Start Omhoog

In reply to Child Sex Abuse: How Did We Get Here?

By BlueRibbon

Intro

(Q52) Was touched by an adult or older child sexually, or was asked to touch their genitals, or adult or older child exposed themselves 
N=574
Yes 23.7%
No 76.0%
Don't know 0.3%
Refused to answer 0.0% 

OK, so to begin with, the number '23.7%' relates to people who say they've had sexual contact as children. It does not mean they were abused, unless you think all sexual contact involving children and older people is abuse. So, already we know we are dealing with dishonest people who will twist their statistics to gain bigger headlines, and obtain more research grants from the government. 

The second thing I notice is that the N of Q52 (574) is precisely the N of Q51 (600) less the N that refused to answer Q51 (25.8). Coincidence? 

I doubt it, yet Q52 reports, strangely enough, that zero respondents refused to answer. It's a little speculative, but if you include those who refused to answer Q51, and divide 136 (the number of respondents who answered yes in Q52) by 600 (the actual total number of participants), the statistic is 22.6%, not 23.7%.

Entire article

I invite anyone with a background in statistics to critique my analysis, or carry it further.

BlueRibbon,

The survey reproduces a statistic that should be very suspect in our minds: 1 in 4 people are victims of sexual abuse. That's an old and debunked favorite. 

Abuse?

How did researchers and clinicians come to believe such a statistic, way back when? They asked questions that got as many "yes" answers as possible, and then interpreted that as if people were saying "yes" to being sexually abused. 

So, 

if you saw your father naked when you didn't want to, i.e. by accident, you were sexually abused. 
If someone at school grabbed your dick when you didn't want them to, you were sexually abused. 
If you came across pornography before the age of 18, even if you liked it, you were abused. 

By these standards, I was abused 100 times over. Yet I would never consider myself sexually abused, because none of the events were traumatic or troubled me. 

(When I was grabbed by a schoolmate in gym class, I was annoyed and grabbed him back. I guess that's the cycle of abuse in full swing. We both ended up being homosexual. Certain religious nuts would probably view that as a direct effect.) 

We'll have to wait to see just how this survey operated to understand how it is they arrived at 23.7%. Then their reasonable-ness will become apparent.

As for his arguments within the article posted, he's not reasonable at all. Take this for example:

The Philadelphians we surveyed almost without exception believe that children who witness physical and sexual violence between family members translate that imagery into their own behavior. They also believe that sexual imagery in the media is an overwhelming influence on adolescent behavior.

So, apparently the survey asked people what their opinion was on how various things influence children. Then Mr. Stinson makes an naive anthropologist's mistake of converting their opinion into fact. We're being led to believe, within the context of the article, that the opinions of those surveyed translate into how things really are. But these are just opinions, not how things really are. 

Why should the average Philidelphian understand how television impacts teenagers? There is an irony here: this survey is used to give a scientific gloss to the facts it 'discovers' (or creates), while at the same time generating those facts out of common opinion. It's as if the Catholic Church tried to disprove Galileo by surveying common serfs: "So does the earth revolve around the sun, yes or no? No? Well, I guess the earth doesn't revolve around the sun." This kind of thinking is not credible at all. 

The author might as well have said: "My mom told me that television is a bad influence on children, and I think so, too." But, of course, it sounds so much better to say, "I did a scientific survey, and everybody thinks X, so X is scientifically proven." 

Let's take another look:

"How we got to this unfortunate place isn't easy to pinpoint. We can't put our finger on just one reason that people sexually abuse children. 

Some genuinely believe that their activities with a child are expressions of love and caring. (They are neither.)

Some abuse children out of a need for power and control. They use their adult status to extract that feeling from a defenseless child.

Some are simply opportunists who satisfy their sexual needs on the easiest target - a child. 
Some suffer from aberrant emotional development. They mature physically, but not emotionally. When they desire emotional connection through sexual behavior, they turn to someone equally immature emotionally - a child.

Some offenders are sadists. Their sexual excitement is amplified by inflicting physical, sexual and emotional pain on a child. Some were themselves sexually abused as a child. 
And for some, their behavior just can't be explained. 

There is simply no single typology for a child sex offender."

Some are this, some are that, some are the other thing. Do these opinions come from his survey, or is he just gesticulating? His survey introduces breakthroughs and challenges and shocking facts, but these 'facts' do not seem to have anything to do with his survey. The survey, it seems, is just an opportunity for the author to riff on his favorite subject.

Notice two things. 

First, he introduces what is basically our position, that Boylove (or pedophilia generally) can be loving and caring. 
He then dismisses this and, secondly, goes on a burn tossing out the typical stereotypes: Boylovers are out for power and control, they are opportunists,
they are emotionally immature, they are sadists, they are enigmas. In the end: 

"There is simply no single typology for a child sex offender." 

We've gone from dismissing the idea that Boylove can be loving, to (re)installing the idea that Boylove is atrocious, and then depositing the reader at a "mystery" where nobody can know for sure. "There is no single typology" yet one of those types is certainly not loving.

Loving has been excluded, not so much by argument, buy by location. By running the idea of love right before hate, the author obliterates love without having to argue against it. Love translates into sadism purely by the flow of his ideas.

That's not reasonable, and it isn't nice. 

Let's look again:

Many people who look at the findings from the Peters survey wonder why child sexual abuse is such a large problem today.

Really? It was released only today. This sentence has more to do with creating the idea in the mind of the reader that 'everybody is talking about it,' and 'everybody is shocked' than giving us real information on how the survey is viewed. The person who wrote the article works for the institution who published the survey and has no interest in downplaying or critiquing it. Basically, the author is spinning the survey. Is that reasonable?

OK, enough about his article. 

I could go on, but let's look at the survey itself.
(< http://www.phmc.org/enews/Philly2006Final.pdf >)

Michael Stinson, who works at the Joseph J. Peters Institute, wants 23.7% to become the buzz number in talk about child sexual abuse in Philidelphia. It will be on the lips of every social worker for years to come. (Stinson hopes.) But where does this number come from? Should we trust it? Let's look at question numbers 51-53:

"My final questions are about sexual abuse in childhood. Again, you do not have to answer if you do not feel comfortable doing so.

51. When you were a child, were you sexually abused by an adult or older
child?
1 --Yes
2 --No
3 --Don't know [DON'T READ]
4 --Refused to answer [DON'T READ]-- go to 54
[AUTOCODE AS 1 IF Q51 [ABUSE] =1]

52. When you were a child, did an adult or older child ever touch you in
a sexual way, ask you to touch their genitals, or expose themselves to you?
1 --Yes
2 --No-- go to 54
3 --Don't know [DON'T READ]-- go to 54
4 --Refused to answer [DON'T READ]-- go to 54

53. Did you tell an adult about this while you were still a child?
1 --Yes
2 --No
3 --Don't know [DON'T READ]
4 --Refused to answer [DON'T READ] 

So far so good. Let's look at the responses. (Check the survey itself, page 19 for a better presentation of the data) 

"Incidence of the following occurrences as a child:

(Q51) Was sexually abused by an adult or older child 
N=600
Yes 16.8%
No 78.6%
Don't know 0.3%
Refused to answer 4.3%

(Q52) Was touched by an adult or older child sexually, or was asked to
touch their genitals, or adult or older child exposed themselves 
N=574
Yes 23.7%
No 76.0%
Don't know 0.3%
Refused to answer 0.0%

(Q53) Told an adult about this 
N=136
Yes 43.0%
No 54.7%
Don't know 2.3% 

OK, so to begin with, the number '23.7%' relates to people who say they've had sexual contact as children. It does not mean they were abused, unless you think all sexual contact involving children and older people is abuse. So, already we know we are dealing with dishonest people who will twist their statistics to gain bigger headlines, and obtain more research grants from the government.

The second thing I notice is that the N of Q52 (574) is precisely the N of Q51 (600) less the N that refused to answer Q51 (25.8). Coincidence? 
I doubt it, yet Q52 reports, strangely enough, that zero respondents refused to answer. It's a little speculative, but if you include those who refused to answer Q51, and divide 136 (the number of respondents who answered yes in Q52) by 600 (the actual total number of participants), the statistic is 22.6%, not 23.7%. 

The survey states the following: 

"...because older residents are frequently more likely to participate in telephone interviews than are younger residents, we weighted the results to reflect the actual age distribution in the survey areas. All data presented in this report is weighted unless otherwise indicated."

Here is the idea of weighting young respondents more than older respondents. This is common practice because traditional telephone surveys can only be conducted over landline telephones, and these devices are disproportionately distributed to older people. Younger people tend to have cel phones, which surveyers are not permitted to cold-call. Older people are also more inclined to respond out of social need, to combat loneliness. You need to weigh younger respondents more, because they are under-represented in the sample.

Look at the age spread of the people who responded to the survey:

18-24     44
25-34     86
25-44   104
45-54   126
55-64   120
65+      120
total    600

If one were to obtain a random sample in the US, it would look something more like (my quick and dirty calculations):

18-24   13.64% * 600 = 81.84 / 78.3% = 104.5
25-34   14.18% * 600 = 85.08 / 78.3% = 108.7
35-44   16.04% * 600 = 96.24 / 78.3% = 122.9
45-54   13.38% * 600 = 80.28 / 78.3% = 102.5
55-64     8.62% * 600 = 51.72 / 78.3% = 66.1
65+     12.44% * 600 = 74.64 / 78.3% = 95.3
total    78.3%

Consequently, the surveyors had to cook the numbers a bit (the weights from one district):

18-24    2.005702
25-34    1.328052
35-44    1.142445
45-54    0.745773
55-64    0.415168
65+      1.073298

What this means is that those in the 18-24 age bracket are counted roughly twice as much as those 65 and over.

This is relevant for us because of what seems to be a ideological shift in the generations. 

 The survey says:

"There are several significant differences among respondent groups, in
terms of the characteristics of respondents who are more likely to
disclose that they were sexually abused as children. Younger residents
are more likely to disclose abuse at the first definition, compared with
older residents; when asked the second, more specific question, however,
this difference disappears, which may indicate that younger residents
are generally more aware of and willing to discuss child sexual abuse. 

Is the difference really due to a shift in belief? Could there be another explanation as to why young people reported being abused more than older people?

The problem is not the statistical correction, but that the correction will multiply any other effects that already exist. If leading questions or the position of a question influences respondents, that influence is multiplied by two. 

Someone might interject that this is what the correction is supposed to do, correct for under-representation by over-representing a group. However, poor implementation results in bad data, and this data will be exaggerated by the correction. Statistical correction is correction only if the original data is good. But there are reasons to think the data isn't good, and therefore reasons to question the correction.

But are younger people disproportionately effected by any bias in the survey? There has to be a bias that disproportionately affects the young. I believe it may come down to a differential in attrition rates between the younger and older respondents. 

Attrition rates

We know that youths are the least likely to respond to a telephone survey. Let's say we have two populations, youth and age, both with 100 people involved. If youth has an attrition rate is 50%, but the older attrition rate is 15%, then we'll end up with 50 and 85 respondents respectively. 

If the attrition rate was independent of any factor being measured by the survey, it should not effect the results. However, we have reason to believe that there is a bias in the attrition rates. 

We know that those who stick with the survey feel child sexual abuse to be more serious than those who get bored and drop out. (See my comments below.) That is a survey wide bias. The specific bias involving youths is this: the elderly are more inclined to complete the surveys, whether they share this bias or not, therefore the proportion of abuse ideology believers to agnostics will be greater in the youth population than in the elderly. We might have 25 abuse victims among the 50 youths left over, and 25 in the group of 85 older folks. That's 50% to 30%.

It is not clear that youth report sexual abuse more readily than the elderly. Differential attrition rates may explain the effect. This bias is exaggerated by statistical correction. 

Let's look at the questions. 

We have two questions, 

one that asks whether the caller believes they were sexually abused, 
and another that just asks them about an inventory of sexual behaviors. 

Two questions are asked because: 

"We asked respondents if they had ever been victims of child sexual abuse
themselves. Estimating the incidence of child sexual abuse in
Philadelphia is problematic. Reporting is likely to vary depending on
how the question is asked, and given the sensitive nature of this
question, we might not expect full disclosure. Sexual abuse of any kind
is therefore likely to be underreported. 

In the 2006 study, we asked the question in two different ways: 

(1) we asked respondents whether they were sexually abused by an adult or older child when they were children; and 
(2) we asked respondents whether an adult or older child ever touched them in a sexual way, asked them to touch their genitals, or exposed themselves to respondents when they were children." 

So, like the survey's of old, this one specifically rejects the idea that a person's own perspective on an experience is of any value. Instead, there can be only one true and valid interpretation of sexual relations between a younger and older person, and that is abuse.

The first question tells us that some people interpret their experiences under the heading of abuse. But this in no way explores what 'abuse' means to the particular person responding. If 'abuse' is the only category you know of to understand your experience, because you have never heard of any other category, then it seems possible that you would term it that way, even if you feel ambiguous about it. 

It's important to note that the previous 40 questions in the survey presumed a wide definition of abuse. At one point the survey taker says: 

"The next issue is child sexual abuse. I know this is a sensitive topic,
but your responses will help develop programs to prevent child sexual
abuse. Again, your responses are totally anonymous and you don't have to
respond to any question that makes you uncomfortable." 

And then after requesting the respondents own view of what constitutes child sexual abuse, the survey taker says 

"In order to avoid misunderstandings, the definition of child sexual
abuse that we're using for this study is any sexual touching, sexual
penetration, or sexual attack of a child by an adult, adolescent, or 
older child. In addition, we would include as child sexual abuse some
non-touching sexual behaviors such as making sexually suggestive
comments to a child, exposing oneself, showing pornography to a child,
and photographing a child in sexual poses." 

The question arises at the end of the survey: is the respondent who says "yes" to Q51 saying yes to their own definition of sexual abuse, or to the definition imposed to "avoid misunderstandings?" 

Furthermore, to what extent has this line of interpretation influenced the respondent to frame his or her experiences as abusive? If you begin a survey of car quality by saying, "Just to avoid misunderstandings, we'll class Ford cars as unreliable and worthless," everybody's bullshit detector would go off.

A real survey about the prelevence of sexual abuse in a community would simply list behaviors without interpreting them in advance as abusive or non-abusive, and would probe to some extent the feelings a person has about those experiences. Then you could do some linking between experiences and subsequent results. But this section of the survey is not serious. In fact it seems to be tacked on at the end of what is, in reality, a survey of public perceptions about child sexual abuse and the organizations involved, the Joseph J. Peters Institute, and Stop It Now! Philadelphia.

The second question, Q52, is a more general and fair question, but becomes worthless in the interpretation. If I ask anyone whether they've had sex, the data I receive does not tell me whether it was abusive. But the abuse paradigm is where this survey is operating, and they permit themselves the liberty of deciding for others how their experiences affected them and what they mean. (There is no meaning outside of trauma and abuse.) 

The survey says: 

"The response rate (the proportion of eligible residents contacted who completed an interview) for the survey was 38.3%." 

The Office of Federal Statistical Policy and Standards has alerted federal agencies sponsoring data collections that the quality of data obtained may not be sufficient if the response rate falls below 75 percent. 

I'm sure response rates become sufficient again when they dip under 40%, right? I doubt it. When you get such a low response rate, we have to ask what it was that made these people respond and not others. Why do some people answer the questions, and some people say 'no thanks', or drop out of the survey?

The survey says: 

"The remaining potential respondents either refused to complete the survey or were unavailable during the data collection period." 

This means that it took 1579 attempts to complete 600 survey's, 200 for each district. In other words, almost 1000 people contacted either did not want to, or could not complete the survey. It seems only those who completed the entire survey were included in the final analysis.

We should be suspicious that the respondents who stuck with the survey until the end were precisely those individuals who already felt that child sexual abuse was a serious issue. One may be inclined to cut a survey short to make dinner or spank the kids if the subject is not terribly important (a commercial survey, for example). One sticks with a
survey that seems compelling or important. 

Who believes that child sexual abuse is important and compelling enough to stick with it through to the end? Certainly this population would sport a greater number of people who feel themselves to have been abused. In other words, compared to the wider population of Philidelphia, those who stuck with the survey until the end would be more likely to feel they were abused, and that explains why there is such a high number. 

Couple that with the differential attrition rates involving youth, and its exaggeration due to statistical correction. I bet you 100 $ that if the question had been asked near the beginning of the survey, the numbers would have been quite different. 

Here's a rundown on why we should question the 23.7%. 

First, there is a noticable N shift between Q51 and Q52. Why? 
 
Second, the data has been 'corrected' which may result in inflation of
bias in the data. 
 
Third, the response rate of the survey was 38.2%, much lower than rates
which are assumed to provide quality data, although it is not uncommon
in telephone surveys. 
 
Fourth, the ordering of the questions influences results. 
 
Fifth, the open-ended and leading questions and statements provide
opportunities for subtle influence. 
 
Finally, the survey interpreters are inherently disinterested in whether
a person perceives themselves as abused. Abuse is assumed and asserted
on neutral data. 

When an 'expert' at an 'institution' takes such questionable data, and marries it with rhetoric that operates in the realm of stereotype, he is not reasonable at all. The survey is a political tool, and the article is a piece of advertising for it, and the ideology of the institution which pays the author's rent.

Start Omhoog