(This is the second half of a longer essay. Unless you are already familiar with the methodology of Cantor’s paedophilia research, much of this essay won’t make sense if you haven’t read the first part )
A doctor who specialises in smoking-related lung cancer notices from his patients’ health records that during their childhoods (defined as the period from birth to the end of secondary education) they consulted their GPs for bronchopulmonary problems at a significantly higher rate than average, many of the consultations taking place during their infancy.
The doctor is, of course, aware of the link between smoking and lung cancer – but these patients’ increased rate of childhood bronchopulmonary problems presumably occurred at a time before they had started smoking.
The doctor concludes from this evidence that a propensity to smoke (and consequently develop lung cancer) must stem from genetic or gestational predisposition to smoke. How else explain the fact that those adults who go on to develop lung cancer had a greater incidence of bronchopulmonary problems before they started smoking?
The doctor is of course wrong. He has not properly grasped the complexity behind the rubric ‘rate of childhood bronchopulmonary problems’. There are many factors other than ‘genetic or gestational predisposition’ that could contribute to the perceived correlation (the most notable being, of course, having parents who smoke).
Moreover the doctor has used a definition of ‘childhood’ that is inappropriate to the phenomenon he’s dealing with. By defining childhood as ‘the period between birth to the end of secondary education’ our doctor has introduced a crucial overlap between the notionally non-smoking ‘children’ and his adult group of smokers. The average age for taking up smoking is variously reported as being between 13 and 16. This means that a significant number of the ‘childhood’ referrals for bronchopulmonary problems of his cancer patients occurred at a time when they were already active smokers.
Let’s quickly revisit the problem with Cantor’s Grade Failure research as outlined in the first part of this essay.
Cantor found that in samples of sexual offenders IQ and brain-functioning diminished as the Age of Attraction got younger. He also found the same pattern with self-reported grade failures that had occurred during their compulsory education.
On the basis that –
“grade failure and identification as having special education needs during primary or secondary education [are] indicators of early developmental pathology”
Cantor concluded that the low IQ and other brain-functioning deficiencies must have been either genetic or gestational.
I hope it’s reasonably clear what (to use the scientific terminology) a crock of shit this is.
Try making the following substitutions to the Lung-Cancer Doctor scenario:
– ‘sexual offending against children’ ↔’having lung cancer’,
– ‘grade failure and other signs of childhood brain deficiency’ ↔ ‘childhood bronchopulmonary problems’,
– ‘being a paedophile’ ↔ ‘smoking’.
The seductive label of ‘childhood’ has led Cantor into assuming that his ‘grade failures’ are uncontaminated by the very factors that would cause functional deficiencies in adult paedophile offenders, most notably Stigma.
Cantor fails to take into account the likelihood that over 50% of paedophiles in his sample knew they were paedophiles whilst still in secondary education, and a significant proportion of these may have known they were paedophiles before the age of 14 (the Visions of Alice survey suggests 20%).
Secondly, because paedophiles become subject to stigma the earliest of the four groups in the study (teleiophile non-offenders, teleiophile sex offenders, hebephile offenders and paedophile offenders), and also endure the most intense stigma, that stigma’s effect on their developing brain, on their social development and on their education will be the most marked.
Thirdly the reporting biases associated with the four groups are such that grade failure which occurred because of paedophile stigma will be best remembered and reported.
All this means that grade failure etc can not be taken as a reliable indicator of “early developmental pathology”, since the stigma associated with paedophilia could be causing, some, most or all of the grade failure reported by the paedophile offender sample. If Dr Cantor had focused exclusively on grade failures that occurred during primary education, when children are much less likely to be identified as ‘paedophiles’ (and thus suffer the associated stigma and stigma-damage), his findings would be more credible.
Confirmation of the role of Stigma in my critique of Cantor’s research has come from a somewhat unexpected quarter:
”One main ramification of having a sexual interest in children, even if one does not act on it, is that these individuals are likely to face intense stigma due to their pedophilic interests and because of this stigma-related stress, be at increased risk of negative mental health and interpersonal outcomes. Stigma for non-offending pedophiles is an important area to research, because the negative outcomes associated with stigma-related stress (e.g., social and interpersonal problems; emotional dysregulation; limited life opportunities) are also theorized to be central risk factors for the initiation of sexual offending.”
(James Cantor in ‘Non-Offending Pedophiles‘ – 2016)
Last year Dr Cantor published a review of recent studies into non-offending pedophiles which appears to mark a break in Cantor’s thinking.
For the first time Cantor acknowledges the existence of non-offending paedophiles. This is quite radical – all his previous work and thinking has focused on offending paedophiles. Indeed, he has, either advertently or inadvertently, eliminated non-offending paedophiles from his samples:
“Patients were also excluded if there was no available information regarding their sexual history or inclinations beyond their own self-report at the time they presented at the Laboratory.” (2)
“[some] patients who came to the laboratory were not included here, such as […] those for whom there was no information available regarding their sexual behavior or interests beyond their self-report. (3)
“The […] sample excluded [those] for whom there was no sexual behavior or interests information available other than self-report.” (4)
Another ‘First’ for this paper is that it contains Cantor’s first use of the word ‘Stigma’ (or any equivalent word or phrase) in connection with paedophilia.
Indeed he veritably binges on the word – making up for lost time maybe? – using it a grand total of 53 times, making it the 20th most common word in the paper, coming in just ahead of ‘sex’ and ‘is’. (he however used the word several times in connection with homosexuality in his 2012 paper “Is Homosexuality a Paraphilia?”)
That’s the good news.
The bad news is that one gets the strong impression reading this paper that Cantor is a man confronting something big that, if viewed in its entirety, would oblige him to wholly re-evaluate not only his views on paedophilia, but also over a decade’s-worth of published, much-cited research.
So in order to best preserve as much he can, he acts like a pachydermophobe locked into a small room with our tusked friend Jumbo. He can only contemplate Jumbo through fearful fingers held over his eyes and, like the blind men and the elephant of legend, notices only disparate, contradictory things.
As the quote at the top of this section indicates, Cantor acknowledges that stigma can be immensely damaging both mentally and socially. But he extends this acknowledgment no further than to ‘non-offending paedophiles’.
He writes as if ‘offending pedophiles’ and ‘non-offending pedophiles’ were two completely separate species, with no permeability between them.
“Non-offending pedophiles are a unique population of individuals who experience sexual interest in children, but despite common misperceptions, have neither had sexual contact with a child nor have accessed illegal child sexual exploitation material.”
Every ‘offending pedophile’ was once a ‘non-offending’ pedophile, of course. And every non-offending paedophile has somewhere in him or her a strong desire to become an ‘offending paedophile’.
Cantor unwittingly acknowledges this fact since throughout the paper he maintains a close focus on the risk factors that can turn the non-offender into the offender. He also asserts that the principal risk factor is Stigma (I would disagree – I think the principal risk factor is ‘temptation’, but acknowledging this has even more subversive implications than acknowledging Stigma as a risk factor ).
But Cantor also resists this fact. Nowhere in this paper does give any indication of applying the concept of Stigma to offending paedophiles – either after they have committed an offense or before (i.e. when they were still non-offending paedophiles).
It is as if at the moment of their first offense, all the stigma the individual previously experienced (stigma which Cantor acknowledges as being the principal contributing factor to them offending) loses its status as ‘stigma’ and becomes, what? Justice? Pre-emptive Justice?
This is the kind of profoundly sloppy thinking that can only remain erect if supported by a rigid, strong and unshakeable scaffolding of prejudice.
This paper is too little, too late, too narrowly applied.
But there is another significant problem with Cantor’s research…
Berkson’s Bias can occur when two attributes that are independent of one another (i.e. un-correlated) in the general population appear to be negatively correlated when measured in a subsample that has been constituted through a selection process based on those two attributes.
If the above paragraph makes as much sense to you as it would have done to me when I first started trying to get to grips with Berkson’s Bias then maybe a simple example will be of use:
A university’s admissions criteria asks that their students either have exceptional academic achievement or exceptional musical talents.
A survey of this university’s student population found that the attributes ‘academic achievement’ and ‘musical talent’ appeared to be negatively correlated, despite them being uncorrelated (or even positively correlated) in the general population.
This is because any candidates accepted with poor exam results could only have been accepted because they were musically talented, and students who had good exam results had no need to be musically talented.
Let’s map Cantor’s methodology onto the above example.
The first part is simple: replace the subsample ‘university student population’ with ‘paedophile offenders’.
But what are the selection criteria for inclusion in the subgroup ‘paedophile offenders’? At first there appears to be only a single criteria: ‘being convicted of for a sexual crime with a child’.
“But, surely”, I hear you protest “Berkson’s Bias requires there to be two independent variables!!”
Well, hush your protests, good reader, and join me in elucidating the variables that lie behind ‘being convicted of for a sexual crime with a child’.
Given a particular level of temptation (let’s imagine a dark-eyed, golden-skinned girl of 7 demanding that you to get into the bath with her and give her a good soaping…) two broad variables will determine whether a paedophile is likely to comply with that child’s request (and thus ‘offend’) or not:
– Propensity to Offend (PO). This measures an individual’s readiness to yield to temptation or engage in illegal activities. High PO correlates with poor impulse control, brain-deficiencies, poor long and short-term memory, and criminality, and has an inverse correlation with IQ and academic success.
[edit 12/06/17 – the ‘Propensity to Offend’ variable is not just applicable to the ‘sexual offender’. It can equally be applied to all ‘offenses’ – tax evasion, murder, stealing paper-clips from the office, illegal parking, burglary…]
– something I’ll describe as ‘Intensity of Paedophilia’ (IP). This measures just how much of a raving paedo you are. It increases in proportion to the youthfulness of the objects of attraction, the exclusivity of the desire, and the intensity of the desire. Someone with a IP of zero is a die-hard geruntophile, who hates children and doesn’t even find them ‘cute’. Someone with a maximum IP might be an exclusive nepiophile who’s overdosed on Viagra.
If we assume that PO and IP are independent variables in the general population (an assumption that Cantor would of course reject…) we get the following graph in which each dot represents an individual in the general population of paedophiles (indeed I suspect that the graph would be valid for the general population of all people, not just paedophiles).
Let’s (crudely) divide this graph up into four quadrants.
The paedophiles in the top-right quadrant with a high PO and a high IP are the most likely to commit a paedophile offense and end up in Cantor’s sample.
The paedophiles in the bottom-left quadrant with a low PO and low IP are the least likely to offend and end up in Cantor’s sample.
The remaining two quadrants (high PO/low IP and low PO/High IP) lie somewhere between the two in terms of risk of offending.
Factors that discourage offending (fear of being caught, of stigma, of damaging the child &c) will push the ‘offending’ area up towards the top right corner of the graph, giving the following negative correlation between the two independent variables.
However Cantor’s analysis of his sample of paedophile offenders does not directly measure ‘propensity to offend’ or ‘intensity of paedophilia’, but instead uses proxy measures: ‘IQ’ for the former and ‘age of attraction’ for the latter. This is not a problem. But since ‘propensity to offend’ is negatively correlated to IQ, as is ‘Age of Attraction’ to IP, the scale of values in the following table are inverted.
So, despite IQ and Age of Attraction being independent and randomly distributed in the general population, the biases in the selection process by which the subgroup of ‘paedophile sex offenders’ is constituted means that a strong negative correlation appears between intelligence and attraction to younger children.
The following table is what Cantor’s research proposes as the distribution of IQ and Age of Attraction in the general population of paedophiles (not just of ‘offenders’).
Who says good things don’t happen to good people?
Out of the blue, as I was working on this essay, I was alerted to research that has come fresh off the press with the title ‘Gray matter anomalies in pedophiles with and without a history of child sexual offending‘ (B Schiffer & al) and which seems to confirm the validity of the ‘Propensity to Offend’ variable I introduce in the ‘Berkson’s Bias’ section above.
I’m not going to go into detail on this paper, partly because it is heavy going and partly because, though its findings support my arguments here, it takes the same stigmatising approach to paedophiles as Cantor does. Nevertheless – the paper concludes that:
“The present study substantiates the idea that CSO [Child Sex Offending] in pedophilia rather than pedophilia alone is associated with changes in GM [Grey Matter] integrity, particularly in the right temporal pole”
This implies that ‘perturbations of neural development’ do not cause paedophilia per se, but rather contribute to a propensity to offend. In other words, that ‘IP’ and ‘PO’ are indeed independent variables.
It may be the case that more intelligent paedophiles are not just less likely to be caught, but are also less likely to offend. The greater impulse control associated with high achievement in education serves to reduce the likelihood of committing crimes of impulse and temptation. This would tend to exclude paedophiles with higher IQ from Cantor’s samples of paedophile offenders.
A letter to James Cantor’s from his research supervisor…
Rummaging round amidst the bags of Cheetos, cans of beer and half-eaten Big Macs in Dr Cantor’s trashcan the other night I came across the following letter. I think it may be of interest to the readers of this blog…
Dear little Jamie-Pooh (I’m sure you won’t mind me calling you that, especially after what happened the other night…)
re your research:
Good show! Ten out of ten for effort!
But I’m afraid I’ll have to mark you down for letting your prejudices run away with you.
Why not look on your research on pedophilia to date as a ‘false start’ from which you can draw valuable lessons? You know, boy, sometimes you have to walk a short path heading East to catch the plane heading West. If you learn from your mistakes I’m sure that you will one day do some really top quality research on pedophiles.
Here are some bullet points for you to think about:
– you know, that ‘Sisyphus Mann’ fellow is quite right: you really should have limited your grade failure data exclusively to primary school grade failures. Surprised that you didn’t spot that one.
– going forward, you should work on accessing samples of pedophiles from the general population. I know that this isn’t easy, but ‘per ardua ad astram’…Using offenders, especially for a condition as stigmatized as pedophilia really sucks – just too many confounding factors – elucidating reliable facts from such a sample is like searching for a stick-insect in a big pile of sticks.
– focus on the factors that make for ‘offending’, rather those that make for ‘pedophilia’. You’ve made a cracking fresh start with the ‘non-offending pedophiles’ study, boy! Now go off and do some research! But don’t forget that every ‘offending pedophile’ was once a ‘non-offending pedophile’.
– if you have difficulty overcoming feelings of contempt and hatred towards those you are studying, try in future to include a pedophile in your team. They will bring a different perspective and may spot errors your prejudice has blinded you to. Also why not try actually talking to one of these perverts? I don’t mean with a clip-board in your hand and your recorder whirring away – but down the bar/pub/café. Or on a park bench by the children’s playground (ha ha ha)…You might be surprised – these pedo fellows may not all be slathering monsters, a straw’s-weight away from jumping the nearest child.
I heard somewhere that Lewis Carroll was a pedo. Maybe you should take a biography of his on vacation. And why not the Alice books? Heard he was a decent chap, despite his lips going a bit dry at the sight of a pink bicycle…
– fools rush in where angels fear to tread &c, but have you given any thought as to what you might find if you treated ‘consensuality’ as a variable of pedophilic offending?
The issue of consent effectively distinguishes your ‘teleiophilic sexual offender’ from the ‘teleiophilic non-offenders’. It sorts the rapist from the normal caring, decent, husband/boyfriend.
Between you and me and the lamp-shade (by which I mean: don’t tell the Dean I’ve said this – this is strictly off-record. He’d have my guts. Especially after the Fresher’s Ball debacle. In fact do us both a favour and eat this sodding letter once you’ve done with it – some extra fibre will do you some good…) if it makes sense to place a man who rapes women in his cellar in a different category to a loving husband who woos his wife with flowers, choccies and Barry White – then maybe you should be making a similar distinction for these pedophile johnnies.
I mean, it should be obvious that a man raping a struggling, crying three-year-old boy is an entirely different kettle of crisps to a man who treats some brat nicely, helps them with their homework, teaches them to ride a bike etc etc and whom the child maybe really quite likes and wants to be intimate with (I have rather fond memories of one of my masters at boarding school)…
‘Effective consent’ may be a valid variable in this paedophilia business. I know that the implications of acknowledging this might be too offensive to the sensibilities of some delicate souls – but, by god Jamie! you’ve got the letters ‘D’ and ‘R’ before your name! You can’t be squeamish when you’re a doctor! What if someone came into your surgery with gangrene? Or industrial-size haemorrhoids? Or an eye-ball hanging down their cheek!?
Last bit of advice: watch out for that Berkson; he’s a tricky johnny!
all the best
PS don’t forget that fiver you owe me