

But, if you create the zones using the spatial analysis routines, you only have to search 4 to 5 zones to find 50% of the offenders (so 10-12.5% of the estimated search area, rather than 52.5-55%). About 25% of the time (though the sample I have on this is a bit small) you only have to search zone 1. To locate 75% of the offenders, you have to search out to zone 20 (so 50% of the search area). The maps and results you will find tend to come from the 25% of the cases where the anchor point was in zone 1; you're seeing a biased sample.

As I say, the routines do help narrow things down, but think about the size of Peter Sutcliff's "territory"! It's something like 45 miles from Leeds to Manchester, so that's a total search area of 1590 square miles. Even getting him in zone 1 is still a search area of almost 40 square miles. so out to zone 4 would be 160 square miles (zone 5 makes it 200 square miles). It's not pinning a sign on his door. But, narrowing a search area down from almost 1600 to even 200 square miles is useful information.

It still requires a lot of hard work to find real evidence. These just produce informed suggestions as to where to look.

- Jeff

[Fleetwood Mac]

In answer to my own question I found this article: https://chalkdustmagazine.com/featur...laria-mammals/ which mentions the YR and JtR. The image below is from the same site - black dots are Sutcliffe's body sites and red squares his two addresses during the murders. Seems a decent enough match apart from the two near Manchester.

.

Add in victims who survived his attacks, and it's a bit of an eye-opener. Two of his early attacks were in Keighley and Silsden which are North of where he lived. So, he went East, West, North and South. 'Pretty much lived central to where he hunted victims.

In the event you're looking to draw a parallel, other traits may prove instructive:

1) Sutcliffe's first four victims survived his attacks.

2) All of the known 21 victims were attacked when it was dark.

[Mark J D]

... spatial analysis isn't specifically about locating the offender's residence (though it is often described that way in the press and movies). It's about locating "anchor points", which are locations that the offender is likely to have a strong association with through their every day lives. Obviously one's residence is an anchor point for non-transients, but an offender's crimes may not necessarily be anchored around their residence. It could be more associated with the offender's place of work...

... or his recent former address and route to work, or his mother's and daughter's current address...




I really must send these graphics to Gareth Norris.

M.

[mpriestnall]

Sorry, I had to 'like' the above comments. Mea culpa.

[MrBarnett]

... or his recent former address and route to work, or his mother's and daughter's current address...




I really must send these graphics to Gareth Norris.

M.

Eddowes looks like an outlier, but she needn’t be if you factor in Pickford’s operating out of Haydon Square prior to the opening of Broad Street goods station. I doubt CAL started work at 18/19. Perhaps he started working for Pickfords at Haydon Square, just to the east of Mitre Square and south of Goulston Street, several years earlier.

[MrBarnett]

‘Anchor points’ is an interesting concept. For Lechmere, his primary anchor point would presumably have been pretty much where the Pinchin Street torso was discovered - identified on Mark’s graphic as his mother’s address.

[JeffHamm]

... or his recent former address and route to work, or his mother's and daughter's current address...




I really must send these graphics to Gareth Norris.

M.

Yes, as potential anchor points, those would be the sorts of things you would look for in the high priority zones. But Lechmere's recent and former address are not in the high priority zone, nor is his mother's and daughter's current address. His route to work does pass through the high priority zone, but generally one expects a location that could be called a "destination" or "stopping point". If, however, one could show that Cross/Lechmere was known to stop at the 10 Bells (for example, as it's in the high priority zone), then that would be interesting. But just passing through is less compelling (but not nothing, of course, but it is a bit down on the list of things one would flag because many people pass through areas).

In short, the spatial analysis does not fit with what we know of Cross/Lechmere. But that doesn't rule him out (nor would fitting with it mean it has to be him) anymore than not being the spouse of a victim precludes someone of being the offender (and just being their spouse doesn't make you the offender either).

I know a lot of documentaries have presented the above as if it is valid, but it's not based upon any empirical data that I'm aware of. The routines that are empirically based (meaning evidence based), produces spatial "profiles" (I personally dislike that term) that do not raise the eyebrows when one looks at Cross/Lechmere.

Now, Robert Paul, his place of work does fall in the high interest zone. Hmmmm, .....

- Jeff

[MrBarnett]

Yes, as potential anchor points, those would be the sorts of things you would look for in the high priority zones. But Lechmere's recent and former address are not in the high priority zone, nor is his mother's and daughter's current address. His route to work does pass through the high priority zone, but generally one expects a location that could be called a "destination" or "stopping point". If, however, one could show that Cross/Lechmere was known to stop at the 10 Bells (for example, as it's in the high priority zone), then that would be interesting. But just passing through is less compelling (but not nothing, of course, but it is a bit down on the list of things one would flag because many people pass through areas).

In short, the spatial analysis does not fit with what we know of Cross/Lechmere. But that doesn't rule him out (nor would fitting with it mean it has to be him) anymore than not being the spouse of a victim precludes someone of being the offender (and just being their spouse doesn't make you the offender either).

I know a lot of documentaries have presented the above as if it is valid, but it's not based upon any empirical data that I'm aware of. The routines that are empirically based (meaning evidence based), produces spatial "profiles" (I personally dislike that term) that do not raise the eyebrows when one looks at Cross/Lechmere.

Now, Robert Paul, his place of work does fall in the high interest zone. Hmmmm, .....

- Jeff

So, the area around Pinchin Street and Berner Street isn’t a high interest zone? Why’s that? Hmmmm, …

[MrBarnett]

Is that your graphic, Mark?

If so, could you add Pinchin Street?

[MrBarnett]

Where would someone living and/or working in the City go to find doss house ‘unfortunates’? Threadneedle Street or Commercial Street?

Don’t the locations of the murders tell us as much - if not more - about the victims than the killer?

[Admin]

This is not a Lechmere thread, it is not in the Cross/ Lechmere board, and it will not turn into a Cross/ Lechmere thread. Anyone attempting to discuss Lechmere, will confine themselves to the forum created for that discussion and not hijack every non-related thread with suspect bias as outlined in the rules.

If you want to discuss Lechmere, there is a board for that. Do not introduce him to every off-topic thread.