Content uploaded by Sanford Braver
Author content
All content in this area was uploaded by Sanford Braver on Apr 29, 2018
Content may be subject to copyright.
Full Terms & Conditions of access and use can be found at
http://www.tandfonline.com/action/journalInformation?journalCode=wjdr20
Journal of Divorce & Remarriage
ISSN: 1050-2556 (Print) 1540-4811 (Online) Journal homepage: http://www.tandfonline.com/loi/wjdr20
Does Joint Physical Custody “Cause” Children’s
Better Outcomes?
Sanford L. Braver & Ashley M. Votruba
To cite this article: Sanford L. Braver & Ashley M. Votruba (2018): Does Joint Physical
Custody “Cause” Children’s Better Outcomes?, Journal of Divorce & Remarriage, DOI:
10.1080/10502556.2018.1454203
To link to this article: https://doi.org/10.1080/10502556.2018.1454203
Published online: 13 Apr 2018.
Submit your article to this journal
View related articles
View Crossmark data
Citing articles: 1 View citing articles
Does Joint Physical Custody “Cause”Children’s Better
Outcomes?
Sanford L. Braver
a
and Ashley M. Votruba
a,b
a
Department of Psychology, Arizona State University, Tempe, Arizona, USA;
b
Department of
Psychology, University of Nebraska–Lincoln, Lincoln, Nebraska, USA
ABSTRACT
Policymakers and researchers are concerned with whether joint
physical custody (JPC) produces better outcomes for children
than sole custody. Although several review articles summarizing
up to 61 empirical articles demonstrate very positive answers,
many of the research designs used compromise the ability to
claim that it is JPC per se—and not selection effects—that
causes the effect. We discuss several research design issues,
such as propensity score analysis, that can more powerfully
probe the question of causality. Some studies have already
been conducted employing these strategies and more are
recommended and likely to soon be forthcoming. On the basis
of this comprehensive review we conclude that JPC probably
does cause benefits to children on average, and that social
scientists can now provisionally recommend rebuttably pre-
sumptive JPC to policymakers.
KEYWORDS
Joint custody; shared
parenting; causal inference;
divorce
Research generally suggests that children of divorce are at increased risk for
social, psychological, and educational difficulties (Braver & Lamb, 2012;
Chase-Lansdale, Cherlin, & Kiernan, 1995; Cherlin et al., 1991). Thus,
when it comes to family law and custody determination, policymakers and
researchers are concerned with knowing what types of postseparation par-
enting arrangements are generally the most beneficial for children. More
specifically, interest has focused on whether joint physical custody (JPC)
produces better outcomes for children than sole maternal physical custody
(SPC). Another related question is how well JPC works when there is
parental conflict or when one or both parents did not want JPC.
A handful of review articles have examined the large number of studies that
have addressed these questions and generally found that children with JPC
arrangements were significantly better off than those in SPC (generally mater-
nal custody) arrangements (Baude, Pearson, & Drapeau, 2016; Bauserman,
2002; Nielsen, 2015,2017). Bauserman (2002) performed a meta-analysis on
33 studies with a combined sample size of more than 2,650 children, of whom
about one-third had JPC arrangements. Children with JPC plans scored
CONTACT Sanford L. Braver sanford.braver@asu.edu Department of Psychology, Arizona State University,
PO Box 871104, Tempe, AZ 85287-1104, USA.
JOURNAL OF DIVORCE & REMARRIAGE
https://doi.org/10.1080/10502556.2018.1454203
© 2018 Taylor & Francis Group, LLC
significantly higher on adjustment measures compared to children in sole
custody. This was true for nearly all categories of adjustment (except academic
adjustment), including general measures of adjustment, family relations, self-
esteem, emotional adjustment, behavioral adjustment, and divorce-specific
adjustment. This suggests that JPC can benefit children in a wide range of
domains. Baude et al. (2016) replicated this finding in a meta-analysis of 17
studies that included some 36,000 families.
Nielsen also updated the Bauserman (2002) review in 2015, citing 40
studies, then again in 2017, summarizing 54 studies, and most recently
summarizing 61 studies (Nielsen, 2018). In all three reviews she found that
children with JPC arrangements were generally better off than children with
sole custody arrangements. Across these studies, children with JPC arrange-
ments showed (a) better grades and cognitive development; (b) lower levels
of depression, anxiety, and dissatisfaction; (c) lower aggression, drug use, and
alcohol use; (d) better physical health and lower smoking rates; and (e) better
father–child relationships. Nielsen (2015) also concluded that the benefits of
JPC arrangements occur even when there is parental conflict. Nielsen (2015)
explicitly noted that in 11 of the 40 studies, the researchers stated that their
sample included high-conflict and litigating parents. Further, in 16 of the
studies, either the parents with a JPC arrangement had as much conflict as
those with sole custody arrangements, or the outcomes remained better for
JPC children even controlling for parental conflict.
Despite being armed with this robust and consistent recent literature
attesting to the substantial benefits of JPC, advocates have run into consistent
opposition in converting the findings into a legal presumption: an assump-
tion made and accepted by a court as a basis for their decision in the case to
be decided. Generally, presumptions in family law are considered rebuttable
and are accepted by the court until disproved. The assumption “will stand as
a fact unless someone comes forward to contest it and prove otherwise”
(Rebuttable Presumption Law and Legal Definition, 2017). In family law, for
example, the child support amount arising from each state’s child support
guidelines constitutes a rebuttable presumption of the proper child support
award (Elrod, 1990). Proponents of JPC have engaged in many unsuccessful
attempts to pass laws making JPC a presumption that is rebuttable on
showing that, for some specifiable legal reason, such an arrangement should
not be considered in the child’s best interest in the particular case.
One of the key sources of opposition to making JPC a legal presumption
arises because of scientific considerations.
1
The objection focuses on a certain
limitation of the research design historically used by the vast majority of the
studies comparing the impact of JPC and SPC arrangements. This research
design has been termed the static group comparison (Campbell & Stanley,
1963) and is often also referred to as a cross-sectional study. A static group
design compares two or more preexisting groups. In this case the research
2S. L. BRAVER AND A. M. VOTRUBA
compared families with JPC arrangements to those with SPC arrangements—
generally maternal custody. The term cross-sectional (implying a single point in
time, with preexisting groups) is generally contrasted with the longitudinal
study (implying multiple points in time). The limitation of this design gen-
erally ensues from the fact that there is “no formal means of certifying that the
groups would have been equivalent had it not been for”the custody arrange-
ment (Campbell & Stanley, 1963, p. 12). Whatever the basis is under which the
individuals become sorted into groups represents selection, which constitutes a
substantial threat to the internal validity of the causal conclusion.
For the issue at hand, it is mostly self-selection that comprises the most
plausible alternative explanation for the differences found between JPC and
SPC children. Specifically, during the historical period when many of the JPC
studies were conducted, families were granted JPC only if both parents (more
or less) freely declared that this was the arrangement they preferred. If either
parent declared that he or she was unalterably opposed to such an arrange-
ment, the courts would typically not grant it. Thus, JPC was virtually never
imposed on consistently unwilling families—in some instances because the
statutes specifically precluded it. Thus, couples in which both parents wished
for JPC—which was a distinct minority (Braver & O’Connell, 1998;Maccoby
& Mnookin, 1992)—were compared to couples in which one or both parents
opposed it. Hence, the sorting into the two comparison groups was based
nearly entirely on the parents’own decisions, resulting in self-selection. It is
also well known that many discernable factors, as identified later in this article,
might discriminate between couples making these two choices and that these
factors often are also associated with better child outcomes regardless of the
custody arrangement. This is an important methodological limitation because
it could be these self-selection factors, rather than the custody arrangement per
se, that accounts entirely for the advantages found for JPC children.
The reality of this methodological limitation found in much of the
research on JPC has implications both for scientific inquiry and for policy
development. Scientifically speaking, when exploring a cause–effect relation-
ship, if any plausible alternative explanation happens to be entirely respon-
sible for the effect, the causal conjecture is thereby invalidated. Specifically, if
selection accounts for the entirety of an effect, then enacting the “cause”
variable will not have the anticipated impact on the “effect”variable. In terms
of custody policy, if the JPC arrangement is not the cause of the benefits, if
instead self-selection happens to account for all the positive findings, impos-
ing JPC (rather than letting parents choose it) will not have the mostly
beneficial effects indicated by the research, as noted by Bauserman (2002),
and Fehlberg and colleagues (2011). Thus, a presumptive law, which would
impose JPC over the opposition of one of the parents, might fail to create the
intended benefits. As Emery, Otto, and O’Donohue (2005) concluded from
research using this methodology, “we cannot extrapolate from voluntary
JOURNAL OF DIVORCE & REMARRIAGE 3
joint physical custody to circumstances when joint physical custody is
imposed upon parents by laws favoring joint physical custody . . . or by
judges who order it”(pp. 16–17).
Although the static group research design with self-selection into compar-
ison groups has clear limitations for assessing causality, other methodological
approaches offer greater promise. The gold standard methodological
approach that can fully overcome the barriers to drawing causal conclusions
is the randomized experiment (Cook & Campbell, 1979). For example, if
couples were assigned at random to either JPC or to SPC, any subsequent
differences in children’s well-being could be unambiguously attributed to the
custody arrangement. It is true that family courts have on relatively rare but
increasingly common occasions been convinced to deploy random assign-
ment for various purposes (Ballard, Holtzworth-Munroe, Applegate,
D’Onofrio, & Bates, 2013; Beck et al., 2009; Braver, Sandler, Hita, &
Wheeler, 2016; Mauricio et al., 2017; Rossi et al., 2015; Sandler et al., 2016;
Winslow et al., 2017). Nonetheless, there has never been and never will be an
instance of judges assigning custody of children at random. Thus, although
conducting a randomized experiment would clearly be the best methodolo-
gical option for assessing the causal mechanism, it is out of the question.
Research designs that probe causality
There are, however, a number of additional methodological approaches that
would allow researchers to probe causality, albeit not prove it. The inability—
for practical, ethical, or physical reasons—to assign treatments at random is an
extremely common one in social science and even physical science. This
problem has prompted a great deal of recent scholarly work devoted to going
beyond static group comparisons to render causal inferences more credible.
These approaches include (a) employing statistical controls; (b) propensity
score analysis; (c) natural experiments; and (d) regression discontinuity or
interrupted time series quasi-experiments. In regard to JPC and SPC children’s
outcomes, two more approaches present themselves: (e) differentiating the
findings on the basis of parents’initial custody preferences; and (f) examining
outcomes in jurisdictions where JPC is already a presumption or a norm.
Statistical controls
The most common approach to strengthen the possibility of establishing
causality is to employ statistical controls. The intent of this technique is to
statistically hold constant, adjust for differences in, covary out, partial out,
control for, correct for, or equate for (all preceding terms are essentially
synonyms) these self-selection factors. According to the Berkeley Glossary of
Statistical Terms,“to control for a variable is to try to separate its effect from
4S. L. BRAVER AND A. M. VOTRUBA
the treatment effect, so it will not confound with the treatment”(in this case,
the custody arrangement; Stark, 2017). It is important to recognize that such
research takes place under one of two distinct statistical approaches. The first is
the group-oriented approach that treats JPC and SPC families as two distinct
classes of people. Group-oriented approaches, in general, use independent
group ttests and analysis of variance (ANOVA) as their main statistical
tools. When they attempt to control for any variables, they move to an analysis
of covariance (ANCOVA), to “covary out”the possible confound. The second
approach treats most of the variables as continuous ones, not as classes or
groups. This approach uses multiple regression as its main statistical tool.
Typically multiple regressions will treat JPC versus SPC as a dummy or binary
variable, and enter it into the analysis after the control variables have been
entered. In this (or in another equivalent) way, the self-selection variables are
controlled for or partialed out, allowing a firmer inference that it is the custody
arrangement per se that is responsible for the outcomes. The regression
approach and the ANCOVA approach yield virtually identical results and
are merely two different but equivalent approaches (Huitema, 2011).
In JPC and SPC studies, most researchers consider parent conflict and
family income to be the two most important self-selection factors in that
both are thought to powerfully affect both the self-selection of JPC arrange-
ments and child well-being. Accordingly, quite a large number of recent
studies have attempted to control for these two factors in evaluating the impact
of JPC. Nielsen (2018) cataloged these 60 studies. Of the 36 studies that
considered parental conflict, JPC children had better outcomes on all measures
in 18 studies, equal to better in 11 studies, equal in 3 studies, and worse
outcomes on one of the measures in 4 studies. In the 42 studies that considered
family income, JPC children had better outcomes on all measures in 25 studies,
equal to better outcomes in 9 studies, equal outcomes in 4 studies, and worse
outcomes on one measure but equal or better outcomes on other measures in 4
studies. As Nielsen (2018) also pointed out, the links between income and
children’s well-being in the vast literature on this topic actually find only weak
and indirect effects, with the exception of children growing up in poverty.
Although the two self-selection factors of income and parental conflict are
often seen as the most consequential, they certainly do not exhaust the list of
potential factors influencing children’s outcomes. This fact is critical because
the effectiveness of the statistical control approach is greatly compromised if
other selection factors are strongly at work. Among these additional factors
might be mother’s and father’s level of education, the child’s age, the parents’
ages, which parent wanted the divorce, each parent’s mental health, how
guilty each parent felt about the breakup, and so on.
One study was quite comprehensive in identifying which factors might set
JPC and SPC parents apart (Gunnoe & Braver, 2001). This study was some-
what unique in that it was both longitudinal and captured data before the
JOURNAL OF DIVORCE & REMARRIAGE 5
divorce was final—that is, before any custody arrangement became official. In
fact, the initial interview with the parents took place within a short 2.5 months
after the initial petition for divorce, which starts the legal process of divorcing.
The study assessed fully 71 predivorce variables, including all the ones men-
tioned earlier, that might plausibly differentiate between families who ulti-
mately obtained joint legal versus sole legal custody (with maternal physical
custody). Twenty of the 71 factors indeed discriminated at a statistically
significant level parents who ultimately obtained sole or joint legal custody.
All 20 factors were then simultaneously controlled in a subsequent ANCOVA
comparison of the 52 sole and 26 joint legal custody families 2 years post-
divorce. The children in the families with joint legal custody continued to have
fewer adjustment problems than children in sole custody families, over and
above the predivorce selection factors. It should be noted that it was legal
custody, rather than physical custody that was at issue here, because the study
was conducted at a time before there were sufficient numbers of JPC cases to
yield adequate statistical power. Note, however, that Bauserman’s(2002)meta-
analysis found that that joint legal custody and JPC bestowed largely equal
benefits. It is also important that the more positive outcomes for JPC children
were not moderated by the level of predivorce conflict between the parents.
In conclusion, statistical controls, the most ubiquitous approach to dealing
with the self-selection confound, have shown rather overwhelmingly that JPC
confers substantial benefits to children over and above, or independent of,
self-selection factors.
Propensity score analysis
Propensity score analysis is a relatively new technique that deals with the issue
of static or preexisting groups by providing another means for equating the
groups on a large number of variables (covariates) measured at a baseline point
(West, Cham, Thoemmes, et al., 2014). Once they are “equated”at baseline
(via matching, stratification, weighting, or ANCOVA) on all the covariates
(e.g., parental conflict) that predict group selection, the comparison of the
groups’differential outcomes rules out the effect of these potential confound-
ing factors (Shadish, Cook, & Campbell, 2002;West,Cham,&Liu,2014). This
strongly enhances the internal validity of the study and thereby the inference of
causal impact.
Propensity score analysis is an upgrade from traditional approaches that
equate groups on only a few variables, instead allowing equating on a large
number of baseline covariates simultaneously by creating a single propensity
score that summarizes all of the covariates. The score is typically constructed
using a logistic regression equation in which the full set of covariates is used
to predict group membership. Unlike traditional approaches, however, it
leaves out simultaneous consideration at this stage of the outcome variable
6S. L. BRAVER AND A. M. VOTRUBA
of interest. Essentially, the “propensity score is the predicted probability that
the person will be assigned to the treatment group based on his or her scores
on each of the full set of covariates”(West, Cham, Thoemmes, et al., 2014, p.
908). If the groups are successfully equated, then it is possible to arrive at an
unbiased estimate of the causal effect of the treatment. In our case, it would
thus be possible to examine what causal effect JPC or SPC had on child well-
being. However, propensity score analysis has advantages over regression-
oriented statistical controls because it assesses overlap of the two groups
being compared; it makes no assumptions about the functional form of the
relationship between the covariate and selection, such as linearity; it allows
nonparametric as well as parametric conditioning; and it allows checks of the
putative selection model.
One of the challenges of performing propensity score analysis is that to get
an accurate propensity score, it is necessary to measure all or nearly all
covariates that might be confounded with self-selection into JPC or SPC
arrangements and child well-being (West, Cham, Thoemmes, et al., 2014).
This could mean measuring a very large number of potential covariates at
baseline. In addition to new data collection efforts, researchers can consider
secondary analysis of data sets that included many potential such covariates
that have already been collected, as did Gunnoe and Braver (2001).
To our knowledge, no researcher has yet attempted to use this powerful
and sophisticated methodology to examine the causal effect of custody
arrangements on child well-being. Because propensity score analysis achieves
results close to those of a randomized experiment (Cook, Shadish, & Wong,
2008; Shadish, Clark, & Steiner, 2008), we believe this is a strong candidate
for future research.
Natural experiments
Natural experiments also often allow causal conclusions to be fairly made. In
natural experiments, the assignment to a treatment condition is not made at
random by the researcher, but is made instead by some independent event;
for example, nature, the weather, sickness, or policy changes. The key to
whether the causal inference is valid in any natural experiment is whether
“the event . . . allows for the random or seemingly random assignment of
study subjects to different groups”(Messer, 2017, italics added).
Because custody laws are a matter of much legal and cultural ferment and
change, new laws and new court holdings are constantly coming into being.
Comparing couples assigned by some means to JPC to couples assigned to SPC
could plausibly constitute a natural experiment that would allow causal infer-
ences about the custody arrangement’s impact on child outcomes. The validity
of such an inference rests completely on the exact nature of the design,
however. Consider, for example, a hypothetical study comparing couples
JOURNAL OF DIVORCE & REMARRIAGE 7
who divorced before a JPC presumption took effect to another group of
couples who divorced after the presumption took effect. Only to the degree
that we might fairly regard as “random or seemingly random”whether the
exact date of each specific case’s divorce decree fell either before or after the
law change would the causal inference about the impact of the JPC presump-
tion on the child’s well-being be valid. When other potential causes of any
differences in child outcomes found might also be plausible, they constitute
clear threats to the internal validity of the inference. For example, if “other
change-producing events”(p. 7) that might affect the children’soutcomeshave
occurred between the two observation points (e.g., economic downturns,
housing collapses), the inference risks invalidity. Such an other event “becomes
a more plausible rival explanation of change the longer”(p. 7) the interval
between the two observations. Thus, studies that let only small intervals (i.e., a
few months) intervene between the divorce dates of the couples in the two
regimes are on more solid footing with causal claims.
We are aware of no solid empirical investigations of JPC’s impact on child
outcomes that employed such a natural experiment, but are mindful that
these could be profitably deployed by alert investigators whenever the pas-
sage of a presumptive law seems imminent. With more than 20 states and
numerous countries currently debating new JPC presumption laws (Leading
Women for Shared Parenting, 2017) researchers should note the important
opportunity that exists to study a random sample of families before and
another random sample after such a law takes effect.
It might appear that another natural experiment opportunity exists by com-
paring two nearby jurisdictions with different custody laws, but this rarely is
valid. For example, Douglas (2003) compared a sample of parents from New
Hampshire, which had recently passed a presumptive joint legal custody law, to
a sample from Maine, which did not have such a presumption. The samples were
chosen from six counties matched on several demographic factors. However,
although matched on some variables, many other differences between the
jurisdictions exist, such as radically different child support regimes. Many or
all of these differences could plausibly account for any impact of the new
presumption. Thus, Douglas (2003)admittedthat“more well-controlled designs
are greatly needed”for sound inference (p. 9). In summary, comparing different
jurisdictions at the same time generally constitutes an invalid variant of natural
experiment with which to evaluate the causal impact of JPC on child outcomes.
Quasi-experimental designs: regression discontinuity or interrupted time
series
One of the most important contributions of Campbell and Stanley’s(1963)work
was to identify an extremely important class of research designs, new at the time,
they termed quasi-experiments. These designs are admittedly less conclusive
8S. L. BRAVER AND A. M. VOTRUBA
than randomized experiments, but, when well conducted, only marginally so.
The two quasi-experimental designs we highlight here are the ones best suited to
the evaluation of JPC arrangements or presumptions on child well-being:
regression discontinuity or interrupted time series. For our purposes, these
terms are largely interchangeable and we refer to it as RD-ITS, accordingly.
Whereas a simple pre- and posttest design is very susceptible to the argu-
ment that other causes might have intervened between the two measurement
occasions, the RD-ITS approach minimizes that threat to internal validity by
considering many pretest points and many posttest points. Figure 1 illustrates
this approach: It gathers a sample of many pre-law-change cases and many
post-law-change cases and plots them all on the horizontal axis by the date of
the final decree. The child well-being measure(s) for each case are plotted on
the vertical axis. If the law had an impact on child well-being (or any other
relevant outcome measure) it should be evident by an abrupt discontinuity or
Month Custody Arrangement
Finalized
Long Term Child Adjustment
Figure 1. Regression discontinuity design.
JOURNAL OF DIVORCE & REMARRIAGE 9
jump in the trend line tracing the average outcomes over time. Any alternative
explanation of the child outcome results other than the causal impact of the
JPC presumptive law taking effect would have to pass the considerable hurdle
of explaining why the impact occurred at that one exact point in time.
Although we are aware of no existing study that used such a design to
study the impact of JPC presumptions on child outcomes, work preliminary
to an analysis of the introduction of Arizona law has been conducted by
Fabricius and Millar. Moreover, the design can be used to evaluate other
interventions in the family law environment. For example, DeLusé and
Braver (2015) used such a design to evaluate a divorce education program
and deemed such an evaluation rigorous.
Differentiating on the basis of parents’initial preferences
In evaluating the causal impact of JPC arrangements on child well-being,
another methodological strategy rather uniquely presents itself. This occurs
because there are two parents, and they might in fact agree initially on a JPC
arrangement, or they might initially disagree. With one parent initially
against it, JPC sometimes nevertheless prevailed, infrequently because a
court decision overruled that parent, and more commonly because the
opposing parent later withdrew his or her opposition, perhaps because of
professional advice or under pressure of some kind. Braver and O’Connell
(1998) and Maccoby and Mnookin (1992) found that initial mutual agree-
ment on joint custody is relatively rare, between 18% and 23%. Fabricius,
Braver, Diaz, and Velez (2010), among others, discussed the many avenues in
which the bargaining process between the ex-spouses can be influenced by
the “guidance about their chances they receive from judges, attorneys, cus-
tody evaluators, parent educators, and mediators”(p. 257). Mnookin and
Kornhauser (1979) famously called this “bargaining in the shadow of the
law.”Braver, Cookston, and Cohen (2002) presented evidence that it is the
parents’lawyers, in particular, that often influence the process, leading
parents to not pursue their initial preferences by advising them about their
“likelihood of prevailing”in seeking the arrangements they prefer. If analysts
have access to information about the two parents’initial preferences prior to
the decree, they could compare the child outcomes of the “both initially agree
on shared”to the “one initially wanted sole but ‘caved’” groups to probe the
impact of the self-selection alternative explanation. If self-selection is respon-
sible for the benefits of JPC that have been documented, we should expect
that children for whom both parents voluntarily selected JPC will have better
outcomes than those for whom one parent initially opposed it. Nielsen (2014)
identified six studies that catalog parents’initial agreements or lack thereof
about the eventual parenting plan (Braver & O’Connell, 1998; Brotsky,
Steinman, & Zemmelman, 1988; Fabricius & Suh, 2017; Luepnitz, 1986;
10 S. L. BRAVER AND A. M. VOTRUBA
Maccoby & Mnookin, 1992; Pearson & Thoennes, 1990). Most of these are
longitudinal, having assessed parents’initial preferences before the decree
was final. The study by Leupnitz (1986), however, is not longitudinal and
simply stated, without explanation of how it was determined, that “in only
54% of the joint cases had parents agreed from the outset on some form of
shared custody. In the remaining cases there was conflict over the question of
custody initially”(p. 3). Finally, Fabricius and Suh (2017) assessed initial
agreement about custody arrangements by retrospective report. The six
studies in general do not find lower benefits of JPC for the group of parents
who initially disagreed; rather, the benefits of JPC held even when one parent
disagreed on the arrangement, undermining the notion that self-selection
accounts for the totality of JPC benefits.
We encourage researchers with longitudinal data sets with parents’initial
custody preferences recorded to harness this power with additional second-
ary analyses. Notably, Maccoby and Mnookin (1992) have a large data set
that is publicly available at http://www.socio.com/fam2527.php. This could
be leveraged to address this and other important causal questions, but to our
knowledge it has not been done.
We should also note the inferential power of longitudinal studies more
generally. Analyses such as cross-lagged panel studies and structural equation
models at different periods of time are generally regarded as greatly enhan-
cing the ability to make causal inferences even without random assignment.
It has long been noted that family law research needs more longitudinal
studies (e.g., Braver & Lamb, 2012; Braver et al., 1993).
Examining outcomes in jurisdictions where it is already a presumption or a
norm
Finally, yet another inferential approach is or is rapidly becoming available in
the present instance to evaluate this article’s central question. However, this
final approach skirts the causal question per se and instead addresses the
related question of whether the benefits of JPC arrangements found in the
literature will continue to hold when such arrangements are a rebuttable
presumption, or when imposed on parents against their will. It turns out we
have such evidence by examining jurisdictions where JPC is already a pre-
sumption, or where there are already strong norms upholding it. Because JPC
practices are rapidly becoming more widespread throughout the United States
and world, several jurisdictions now have large portions of the recent divorce
cases adopting JPC, some of which were presumably initially disinclined.
Among these jurisdictions are several European countries, including Sweden,
Belgium, and Australia, and several states, including Arizona and Wisconsin.
By examining child well-being or other relevant outcomes in samples of recent
divorces in these locales it is possible to glean answers regarding how well it
JOURNAL OF DIVORCE & REMARRIAGE 11
works when it is imposed, perhaps over the initial objections of one of the
parents.
Most of these law reforms are too fresh to permit sensitive analyses of
longer term impacts of the presumption or practice. Consequently, it is too
soon to have many published evaluations. The Arizona presumptive law,
however, had a recent cursory evaluation that is summarized in Fabricius,
Aaron, Akins, Assini, and McElroy (2018). It found that the law appears to be
having a positive effect and is in the child’s best interests.
The country with the most mature law and practice as well as rigorous
recent evaluations is Sweden. The articles in this issue by Nielsen and by
Bergstrom summarizing the Swedish research indicate both that the arrange-
ment has become a “new norm”and that children who spent equal time
living with both parents after a separation reported better well-being than
children in predominantly single parent care.
As noted, the move toward making JPC the substantially normative option
is very recent. Thus it is premature to expect a plethora of these types of well-
designed studies assessing what happens when large swaths of couples, which
include the many couples where at least one of the parents is unenthusiastic
about the arrangement, have JPC imposed on them because of legal reform.
Scholars, advocates, and decision makers should be very alert for when
evaluations of these situations emerge and become part of the literature. It is
noteworthy, though, that virtually all the studies to date support the proposi-
tion that JPC is in children’s best interests even when one parent opposes it.
Conclusion
The central question posed by this article is whether JPC causes better outcomes
for children, and to describe those research designs that can better help us
answer this question. It is difficult to draw causal conclusions from older
research in this area because the studies use primarily static group comparison
research designs with self-selection into comparison groups, which confounds
the causal question. Because a random assignment experiment is unlikely to ever
occur, it is a certainty that such causality will never be answered conclusively.
However, several other approaches are beginning to be employed with more
frequency that can probe causality. Some recent studies exploiting such analyses
have already been reported, and others should be expected in the near future.
The weight of the recent evidence indicates that self-selection effects do
not largely account for the benefits of JPC in the empirical literature. Over a
wide variety of methodological approaches and for the vast majority of
findings to date, it appears that the benefits of JPC for children are not
primarily due to the fact that a unique set of families choose it. Thus,
evidence from recent research is discrediting the major rival explanation—
that the better child outcomes observed in JPC are merely the result of self-
12 S. L. BRAVER AND A. M. VOTRUBA
selection. Infirming the primary alternative explanation has the compensa-
tory effect of supporting the original causal proposition (Cook & Campbell,
1979). Thus, we conclude that JPC probably does cause benefits to children
on average. It should go without saying that the final two words in the
preceding sentence are absolutely necessary. Although the general tendency
across all individuals merits this conclusion, it certainly might not apply to all
individual child custody cases. However, whether we currently have the
requisite expertise to permit inferences about the likely impact in any parti-
cular case is debatable (Emery, Otto, & O’Donahue, 2005; Kelly & Ramsey,
2009; Stevenson, Braver, Ellman, & Votruba, 2012). “Bottom line: much as it
may be desirable, we may really not know how to properly individualize,
tailor, or custom-fit parenting plans to achieve the best possible outcomes in
each case. If this is true, the effort and expense and time and trouble taken in
the futile pursuit of case-specific fittings come with little in the way of
corresponding benefits. And, in such a case, it is better to have a rule or
starting place that covers the majority of cases and families, with, of course,
the ability to deviate when the fit is obviously bad”(Braver, 2014, p. 177).
Similarly, with the recent increased use of methodologically advanced
research designs, we regard the evidence to now be sufficiently deep and
consistent to permit social scientists to provisionally recommend presump-
tive JPC to policymakers. As always, the presumption should be rebuttable;
that is, although on average JPC can now be confidently predicted to bestow
benefits on children, there are certainly situations where JPC would be
unwise. Researchers can assist the enterprise of identifying these exceptions
by engaging in systematic efforts to identify subgroups for whom the usual
conclusion does not fit. One way to do this is to investigate interaction effects
(e.g., custody arrangement by conflict interactions) on the child outcomes.
The term provisionally is used here, because we hope and expect research-
ers will keep studying the matter, especially with rigorous analyses of the type
identified in this article. Consumers of this research also need to be alert to
new findings that continue to affirm the conclusions here—or perhaps that
oppose it. We might aptly characterize the current state of the evidence as
“the preponderance of the evidence,”meaning that there is substantially
more evidence for the presumption than against it. A great many studies,
with various inferential strengths, suggest that JPC will bestow benefits on
children on average, and few if any studies show that it instead harms them.
We note a kind of personal natural before and after experiment in this
regard. About 20 years ago, the first author wrote, “There is simply not
enough evidence available at present to substantiate routinely imposing joint
residential custody . . . there are too few cases adopting [it] to perform
statistical analyses”(Braver & O’Connell, 1998, p. 223). That was before. A
large number of those studies have since been performed, and the state of the
newer evidence is almost completely supportive. On this basis, we contend
JOURNAL OF DIVORCE & REMARRIAGE 13
the burden of persuasion has shifted to those who oppose a presumption
of JPC.
Note
1. There is also, of course, a very substantial literature that opposes shared parenting
presumptions when domestic violence is evident or alleged (e.g., Greenberg, 2004;
Morrill, Dai, Dunn, Sung, & Smith, 2005). Although these voices are persuasive, in
general, the articles provide arguments, not quantitative empirical research findings.
Because this article is devoted to research design issues within the quantitative empirical
research literature, papers presenting arguments only are outside the scope of this article.
In any event, proposed statutes often explicitly note that the existence of chronic, one-
sided domestic violence should be a rebuttal factor. There are also voices that oppose
shared parenting when there is high interparental conflict. For example, Stahl (1999), in
his guide for professional custody evaluators, opined, “high conflict parents cannot share
parenting”(p. 99). Similarly, Buchanan (2001) wrote, “when parents remain in high
conflict, joint custody is . . . ill-advised”(p. 234). Emery (2009)wrote,“joint physical
custody is the worst arrangement for children when [it] leaves [them] in the middle of a
war zone. .. . In high conflict divorces, children do worse in joint physical custody than
in other arrangements.”Such claims are supposedly based on the quantitative empirical
literature and therefore are included in our review here.
References
Ballard, R., Holtzworth-Munroe, A., Applegate, A., D’Onofrio, B., & Bates, J. (2013). A
randomized controlled trial of child-informed mediation. Psychology, Public Policy, and
Law,19, 271–281. doi:10.1037/a0033274
Baude, A., Pearson, J., & Drapeau, S. (2016). Child adjustment in joint physical custody
versus sole custody: A meta-analytic review. Journal of Divorce & Remarriage,57, 338–360.
doi:10.1080/10502556.2016.1185203
Bauserman, R. (2002). Child adjustment in joint-custody versus sole-custody arrangements: A
meta-analytic review. Journal of Family Psychology,16,91–102. doi:10.1037/0893-3200.16.1.91
Beck, C., Holtzworth-Munroe, A., D’Onofrio, B., Fee, H., William, C., Hill, H., & Frances, G.
(2009). Collaboration between judges and social science researchers in family law. Family
Court Review,47, 451–467. doi:10.1111/j.1744-1617.2009.01267.x
Braver, S. (2014). Costs and pitfalls of individualizing decisions and incentivizing conflict: A
comment on AFCC’s think tank report on shared parenting. Family Court Review,52, 175–
180. doi:10.1111/fcre.12079
Braver, S., Cookston, J., & Cohen, B. (2002). Experiences of family law attorneys with current
issues in divorce practice. Family Relations,51,325–334. doi:10.1111/j.1741-3729.2002.00325.x
Braver, S. L., & Lamb, M. E. (2012). Marital dissolution. In G. W. Peterson & K. R. Bush
(Eds.), Handbook of marriage and the family (3rd ed., pp. 487–516). New York, NY:
Springer.
Braver, S. L., & O’Connell, D. (1998). Divorced dads: Shattering the myths. New York, NY:
Tarcher/Putman.
Braver, S. L., Sandler, I. N., Hita, L. C., & Wheeler, L. A. (2016). A randomized comparison
trial of two court-connected programs for high conflict families. Family Court Review,54,
349–363. doi:10.1111/fcre.12225
14 S. L. BRAVER AND A. M. VOTRUBA
Braver, S. L., Wolchik, S. A., Sandler, I. N., Sheets, V., Fogas, B., & Bay, R. C. (1993). A
longitudinal study of noncustodial parents: Parents without children. Journal of Family
Psychology,7,9–23. doi:10.1037/0893-3200.7.1.9
Brotsky, M., Steinman, S., & Zemmelman, S. (1988). Joint custody through mediation.
Conciliation Courts Review,26,53–58. doi:10.1111/j.174-1617.1988.tb01038.x
Buchanan, C. M.. (2001). Divorce. In J. V. Lerner, R. M. Lerner, & J. Finkelstein (Eds.),
Adolescence in America: An encyclopedia (pp. 232–235). Santa Barbara, CA: ABC-CLIO.
Campbell, D. T., & Stanley, J. C. (1963). Experimental and quasi-experimental designs for
research on teaching. Boston, MA: Houghton Mifflin.
Chase-Lansdale, P. L., Cherlin, A. J., & Kiernan, K. E. (1995). The long-term effects of
parental divorce on the mental health of young adults: A developmental perspective.
Child Development,66, 1614–1634. doi:10.2307/1131900
Cherlin, A. J., Furstenberg, F. F., Jr., Chase-Lansdale, L., Kiernan, K. E., Robins, P. K., Morrison,
D. R., & Teitler, J. O. (1991). Longitudinal studies of effects of divorce on children in Great
Britain and the United States. Science,252, 1386–1389. doi:10.1126/science.2047851
Cook, T. D., & Campbell, D. T. (1979). Quasi-experimentation: Design and analysis for field
settings. New York, NY: Rand-McNally.
Cook, T. D., Shadish, W. R., & Wong, V. C. (2008). Three conditions under which experi-
ments and observational studies produce comparable causal estimates: New findings from
within-study comparisons. Journal of Policy Analysis and Management,27, 724–750.
doi:10.1002/pam.v27:4
DeLusé, S., & Braver, S. L. (2015). A rigorous quasi-experimental evaluation of a mandatory
divorce education program. Family Court Review,53,66–78. doi:10.1111/fcre.12131
Douglas, E. M. (2003). The impact of a presumption for joint legal custody on father
involvement. Journal of Divorce & Remarriage,39(1–2), 1–10. doi:10.1300/J087v39n01_01
Elrod, L. H. (1990). The federalization of child support guidelines. Journal of the American
Academy of Matrimonial Law,6, 103–130.
Emery, R. (2009, May 18). Joint physical custody: Is joint physical custody best—or worst—for
children? [Web log post]. Retrieved from https://www.psychologytoday.com/blog/
divorced-children/200905/joint-physical-custody
Emery, R., Otto, R., & O’Donohue, W. (2005). A critical assessment of child custody
evaluations—Limited science and a flawed system. Psychological Science,6(1), 1–29.
Fabricius, W., Aaron, M., Akins, F. R., Assini, J. J., and McElroy, T. (2018). What happens
when there is presumptive 50/50 parenting time? An evaluation of Arizona’s New Child
Custody Statute. Journal of Divorce & Remarriage. doi:10.1080/10502556.2018.1454196
Fabricius, W. V., Braver, S. L., Diaz, P., & Velez, C. E. (2010). Custody and parenting time:
Links to family relationships and well-being after divorce. In M. E. Lamb (Ed.), Role of the
father in child development (5th ed., pp. 245–289). Hoboken, NJ: Wiley.
Fabricius, W. V., & Suh, G. W. (2017). Should infants and toddlers have frequent overnight
parenting time with fathers? The policy debate and new data. Psychology, Public Policy, and
Law,23,68–84. doi:10.1037/law0000108
Fehlberg, B., Smyth, B., Maclean, M., & Roberts, C. (2011). Legislating for shared time
parenting after separation: A research review. International Journal of Law, Policy and
the Family,25, 318–337. doi:10.1093/lawfam/ebr015
Greenberg, J. G. (2004). Domestic violence and the danger of joint custody presumptions.
Northern Illinois University Law Review,25, 403–515.
Gunnoe, M. L., & Braver, S. L. (2001). The effects of joint legal custody on mothers, fathers,
and children, controlling for factors that predispose a sole maternal vs. joint legal award.
Law and Human Behavior,25,25–43. doi:10.1023/A:1005687825155
JOURNAL OF DIVORCE & REMARRIAGE 15
Huitema, B. (2011). The analysis of covariance and alternatives: Statistical methods for
experiments, quasi-experiments, and single-case studies (2nd ed.). Hoboken, NJ: Wiley.
Kelly, R. F., & Ramsey, S. H. (2009). Child custody evaluations: The need for systems-level outcome
assessments. Family Court Review,47, 286–303. doi:10.1111/j.1744-1617.2009.01255.x
Leading Women for Shared Parenting. (2017). Twenty five states consider shared parenting
bills in 2017. Retrieved from http://lw4sp.org/blog/2017/2/24/twenty-five-states-consider-
shared-parenting-bills-in-2017
Luepnitz, D. (1986). A comparison of maternal, paternal and joint custody. Journal of
Divorce,9,1–12. doi:10.1300/J279v09n03_01
Maccoby, E. E., & Mnookin, R. H. (1992). Dividing the child: Social and legal dilemmas of
custody. Cambridge, MA: Harvard University Press.
Mauricio, A. M., Mazza, G. L., Berkel, C., Tein, J. Y., Sandler, I. N., Wolchik, S. A., &
Winslow, E. (2017). Attendance trajectory classes among divorced and separated mothers
and fathers in the new beginnings program. Prevention Science. Advance online publica-
tion. doi:10.1007/s11121-017-0783-3.
Messer, L. C. (2017). Natural experiment. In Encyclopaedia Britannica. Retrieved from
https://www.britannica.com/topic/natural-experiment
Mnookin, R. H., & Kornhauser, L. (1979). Bargaining in the shadow of the law: The case of
divorce. The Yale Law Journal,88, 950–997. doi:10.2307/795824
Morrill, A. C., Dai, J., Dunn, S., Sung, I., & Smith, K. (2005). Child custody and visitation
decisions when the father has perpetrated violence against the mother. Violence Against
Women,11, 1076–1107. doi:10.1177/1077801205278046
Nielsen, L. (2014). Shared physical custody: Summary of 40 studies on outcomes for children.
Journal of Divorce & Remarriage,55(8), 613–635.
Nielsen, L. (2015). Shared physical custody: Does it benefit children? Journal of the American
Academy of Matrimonial Lawyers,28,79–139.
Nielsen, L. (2017). Re-examining the research on parental conflict, coparenting, and custody
arrangements. Psychology, Public Policy, and Law,23, 211–231. doi:10.1037/law0000109
Nielsen, L. (2018). Joint versus sole physical custody: Outcomes for children independent of family
income or parental conflict. Journal of Child Custody.doi:10.1080/15379418.2017.1422414
Pearson, J., & Thoennes, N. (1990). Custody after divorce: Demographic and attitudinal
patterns. American Journal of Orthopsychiatry,60, 233–249. doi:10.1037/h0079166
Rebuttable Presumption Law and Legal Definition. (2017). USLegal. Retrieved from https://
definitions.uslegal.com/r/rebuttable-presumption/
Rossi, F. S., Holtzworth-Munroe, A., Applegate, A. G., Beck, C. J., Adams, J. M., & Hale, D. F.
(2015). Detection of intimate partner violence and recommendation for joint family
mediation: A randomized controlled trial of two screening measures. Psychology, Public
Policy, and Law,21, 239. doi:10.1037/law0000043
Sandler, I. N., Wolchik, S. A., Berkel, C., Jones, S., Mauricio, A. M., Tein, J.-Y., & Winslow, E.
(2016). Effectiveness trial of the New Beginnings Program (NBP) for divorcing and
separating parents: Translation from and experimental prototype to an evidence-based
community service. In M. Israelashvili & J. L. Romano (Eds.), Cambridge handbook of
international prevention science (pp. 81-106). Cambridge, UK: Cambridge University Press.
Shadish, W. R., Clark, M. H., & Steiner, P. M. (2008). Can nonrandomized experiments yield
accurate answers? A randomized experiment comparing random and nonrandom assign-
ments. Journal of the American Statistical Association,103, 1334–1344. doi:10.1198/
016214508000000733
Shadish, W. R., Cook, T. D., & Campbell, D. T. (2002). Experimental and quasi-experimental
designs for generalized causal inference. Boston, MA: Houghton Mifflin.
Stahl, P. M. (1999). Complex issues in child custody evaluations. New York: Sage.
16 S. L. BRAVER AND A. M. VOTRUBA
Stark, P. B. (2017). Control for a variable. In Berkeley glossary of statistical terms. Retrieved
from https://www.stat.berkeley.edu/~stark/SticiGui/Text/gloss.htm
Stevenson, M. M., Braver, S. L., Ellman, I. M., & Votruba, A. M. (2012). Fathers, divorce and
custody. In N. C. Cabrera & C. S. Tamis-LeMonda (Eds.), Handbook of father involvement:
Multidisciplinary perspectives (2nd ed., pp. 379–396). New York, NY: Psychology Press.
West, S. G., Cham, H., & Liu, Y. (2014). Causal inference and generalization in field settings:
Experimental and quasi-experimental designs. In H. T. Reis & C. M. Judd (Eds.),
Handbook of research methods in social and personality psychology (2nd ed., pp. 49–80).
New York, NY: Cambridge University Press.
West,S.G.,Cham,H.,Thoemmes,F.,Renneberg,B.,Schulze,J.,&Weiler,M.(2014). Propensity
score as a basis for equating groups: Basic principles and application in clinical treatment outcome
research. Journal of Consulting & Clinical Psychology,82, 906–919. doi:10.1037/a0036387
Winslow, E. B., Braver, S., Cialdini, R., Sandler, I., Betkowski, J., Tein, J. Y., . . . Lopez, M.
(2017). Video-based approach to engaging parents into a preventive parenting intervention
for divorcing families: Results of a randomized controlled trial. Prevention Science.
Advance online publication.
JOURNAL OF DIVORCE & REMARRIAGE 17
