However, I will not take up readers’ time with the loaded language. You will all be familiar with its like (and citation of Elizabeth Sheehy and Linda Neilson is all you need register to have it accurately pigeon-holed). The paper in question is “U.S. child custody outcomes in cases involving parental alienation and abuse allegations: what do the data show?”, By Joan S. Meier, Journal of Social Welfare and Family Law 42:1, 92-105 (2020).
The site Researching Reform summarised the paper’s findings thus,
“The first ever national study has confirmed that mothers who make complaints of child abuse against a father in court are more likely to lose contact rights with their children. The research also found that this risk doubled when a father made a counter-claim of parental alienation, leading the researchers to conclude that ‘alienation trumps abuse’.”
What image does this conjure? A poor abused mother who plucks up the courage to mention in court the abuse she has suffered at the hands of her male partner, only to find that her children are taken off her as punishment for her temerity. This faithfully reflects the tenor of the paper itself. We read this in its Conclusion,
“The data support the widespread critiques of family court proceedings sending children into the care of destructive and dangerous parents.”
It is eminently clear from the rest of the paper that “dangerous parents” means fathers. There is nothing within the paper to support this conclusion. For all we know from the paper, the courts might have been 100% perfect in their rulings, always ensuring a safe adult was given custody of the children. We simply know no different based on the contents of the paper. The “conclusion” comes from nowhere – or, rather, it comes from pre-existing prejudice but not from the data within the study.
The blame for this failure to uphold academic standards lies not only with the author but also with the peer reviewer(s), and the journal itself. I have reviewed a great many academic publications. A standard requirement is to confirm that “the conclusions follow from the arguments of the text”. Clearly, in this case, the reviewer(s) failed in their duty. That conclusion should not have been permitted.
In the Abstract we read,
“The findings confirm that mothers’ claims of abuse, especially child physical or sexual abuse, increase their risk of losing custody, and that fathers’ cross-claims of alienation virtually double that risk. Alienation’s impact is gender-specific; fathers alleging mothers are abusive are not similarly undermined when mothers cross-claim alienation.”
Consider the first part of that conclusion, “The findings confirm that mothers’ claims of abuse, especially child physical or sexual abuse, increase their risk of losing custody”. I can find nothing within the paper to substantiate it. Whilst the paper tells us that mothers reporting a father’s abuse lost custody in 26% of cases, we are not told what percentage of mothers lost custody when no allegations of abuse were made, so the conclusion in the Abstract appears from nowhere. It might be that Meier’s dataset does support the conclusion, but the text of the paper does not support the statement in the Abstract and so is another failure of the peer reviewers to do their job.
The paper purports to reveal certain correlations, or associations, between allegations of abuse, allegations of alienation, and courts ruling to change the custodial parent from mother to father, or vice-versa. The wording of the above quote (and throughout the paper) invites us to interpret these associations thus: if a poor abused mother has the nerve to allege that her ex-partner is abusive, she is likely to be punished for it by removal of her children. Worse – the wicked father may make a counter-allegation of alienation which will have the effect of further increasing the likelihood that she will lose her children. What woman reading this will not be hopping mad?
Well, the better-informed ones, actually.
Let me unpick the paper a little so you can see how propaganda gold is spun out of factual straw.
Start with this quote referring to the source of Meier’s data (published court opinions),
“the majority of the opinions analyzed were appellate decisions.”
So, the data derives predominantly from appeal courts. What does that tell you in the context of child custody cases? In the UK, 92% of resident parents after parental separation are mothers. I believe the percentage is slightly lower in the USA – perhaps 85%. (If an American reader has a better figure I’d be obliged). Straightaway we see that the data source used by Meier is severely skewed.
Who is likely to be the appellant in such cases? That would be the parent who failed to gain custody in the preceding case – which means fathers in about 85% of cases. If the outcome of the appeal is to change the children’s residency, in 85% of cases it can only go from mother to father. So, simply due to sample bias we can anticipate as obvious what the paper does in any case reveal…
“We found that mothers losing custody were over-represented in the appeals”
Well, bloody obviously! And, surprise, surprise…
“There were lower custody loss rates among the non-appealed cases.”
Clearly, in cases where the father sought custody in the initial hearing, but failed, and has subsequently made an appeal, any decision by the court to change residency can only go one way.
Can Joan Meier truly be unaware that this is simply sample bias? This same sample bias contaminates all the observations which follow, irrespective of allegations of abuse or alienation.
This is another issue over which the peer reviewer(s) failed badly. This sample bias is so serious the paper should have been rejected as it invalidates all that follows.
However, let’s press on, ignoring that flaw, fatal though it actually is.
Recall that the claims revolve around the effect that allegations of abuse or of being alienated have upon the likelihood of custody being reassigned. From this point on I shall simply assume the data that Meier quotes is valid (as I have no way of checking it). What is in contention is how the data may be interpreted.
To pull the argument apart most cogently the best place to start is this revealing quote,
(A) “Mothers and fathers fared equally in several circumstances: First, when a parent’s claim of alienation was credited (across abuse and non-abuse cases) mothers and fathers lost custody at identical rates (71%). More broadly, win rates were also identical (89%) for mothers and fathers when the other parent was found to have committed alienation. Second, and notably, virtual parity is apparent in the non-abuse alienation cases, where win rates are 58% (fathers) and 56% (mothers).”
I think we may interpret this to mean that, when the court believes an allegation of alienation, it has a substantial effect on the likelihood that they rule to change the children’s residency. Since alienation involves psychological trauma to the child, and the courts are mandated to make the child’s welfare paramount, if the court believes this is happening it is hardly surprising that the result is frequently a court ruling for change of residency. Nothing odd here.
Moreover, Meier’s quote, above, indicates that this is not gendered: both mothers and fathers as resident parents guilty of alienating run the same risk of having the children moved to the other parent.
So far, so good.
Now consider this quote from the paper, which encapsulates Meier’s main claim,
(B) “Across all alienation cases (both with and without abuse claims), when a father alleged a mother was alienating they took custody from her 44% of the time (166/380). When the genders were reversed, mothers took custody from fathers only 28% of the time (19/67).”
Meier’s interpretation is that this observation implies that allegations of alienation are a cunning plan to obtain custody which works far more effectively for fathers than mothers. (Note in passing how statistically weak is the 28% figure, resulting as it does from just 19 cases out of an original dataset of 4,338).
How does quote (B) differ from quote (A), above? The difference lies in whether the court believes the claim of alienation. The custody reversals in (B), i.e., 44% or 28%, are less than in (A), i.e., 71%, because the latter is conditional upon the alienation claim being believed. But alienation will not be believed – or not believed to be sufficiently serious – in all cases.
Taking into account the non-gendered nature of the courts’ response to allegations of alienation when they believe them, the reason for the differing percentages in (B), i.e., 44% versus 28%, can only be because the courts disbelieve more claims by mothers of being alienated than claims by fathers of being alienated. The possible reasons for this are, (a) the courts are biased against mothers, or, (b) the courts are broadly accurate in their assessments but a larger percentage of alienation claims by mothers than by fathers are false within Meier’s skewed dataset.
Strictly, from the data within the paper, we cannot decide between these possibilities. Meier simply takes as axiomatic that the explanation is (a), bias against mothers. But a systemic bias against mothers in the courts seems unlikely, both in view of quote (A), above, and also because mothers are awarded custody far more often (~85%).
The more likely interpretation of Meier’s own data is that they reveal that a larger percentage of claims of alienation by mothers are false than claims of alienation by fathers, bearing in mind that this relates only to Meier’s skewed dataset. This need not be inconsistent with a claim that, other things being equal, mothers and fathers are equally likely to be alienators. But – crucially – other things are not equal in Meier’s dataset. We have already seen the skew due to being predominantly appellate court data, and hence the huge gender-skew in which sex is the appellant. But there is another skew: quote,
“This article focuses primarily on findings related to cases where a mother accused a father of abuse. There were some – although not many – cases where the genders were reversed.”
Of the 4,338 cases in the total dataset, at least 52% involved allegations of abuse, almost all being allegations by mothers against fathers. Some proportion of these the court of appeal will conclude to be false. Necessarily, then, virtually all allegations of abuse which the court does not believe will be allegations by mothers against fathers. Here we have another crucial skew in the data. These allegations are likely to have played a part in the decision to place the child with the mother (in 85% of cases). But where the court of appeal has failed to find these abuse allegations to be convincing, is it not reasonable to expect this to count against the false accuser? This is the further skew in the data sample which renders Meier’s findings unsurprising.
There is no need to opt for an interpretation that the courts are biased against mothers, as Meier’s findings are only what would be expected on purely statistical grounds given the multiple biases in the sample of data analysed. The entire “study” is an exercise in statistical misdirection, and the claims made in the Conclusion and Abstract are invalidated by serious statistical shortcomings.
No comments:
Post a Comment