UK Family Law Reform
A Short Treatise on Woozles and Woozling
National Parents Organization
May 14, 2014 by Robert Franklin.
The excitement in the family law world occasioned by two papers –
one by Dr. Richard Warshak that was endorsed by 110 eminent social scientists
around the world and one by Dr. Linda Nielsen – has died down
a bit with no blood actually having been shed. Whew, that was a close
call. Warshak, et al and Nielsen thoroughly skewered work done by Dr.
Jennifer McIntosh and colleagues that was widely used by various Australian
organizations involved in parenting post-divorce to marginalize fathers,
particularly regarding children under the age of three. McIntosh’s
work was cited time and again for the proposition that fathers having
their very young children overnight was a bad idea. Australian courts
embraced the notion as did custody evaluators, lawyers and the like.
About that, Warshak, et al said this:
Advocates are promoting a report issued by an Australian government
agency (McIntosh, Smyth, & Kelaher, 2010) as a basis for decisions
regarding parenting plans for children of preschool age and younger.
Accounts of the report appearing in the media, in professional seminars,
in legislative briefs, and in court directly contradict the actual data,
overlook results that support opposite conclusions, and mislead their
A “background paper” describing the Australian report, posted
on the Internet (McIntosh & the Australian Association for Infant
Mental Health, 2011), illustrates all three characteristics.
In short, according to 111 prominent scientists, both McIntosh and fellow
authors, and advocates that use their work to marginalize fathers in
the lives of their children, make claims that “contradict the
actual data, overlook results that support opposite conclusions, and
mislead their audience.”
If you’re a researcher in the field of children and their well-being
under various parenting arrangements, as McIntosh is, that’s a
blow to your reputation that could conceivably be fatal. Face it, when
111 scientists around the world combine to condemn your work and attempt
to set straight the scientific record on overnights for kids, Houston,
you have a problem.
But that wasn’t all. Certain Australian organizations began publicly
proclaiming their intention to stop using McIntosh’s work as a
guide to their decisions about parenting and to heed the extensive literature
review by Warshak, et al. People and organizations have begun to run
away from McIntosh and her views on the subject of parenting plans for
very young children. Academic researchers rely on grants to do their
work. A bad reputation, particularly a biased bad reputation is not
the stuff of funding for future research.
Not only that, but, shortly before Warshak, et al published their review
of the science, Dr. Linda Nielsen published her own take on McIntosh’s
work. It wasn’t flattering. Entitled “Woozles: Their Role
in Custody Law Reform, Parenting Plans and Family Court,” it was
nominally a cautionary tale about the uses and misuses of social science,
but its main targets were McIntosh and her colleagues and the uses of
their work to further marginalize fathers in children’s lives.
What’s a Woozle? The term is taken from the delightful Winnie
the Pooh by A.A. Milne. Recall that, at one point, Winnie and friends
had gotten the idea that a sinister beast – a Woozle - was following
them. Around and around a tree they went, and with each revolution,
they discovered more and more evidence of same in the form of footprints.
And the more footprints they found, the more certain they became that
there was a Woozle on the loose. Of course there was no Woozle; their
“proof” of its existence was their own footprints.
The woozle effect in the social sciences has been known for a long time.
Woozling involves a number of behaviors that, all taken together, serve
to promote at best dubious and at worst downright false conclusions.
Therefore, things like cherry picking of data, confirmation bias and
ignoring contradictory data are parts of the woozling process. So are
evidence by citation in which one or two studies cited again and again
attain an authority far beyond their actual scientific value. White
Hat bias enters the picture too. That occurs when getting the “right”
conclusion, the one that’s presumed to be beneficial to society,
becomes more important than drawing only the conclusions merited by
The press and other researchers contribute to woozling too. Journalists
are seldom equipped to rigorously analyze scientific papers and, always
on the lookout for a scintillating story, often promote certain scientific
points of view that are far less nuanced than they ought to be and ignore
conflicting data. And then there are always scientists who are either
too lazy to look at what studies actually are able to conclude or who
have their own bias they want to promote, so they consciously or unconsciously
contribute to the woozling of data.
That brought Nielsen to the science on overnights for children of separated
parents. Should very young children, i.e. those under five, spend some
nights with each parent or every night with one? If “some with
each” is the correct answer, then how much with each? Some 31
studies have delved into the issue of shared versus sole or primary
parenting when children have some overnight visits with Dad. Of those,
eight have dealt with children under six, among them one by McIntosh,
Smyth, Kelaher and Wells published in 2010.
As Nielsen makes clear, the first six of those studies offer essentially
nothing to back up the claim – the woozle - that even very young
children should not spend overnights with their fathers. Indeed, children
whose parents had previously been married and who frequently spent overnights
with their fathers did better than those who didn’t. There were
“no differences in social or behavior adjustment… and frequent
overnighters had better relationships with their fathers and were better
Three other studies examined the children of parents who may not have
ever been married. As such, some of the children had not been able to
form secure attachments to their fathers and so, their responses to
spending overnights away from Mom could be seen as more problematical.
But, “overall the frequent overnighters had marginally better
outcomes, even after accounting for parents’ levels of violence,
conflict, and education. More important, violence between the parents
had no worse impact on the frequent overnighters than on the other children.”
That leaves two studies, McIntosh, et al previously mentioned and another
by Tornello, Emery, et al. The Tornello study was astonishingly flawed
in at least two respects. First, it relied exclusively on data from
the Fragile Families and Child Wellbeing data that have been gathered
by McLanahan and her colleagues since 1998. The people making up McLanahan’s
cohort are utterly unrepresentative of the population of Americans and
certainly not of Australians. Nielsen describes the cohort thus:
65% had no high school degree; 85% were African or Hispanic American;
and 60% were below the poverty level. Slightly more than 85% were not
married when their children were born. Of these, 30% were not living
together and 20% no longer had a relationship with each other when their
child was born. Before their children’s fifth birthday, 50% of
these fathers and 10% of these mothers have served time in jail.
So whatever else may be said about the Tornello study, it’s clear
that extrapolating it to the population at large is intellectually impossible.
Worse, Tornello and colleagues chose to measure the children’s
behavior using a method that had been shown to be invalid. The Toddler
Attachment Q Sort has validity only if the data is gathered by trained
observers who gather information from observing the interactions of
mothers and their children over a period of hours. The idea is to rate
the level of attachment by the child and any disorganization of same.
But when mothers are allowed to self-report, the TAQ loses validity.
So which method of gathering observations did Tornello, et al use? That’s
right, they allowed mothers to tell scientists what they experienced.
Given the choice between the valid and invalid ways of observing, they
chose the invalid method.
Even so, their findings revealed no reason why children should not spend
overnights with their fathers.
Consistent with the seven studies already described, there were virtually
no differences between the overnighters and nonovernighters. On 14 regression
analyses for the seven measures of well-being, only one statistically
significant difference emerged: The children who frequently overnighted
at age 3 years displayed more positive behavior at age 5 years than
the rare or no overnights groups.
That brings us to McIntosh, et al, about whose work I’ll say more