On child abuse and drawing conclusions from data.

If you’re looking for a good strategy for having a bad weekend, I’ve got one: you could go to your local library and borrow Ross Cheit’s book The Witch-Hunt Narrative.

Cheit ruined my weekend.  And his work is out there, ready to ruin yours too!

Not that his book isn’t good.  It is.  I’d write that I’m glad he wrote it, except that I wish he hadn’t needed to.  But he did.  And, sadly, because I am working on a project that involves many sad occurrences, I needed to read it.

So, his book?  He (and a huge team of helper monkeys) went to considerable trouble to investigate several large child abuse trials that were described as “witch hunts” in the media.  He makes a personal statement revealing his bias going into the work right from the beginning, and then presents the evidence he found as clearly and exhaustively as possible.  I found his reasoning to be extremely cogent, his writing lucid, and his conclusions persuasive.  I think he makes a compelling case that many children were failed by our criminal justice system – the children who were deemed dishonest, the children who were denied compensation for harm wrought upon them, and those who were later harmed by perpetrators who perhaps could have been stopped in a less hostile legal environment.

For me, reading this book, one of the main things I took away was that the same individuals were often serving as expert witnesses for the defense, testifying against the children’s claims.  And in his research, Cheit identified some highly suspicious potential motives for these people – several seem to have believed sexual contact between adults and children to be reasonable.  But one of the major adversaries discussed in the book is Dr. Maggie Bruck, whom Chiet at one point accuses with the statement: “Bruck’s position in these cases is not based on science alone.”

That sounded like the kind of statement I could write an essay about.  For one thing, my own project concerns academic science, the difference between how objective it is and how objective the population at large often believes it to be.  And I was interested in why Bruck would make the claims quoted in the book.  For instance, there is this compelling passage from Chiet:

“The remarkable blind spot in Bruck’s position is how it failed to apply her vast knowledge and concern about coercion to the most likely and powerful source in the case: the man who used a baseball bat to make his points.  In Bruck’s view, the interview of July 14, 1994, was completely tainted because in the preceding months David had had contact only with adults who believed Ardolino was guilty.  Bruck never explains why this is so objectionable, while an interview conducted immediately after years under the control of the defendant doesn’t rate any similar concern–even when the allegation, backed up with medical evidence, is that the defendant slowly beat his brother to death.  Apparently, even a witness to such horror is suspect in Bruck’s view if he does not disclose the abuse immediately, on the first interview, and all in one piece.

“There is, of course, a hidden value choice in the position–one that deems the coercive effects of the police interview as a far greater concern than the coercive effects of living with Robert Ardolino for years.  In that respect, Bruck’s position is de facto pro-defense.  This is the only explanation that makes this position consistent with the Rouse case, where Bruck readily accepted recantations that occurred only after the children moved back into a residence where the adults did not believe the original claims–the mirror image of the reason she used to reject David Ardolino’s testimony.  So Bruck does not see the Rouse adults as tainting the recantation in that case, yet she sees David’s foster mother as tainting his testimony, which was, in effect, a recantation of his earlier denials.  These positions are logically incompatible, except to the extent that they both favor defendants.  This “pro-defense orientation” is precisely how Ceci described the design of his research studies when testifying in the Foeller case in Michigan.  Why so many psychologists have aligned themselves with the defense, to the point of demonstrating and even admitting bias in that direction, is a puzzle for another time.  What matters for this book is that these politics exist and work to diminish the credibility, and ultimately the safety, of children.”

Yeah, there are a lot of names in that passage.  I didn’t fill in the background for them – Cheit explains them all clearly in his book, and, honestly, I’m not sure they matter so much to understand the point he’s making here.  There is a clear bias against children, and clearly illogical, inconsistent reasoning going on.  So how does that square with the claim that these people, such as Dr. Bruck, are scientists?

I have only read one paper authored by Bruck – I chose the most recent publication listed on her faculty page at Johns Hopkins, a review of the way children disclose abuse.  She cites a number of other studies and presents some of their data to argue that although children clearly do wait long periods of time before disclosing abuse, she feels that other behaviors described as comprising “childhood sexual abuse accommodation syndrome,” such as possible waffling in a story about what happened once a disclosure is made, do not occur.  And the impression I got from Cheit’s book is that she has testified as such on the behalf of the defense in several cases – testified that children who behave that way are probably not real victims and might be imagining or fabricating their stories of abuse.

I could go through her article bit by bit and explain why I think her interpretations are unreasonable, but that would belie my main point – just because something is “science,” and just because there are numbers attached to it, does not mean there is only a single conclusion that can be drawn.  In my opinion, the closer one gets to mathematics, the more agreement there will be that a certain result implies a certain underlying truth.  But by the time you’re dealing with physics, or chemistry, or biology, or psychology, or economics (which I see as being roughly scalar jumps from one to the next, in terms of what objects are under study), a single experimental result can be interpreted in various ways.  Because most experimental set-ups use model systems.  You always have to accept that there will be some underlying assumptions made about what features of your system will and will not affect your interpretation.

Let’s say you’re replicating some biological process in a test tube.  You have to assume that the tube itself won’t matter – and for some experiments, this is known to be false, and researchers doing those studies might take a few aliquots of purified protein and incubate them in their test tube before even starting the experiment, hoping to block off potential sites along the wall of a tube where the protein might stick.  And then, if that researcher, someone who’d realized the tube was a problem and was doing those pre-incubations, later read a study that involved their model system from someone who did not pre-incubate protein in the test tubes, obviously the latter results would seem suspect.  But it’s possible that no one else would realize that there would be problems with the data.

For a study in psychology, this problem seems even bigger.  Some researchers have argued that the specific wording for questions on psychology surveys can result in different results (and, sure, you might worry about results that are so exquisitely sensitive to seemingly-extraneous details.  I do) – so there are possibly many, many nearly-imperceptible details of a model system that might affect a study’s results, and that you might therefore consider relevant when analyzing someone else’s work.

So, instead of going through Bruck’s paper bit by bit, I’ll point out only what I see as the largest difference between the studies she cites and the real-world situations she has tried to extrapolate those findings to: timing of interviews to identify abuse.  If a case involving many children is going to be prosecuted, the trial happens all at once, and so data collection has to happen all at once – there isn’t time to wait 4 years for one child to be ready to say what happened, 5 months for another, 7 years for another.  Every interview might be conducted within that first year.  And her review cited a study reporting that only 43% of “young” girls (oft older than those involved in the trials described in Cheit’s book) disclosed abuse after STD evidence known to the girls and the interviewer was found that demonstrated that they had been abused.  So, given that many children won’t report abuse right away even when there is already proof known to the person they’re talking to, trying to extrapolate to the way disclosures are made by children who are younger, who do not have proof, and who are reporting what happened perhaps long before they are psychologically ready to do so… to me, seems foolish.

Not that her analysis doesn’t qualify as “science.”  Bruck has made some assumptions, and she has looked over some data, and she’s made some conclusions.  Her methodology is consistent with the practice of science.  Her conclusions may be wrong, but that’s most likely a problem with her underlying assumptions.

Anyway, that’s all I have to say about that.  I had originally thought I’d work in some clever references to other anti-child science (here is one of the more horrifying examples I was considering – which is even salient because you could argue that the children’s susceptibility to suggestion goes along with Bruck’s pro-defense stance), or work in a passage comparing the ways in which children’s testimony was discounted to the way that women’s testimony was (and often still is) discounted in violative assault cases, or a passage contrasting all the attacks on child reliability with people’s acceptance of adult witness testimony, even though there have been many studies showing that adults aren’t very good witnesses and some showing that children might not be much worse.  But, you know, putting all that in would make this much longer and sadder, and it’s pretty long already.