Judicial decision-making in contested custody cases


Material Information

Judicial decision-making in contested custody cases the influence of sociological, psychological, and legal variables
Alternate title:
Influence of sociological, psychological, and legal variables
Physical Description:
v, 115 leaves : ill. ; 29 cm.
Sorensen, Erik D., 1963-
Publication Date:


Subjects / Keywords:
Research   ( mesh )
Decision Making   ( mesh )
Jurisprudence   ( mesh )
Child Custody   ( mesh )
Child Abuse   ( mesh )
Department of Clinical and Health Psychology thesis Ph.D   ( mesh )
Dissertations, Academic -- College of Health Related Professions -- Department of Clinical and Health Psychology -- UF   ( mesh )
bibliography   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph.D.)--University of Florida, 1990.
Bibliography: leaves 102-107.
Statement of Responsibility:
by Erik D. Sorensen.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 002357959
oclc - 50908084
notis - ALW2449
System ID:

This item is only available as the following downloads:

Full Text








I would like to express my appreciation and affection

for my chairperson, Jacquelin Goldman, Ph.D., whose

knowledge and concern was essential both in this project and

in my broader development as a psychologist. I would also

like to express my appreciation to my committee members,

James Johnson, Ph.D., Sheila Eyberg, Ph.D., Steve Boggs,

Ph.D., and Steve Willis, J.D., each of whom has provided me

with essential knowledge and support. Additionally, I would

like to thank the Guardians for their dedicated efforts at

collecting the data for this project. In particular, I

would like to acknowledge the directors and coordinators of

the Guardian Ad Litem programs who were instrumental in

generating support and interest among their staff. Finally,

many thanks to my wonderful coworkers, Ilana Interrante,

M.S., Martin Ward, Linda Graves, and Cindy Sutton, whose

efforts were truly indispensable.





Overview . . .
Types of Custody. . .
Historical Background . .
Problems with the Use of Rules in
Determining Custody. ... .
Joint Custody .. . .
Legal Guidelines for Custody Determinations .
Judicial Decision-Making. ... .
Development of Children Following Divorce .

Parental Characteristics. .
Parent-Child Relationships. .
Specific Aims . .
Hypotheses. . .


Subjects. . .
Measures . .
Procedure . .
Analyses. . .
Statistical Procedures .
Variable Selection .


Case Characteristics. . .
Frequency of Custody Awards and Recommendations
Prediction of Custody Awards. . .
Awards of Shared Versus Sole Custody .
Awards of Maternal Versus Paternal
Primary Physical Residence .
Guardians' Decisions to Report Ratings. .






S ii


S 1

S 1








. .
. .
* .
* .

Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy



Erik D. Sorensen

December 1990

Chairperson: Jacquelin Goldman, Ph.D.
Major Department: Clinical and Health Psychology

In contested custody cases judges are responsible for

determining the custody award that serves the best interests

of the children involved. While state statutes provide

judges with guidelines regarding these determinations, they

are not worded very specifically. The judge's role is to

interpret these guidelines and apply them to individual

cases. Previous literature concerning judicial decision-

making has suggested that judges often rely on a general

preference for the mother as custodian rather than on

individual case characteristics. This study examined the

relationship between specific case factors and custody

awards in order to describe the decision-making process in

Florida. Information was collected on 60 contested initial

disposition or disposition modification cases and was used

to model judicial decision-making. Our results indicated

that judges were generally following the legal statutes and

did not rely on a general preference for one parent.

Additionally, although several case factors were used

consistently by judges, some notably important information

did not appear to be given much weight in their decision-

making. In some cases substantiated reports of child

abuse/neglect were overlooked. Additionally, the degree of

parental conflict did not appear to be considered by judges,

although this variable has been identified in the literature

as a strong predictor of post-divorce child adjustment.



Before 1900, determinations of responsibility for

children after divorce were most often based on socially

defined parental roles. In other words, the parent socially

deemed responsible for raising children was given custody

unless declared unfit by the courts. Specific criteria that

influence child development were not considered in the

initial adjudication of custody. The needs of children were

expected to be met most effectively by the parent considered

socially responsible for child rearing. However, as

researchers in child development, divorce, and families

demonstrated the importance of specific variables in normal

child development, the legislature and courts have

responded. Guidelines used by judges in adjudicating

custody have become more detailed and the general doctrine

has shifted from following social definitions of parental

roles to a direct consideration of "the best interests of

the child." Although criteria used to determine custody

have become much more clearly delineated than ever before,

judges may employ any additional information that they deem

important to make a determination in a given case. The

purpose of this research is to identify the criteria

utilized by judges in contested custody cases. Previous

research on judicial decision-making in custody cases have

studied judges' reports of what they do, but have not used

other sources of data. The present paper will study the

decision-making process from court records and ratings made

by other court personnel (Guardian Ad Litem

representatives). This methodology will provide additional

estimations of the importance of specific criteria in

custody decision-making than previously have been available.

This approach also will make possible a comparison between

what judges report is happening, what the Florida statutes

dictate should be happening, what literature on post divorce

child adjustment suggest is most important for these

children, and what occurs in custody cases. A review of the

types of custody available to most judges and a summary of

how these custody arrangements have been applied

historically will provide useful background information to

the reader.

Types of Custody

Prior to 1957 legal statutes stipulated that only one

type of custody arrangement was available to families: one

parent held complete legal control in all areas of the

children's lives, except for the court's determination of

visitation rights and financial responsibilities. This

arrangement was termed sole parental responsibility. The

lack of other custody alternatives presumably followed from

an assumption that divorced parents could not work well

enough together to make joint decisions about their

children. However, since social-science research has

documented the importance of a child's contact with both

parents following divorce, the courts have developed

alternative custody arrangements (Barnard & Jenson, 1984;

Felner, 1987).

North Carolina was the first state to allow an

alternative arrangement for custody in 1957, which was

termed joint custody or shared parenting. Under joint

custody parents are given approximately equal legal

authority for making decisions about the children, though

the court may determine which parent is expressly

responsible for specific needs of the children such as

school issues (Felner, 1987). In fact, usually one parent

is given primary physical custody under a joint custody

arrangement to maintain stability and continuity in the

child's environment. However, some states have made another

option available to the courts where the parents not only

share the legal responsibility of the children, but also the

residential care of them. Since this custody arrangement

includes all of the characteristics of joint custody, states

that allow this option make a distinction between joint

legal and joint physical custody, with the latter referring

to the arrangement in which the child moves between homes

(Felner, 1987). A few states have also established

legislation for a fourth type of custody arrangement, termed

split custody. Under this arrangement, each parent is

awarded sole custody of one or more child following the

divorce. This arrangement is relatively uncommon.

Historical Background

Prior to the eighteenth century, sole custody was the

only option available, and it was awarded to the father

universally. In fact, the father held complete legal

control over the family. He owned all of the family goods,

as well as the children, and completely determined what

happened to children following a divorce. Furthermore,

women held no rights over their husbands; they could neither

question the father's ability as a parent nor exert any

influence over him. The father held an unquestionable right

to decide all facets of his child's life, as children were

considered the property of the father (Derdeyn, 1976).

By the early eighteen hundreds, fathers became

responsible for their children's welfare. Children were

awarded rights as individuals for the first time in history.

The English Parliament established this change in the social

status of children in 1817 by instituting the doctrine of

parents patriae which held that the Crown should defend the

rights of those who had no other protection (Derdeyn, 1976).

Children were considered part of this group, as they could

not defend themselves against their parents. Custody

decisions were said to be based on the "best interests for

the child" rather than on the right of a father to dictate

his children's lives.

In the early 1800s, the social status of women also

increased with regard to their role and rights in the

family. By the 1850s, a mother's role in the family,

separate from that of serving her husband, began to be seen

as important (Derdeyn, 1976). In fact, the mother's care

was seen as necessary for the development of children

through the early years of life (Folberg, 1984b). This

change in social roles and status was reflected in the

courts, as judges began to refer to the assumption that

young children needed a mother's care to facilitate their

growth and development (termed tender years presumption)

when making custody determinations (Derdeyn, 1976). Under

this presumption, custody awards were given to mothers more

often than before, though still less often than to fathers.

While, initially the period in a child's life believed

to require maternal care included only the first four years

of life, the period was soon extended. By the 1920s,

mothers were awarded custody almost as often as fathers, and

by the mid 1900s, the tender years presumption had gained

sufficient popularity that mothers received custody an

estimated 90% of the time (Derdeyn, 1976). A few

researchers have suggested that the courts' support of the

mother as primary custodian of children following divorce

reflected that her role was considered fundamental not only

to the care of the children but also to the maintenance of

stability and continuity in her children's interpersonal

relationships and their physical environment for children

(Folberg, 1984b). Awarding primary custody to mothers in

order to maintain continuity for the child is consistent

with the fact that mothers were the primary care-givers of

children within most families prior to their divorce. A

preference for the mother as custodian may be, at least

partially, a reflection of the importance of case specific

variables and not entirely a simple presumption for the


The common social view of each parent's role in the

family and the consistency with which actual family role

structures adhere to the social view, though shifting from

supporting as custodian first the father and more recently

the mother, continued almost universally to dictate custody

decisions through the mid 1900s. The only exception to this

policy occurred when the parent assumed to be most important

to the child was considered unfit as a guardian. If a

parent was guilty of adultery or desertion, for example,

then custody would be given to the other spouse regardless

of the strength of the social preference for that parent

(Charnas, 1981; Derdeyn, 1976). In addition to moral

infractions, mental illness also served as a reason for

declaring a parent unfit for shared or sole parental

responsibility (Horowitz & Davidson, 1984). The

adjudication of custody appeared to be based upon rules

which assumed the overriding importance of one parent for

the children's growth and development, often without even a

consideration of the other's abilities (Mnookin, 1975).

Problems with the Use of Rules in Determining Custody

Using a general rule that considers one parent a more

worthy custodian of children and allowing for only one

custody arrangement (sole custody) presents some problems

when a parent's desired post-divorce role structure does not

adhere to the traditional social role definitions. For

example, in traditional western post-divorce families, men

are expected to earn the money to support two households,

often leaving them with little time for their children; if

they choose to spend time with their children at the expense

of career advancement, they must respond to accusations of

avoiding child-support payments (Victor & Winkler, 1976).

Similarly, mothers are expected to fight for custody of

their children; if a mother wants to begin or pursue a

career that requires more than a few hours a day, she must

stand up to challenges against her love for her children

(Victor & Winkler, 1976). Consequently, the

socially-defined parental roles and the use of sole custody

as the only option can, at least partially, dictate parents'

involvement with their children.

The primary reliance of the courts on sole custody

alone and on a general rule to determine which parent

receives custody also implies limits of factors that are

considered important for custody determinations (Victor &

Winkler, 1976). Factors important in individual cases could

be overlooked, if they do not support the traditional

caretaker as sole custodian. It may be that each parent's

relationship with the children, which could be different

from traditional parental roles, could provide a unique

contribution to the child's development. Likewise, the

possibilities available to the family through the parents'

combined resourcefulness or special circumstances might make

a custody arrangement other than sole custody more

beneficial for the children. Recent research has, in fact,

consistently indicated the importance of many specific

variables that describe the influence of intrafamilial

relationships on the development and adjustment of children

following divorce (e.g., Santrock & Warshak, 1979).

Additionally, unique benefits have been shown to arise from

the individual relationships of children with each parent

following divorce, suggesting the importance of both

parents' continued involvement with the children over a sole

custody arrangement. These findings will be discussed in

detail later.

Another problem arising from the use of traditional

social role definitions to adjudicate custody and a reliance

on sole custody as the only option for custody follows from

the court process itself. Parents must engage in

adversarial relationships with each other if one or both of

them wishes to challenge the court's established view of the

family's post-divorce roles and the concomitant

specifications of visitation and custody rights (Fischer,

1983; Victor & Winkler, 1976). With custody determinations

being based primarily on a general rule, the only issues

that can result in a shift from awarding custody to the one

expected primary caretaker include moral and psychological

infractions that render that parent unfit. The interaction

between parents in the court room has become, in effect, a

battle over father's and mother's moral and psychological

fitness, rather than a direct defense of the best interests

of the child (Goldstein, Freud, & Solnit, 1973; Pringle,


Within this context, answers to the court's question of

what is the best interest of a child have been centered

around individual parental roles rather than the

consideration of each parent's unique contribution to the

child's development or the possibilities available to the

family through the parents' combined resourcefulness.

Wallerstein and Kelly (1980) described the effects of using

simple rules that follow role expectations in determining

the primary caretaker for children. For their study, 60

cases where the mothers were awarded sole custody were

followed for five years after divorce. One effect was that

three-fifths of the women had a substantial decline in their

standard of living. Additionally, their periodic income

also became increasingly more variable. Because mothers

were forced to either stay at home with the children or

resort to uninteresting jobs that demanded little time, they

were unable to develop resources to stabilize the family


These researchers also found that many older children

living with their mothers rejected their fathers, accusing

them of desertion when in fact the separation and

post-divorce relationship was decided upon either equally by

both parents or by the courts. Furthermore, fathers were

allowed little share in decisions concerning the development

of their children. These results suggest that limited

custody options and decision-making based on traditional

roles affect the parents' involvement with their children

after divorce. They may even force the family into

situations of conflict and frustration rather than helping

them to resolve problems. A preference for the mother or

father as a single custodian is not supported by the above


Joint Custody

Shared parental responsibility, where both parents

participate in their child's development, has been offered

in response to these problems. Under this arrangement,

there is greater flexibility with regards to individual

parent's responsibility for and his or her availability to

the children. Ideally, joint custody allows time for

mothers to pursue their careers and fathers to develop

relationships with their children (D'Andrea, A., 1983;

Derdeyn & Scott, 1984; Rothberg, 1983; Steinman, 1981;

Victor & Winkler, 1976). Rosen (1979) and Frankel (1985)

also found that when the child was allowed free access to

both parents, divorce was perceived by children as less


Third, shared parental responsibility offers the courts

an option that does not force parents to do battle over

their children, avoiding the usual problems of the legal

system's adversarial process. For example, Ilfeld, Ilfeld,

and Alexander (1982) and Frankel (1985) found that joint

custody with open visitation arrangements resulted in less

conflict within the family as assessed by the low percentage

of cases involved in relitigation relative to sole custody

cases. Frankel further found that when joint custody

families did require modifications in their arrangements,

they were effected with little conflict.

Fourth, as fathers have been taking on more of the

daily responsibilities of child care within intact families

than was previously common, concerns about the continuity

and stability of the child's environment and interpersonal

relationships no longer support simply an award for the

mother as primary caretaker. In fact, Kline, Tschann,

Johnston, & Wallerstein (1989) found that more joint custody

fathers continued to have frequent regular contact with

their children than noncustodial fathers. Fifth, Steinman

(1981) found that joint custody can affect the parent-child

relationships and interparent conflict. Joint custody

parents held much more respect for each other in terms of

their parenting skills and applied a lot of effort towards

working together in dealing with their parental concerns.

These issues suggest that joint custody may offer a solution

to many of the problems arising from using social role

definitions for determining custody. However, Ilfeld et al.

(1982) and Frankel (1985) stated that their results might

have been due to a sample bias in that joint custody may

have been awarded primarily to families with less overall

conflict. This limitation, in fact, probably is present for

a large number of the studies on post-divorce family

relationships as samples have consisted of primarily middle-

class families who are recruited through advertisements.

Despite its benefits, joint custody also requires

several concessions on the part of the parents before it can

be used effectively (Derdeyn & Scott, 1984). Parents must

be willing to decide together what is best for their

children and to live close enough to one another so that

they can both maintain frequent contact with the children.

Joint custody, then, is not the total solution to the

problems of custody adjudication.

In many cases where joint custody has been awarded,

parents may find it difficult to work out visitation rights

and economic responsibilities together. Even more

critically, when parental conflict is high, each episode of

visitation or decision-making may subject a child to

repeated psychological trauma. In some cases, parents may

resort to relitigation: at this time the courts re-evaluate

whether joint custody is in the best interests of the child

as well as addressing other specific issues.

Consequently, even under joint custody and with the

establishment of the "best interests of the child" doctrine,

the child's welfare depends upon sensitive judicial

decision-making. Examination of the factors which may best

support the post-divorce adjustment of children could have

important implications for the courts' decision-making

processes. The remainder of this paper compares current

legal guidelines for custody determinations in terms of

their congruence with empirical studies of children's

post-divorce adjustment.

Legal Guidelines for Custody Determinations

States presently provide some general criteria in

adjudicating custody disputes, but there are very few

specifically defined legal guidelines to aid judges in

decision-making, and even fewer that are known to influence

directly the future development of these children. In

Florida, the courts follow the "best interest of the child"

doctrine, and shared parental responsibility is assumed

unless determined detrimental to the child because of a

parent's unfitness (Joint Legislative Management Committee,

1985). A parent can be considered unfit as a caretaker due

to a past major felony conviction, spousal or child abuse,

psychological instability, or a history of not caring for

his or her children. If joint custody is rejected as the

custody arrangement, then one parent is given full

responsibility for raising the children after a divorce

through an award of sole custody.

To date in Florida nine factors have been identified by

the state as important in adjudicating custody disputes to

determine both the type of custody, and who will be the

primary residential caretaker if joint custody is chosen.

In section three, chapter 61.13 of the Florida Statutes

(Joint Legislative Management Committee, 1985) the

guidelines are as follows: (a) the parent who is more

likely to allow the child frequent and continuing contact

with the nonresidential parent; (b) the love, affection, and

other emotional ties existing between the parents and the

child; (c) the capacity and disposition of the parents to

provide the child with food, clothing, medical care or other

remedial care recognized and permitted under the laws of

this state in lieu of medical care, and other material

needs; (d) the length of time the child has lived in a

stable, satisfactory environment and the desirability of

maintaining continuity; (e) the permanence, as a family

unit, of the existing or proposed custodial home; (f) the

moral fitness of the parents; (g) the mental and physical

health of the parents; (h) the home, school, and community

record of the child; and (i) the reasonable preference of

the child, if the court deems the child to be of sufficient

intelligence, understanding, and experience to express a


While these criteria offer some guidelines on what

issues to consider, the standards for determining when a

problem exists under one criterion are still quite vague.

There remains the question of assessing when one of these

criteria has been sufficiently satisfied to declare a parent

unfit. Keenan (1985), for example, argues that when proof

of abuse is not present, a parent alleging the abuse may be

refused custody because of appearing uncooperative or

emotionally unstable to the court. The criteria established

by the state legislatures then do not provide clear guidance

in determining parental fitness. Furthermore, it even may

be argued that no support is offered regarding how to

evaluate those issues.

Similarly, when both parents are considered equally

fit, there remains the issue of judging which of the above

criteria take precedence over others if primary physical

residence is to be awarded. If one parent has predominantly

cared for the children during the marriage while the other

has provided for its economic needs, then a judge must

determine if economic capacity to care for the child is more

or less important for the best interests of the child then

maintaining the consistency in the care giving environment

of the child (Trudrung-Taylor, 1986). Again, a judge is not

given guidelines upon which an evaluation of state statute

guidelines can be based. Additionally, it can appear to

families that the judge is biased against one parental role

no matter which criteria is given the deciding weight

(Trudrung-Taylor, 1986).

A third issue relevant to custody determinations for

which few guidelines are offered centers on how to evaluate

when a previously awarded custody arrangement needs to be

modified. A modification of child custody or visitation

orders are generally justified only if there has been some

"substantial change" of circumstances since the initial

award of custody and only if the modification is in the best

interests of the child (Price, 1983). Changes in custody or

primary physical residence, for example, might be based on

such issues as voluntary relinquishment of custody or on

some event that proves a parent is unfit (e.g.,

substantiated physical abuse of the child). Visitation, on

the other hand, might be modified if a previous arrangement

has become unworkable or if one parent has moved (Price,


While these general standards provide some guidelines

upon which to base a modification, what a judge considers

sufficient to award a modification varies greatly from judge

to judge. In fact, a judge may not even identify what

factor was considered important in his determination, but

rather list all issues he considered in a given case without

indicating how they were used. Thus, a fact considered

important in one case might not be important at all in

another case (Price, 1983). Consequently, the discretion of

the judge remains paramount over any statement by state

statutes or court precedence in the decision-making process.

In addition to the lack of specificity of these

guidelines, there are some issues common to custody cases

for which no guidelines are offered. Judges are left with

only court precedence, which is often equivocal, in

considering these issues. One such issue, defined as

kidnapping, occurs when one parent takes and holds the

children so that the other parent cannot gain access to

them. While many people have described the traumatic effect

of this event (e.g., Green, 1985), judges may not interpret

kidnapping as sufficient grounds to declare a parent unfit.

Even when it is illegal, a judge may not consider this issue

when determining which parent will be awarded primary

physical residence (Green, 1985).

A similar issue for which no guidelines have been

offered, though its relevance to custody decision-making is

unquestioned, is when a parent motions the court to allow

him or her to move out of the state and take the children

along (Quinn, 1986). Allowing a parent to move presents

some problems for divorce families, as it effectively limits

the other parent's contact with the children. In Florida,

it is generally accepted according to court precedence that

the moving spouse must show that the move a) will provide

"substantial changes" to the family issues upon which an

initial custody arrangement has been based, and b) be in the

best interest of the child before it will be allowed (Quinn,

1986). While the above standard provides some accepted

guidelines for judges, the issue of what constitutes

"substantial change" remains sufficiently vague. In fact,

in very similar cases judges may decide to allow a move in

one case but not in another. For these issues where no

state statutes are available, judges are offered only

general precedence that may not provide any guidelines upon

which to base their determinations.

This lack of information and guidance forces judges to

continue to depend upon their own views of the family's

roles and structure to evaluate where the best interests of

the child lies. Though some guidelines and court

precedences are available, they do not provide sufficient

support upon which to base decisions of how to use those

guidelines. Furthermore, no information is provided on how

to combine different components of custody arrangements,

such as visitation, financial responsibilities, and child

care to deal with the characteristics of families.

Empirical studies examining the factors associated with

children's post-divorce adjustment could be used to aid in

the determination of custody and primary physical residence.

Research could offer some more specific information on how

the variables discussed previously relate to post-divorce

child adjustment, and thus present some empirical support

for what constitutes the best interest of the children.

Research could also suggest when specific custody

arrangements are clearly contraindicated, such as joint

custody when the parents are extremely hostile towards one

another. It would then be the task of the court to

determine how these findings would apply to specific cases.

The recent advent of joint custody and other new types

of custody as options for judges and the inclusion of some

specific criteria in state statutes suggests that there

already has been some progression in judicial

decision-making away from a simple bias for one parent.

These changes present a question to those who are searching

for reform in custody decision-making of what judges

currently use. Research on judicial decision-making is

discussed in the next session. What criteria are used by

judges currently will then be compared to research on

post-divorce child adjustment as the next step in helping to

evaluate where the best interest of the child lies.

Judicial Decision-Making

The new custody arrangements established by state

legislatures have been argued as more supportive of the best

interests of the child doctrine than sole custody awarded on

the basis of social role expectations. If judges have begun

to use more case specific variables than was previously

common, it can be expected that judges will rely less on a

general preference for the mother as primary custodian.

Pearson and Ring (1981) examined recent custody

determinations in three Colorado counties and found that

mothers received sole custody in 51.5% to 76.3% of the time,

while fathers had only a 15.5% to 32.3% possibility in being

awarded sole custody; split custody, where children are

separated between parents, was awarded between 7.2% to 11.2%

of the time; and joint and third party custody were the

least popular awarded 1.0% to 6.1% of the time. Pearson and

Ring concluded that the father only gained sole custody if

the mother was proven to be unfit due to either mental

problems, remarriage to a husband with a history of abusing

children, or excessive drinking. Furthermore, custody

determinations made use of the new custody arrangements in

only a small number of cases. These results were also

supported by Lowery (1981) and Phear, Beck, Hauser, Clark,

and Whitney (1984).

Felner, Terre, Farber, Primavera, & Bishop (1985) also

suggested that a preference for the mother continues to

exist. Their study interviewed 43 judges; the results of

their study revealed that 42% of the judges endorsed sole

maternal custody as the arrangement of choice when both

parents are considered approximately equally competent as

parents. Only 26% of the judges reported that they would

consider joint custody as an option of choice in these

cases, while none endorsed sole paternal custody as an

option that they considered. These findings suggest that

judges have continued to rely highly on the use of social

views of the family, as represented by the tender years

doctrine, to determine sole custody rather than examining

specific case variables.

Felner et al. (1985) found that one reason judges

continued to prefer maternal custody regarded their

estimation of each parent's motivation for obtaining

custody. Two primary motivators described by judges for

fathers included revenge (endorsed by 26% of the judges) and

use of the child as a bargaining tool in financial

arrangements (endorsed by 23% of the judges). In addition,

more fathers than mothers were seen as desiring sole custody

because of spousal incompetence. These results indicate

that fathers still were assumed to be less concerned for

their children then are mothers, except when she is unfit as

a custodian.

While the above studies appear to support the

conclusion that judges continue to rely on a general

preference for the mother as custodian, this may not be the

whole truth. Pearson & Paul (1984) found that custody

awards to mothers often were based upon the issue of the

psychological parent. Goldstein et al. (1973) defined the

psychological parent as the one to whom the child has bonded

and with whom the child has been most involved. Since

mothers were the primary caretakers of the children for the

families studied while the parents were married, the award

of custody that best maintained consistency in the child's

emotional environment would be to the mother. A preference

for the mother as custodian may then be supported by

reliance on a specific case variable, the parent who

attended most to the needs of children prior to the divorce.

A more recent study (Racusin, R., Albertini, R.,

Wishik, H., Schnurr, P, & Mayberry, J., 1989) also found

support for the conclusion that judges respond to specific

case factors when determining custody of children. Joint

versus sole custody was awarded significantly more often

when fathers petitioned for it, when mothers and fathers had

greater than 12 years of education, when fathers were over

25 years of age, and when no one in the family received

welfare. Unfortunately, the influence of other factors such

as interparent conflict, parenting skills, or the identity

of the psychological parent was not assessed in this study.

However, at least some case specific variables were

considered by judges.

Another reason to question the finding of general

preferences for one parent is that the above studies did not

include in their samples families where both parents are

fighting for the custody or primary physical residence of

their children. Pearson, Munson, and Thoennes (1984) report

that as few as 10-15% of the custody cases brought before

judges may actually involve disputes between the parents.

The large number of maternal sole custody awards may then

include in part a preference for this arrangement by the

divorcing families themselves and not be solely a reflection

of the views of judges.

Clearly, a decision to award a particular custody

arrangement can be made for a variety of reasons.

Consequently, simple frequency counts of custody or primary

physical residence awards to each parent are insufficient to

conclude what issue or issues influenced the court's

decision-making. Furthermore, families often determine

their own custody arrangements. Thus, the frequency of

specific custody awards may evidence common family patterns

as much as judicial preferences or values.

Judicial attitudes towards joint and paternal custody

may also be misrepresented because judges have been

considered as one homogeneous sample. In fact, there may be

a group of judges who consider more case specific variables

than other groups, or different clusters of variables. This

difference would not be evident in judges' attitudes towards

specific criteria if groups of judges are not compared.

Pearson and Ring (1981) compared younger to older judges and

found that while older judges continue to support a

preference for maternal sole custody, younger judges

focused more on parental motives and the relationships

between the parents and children. Consequently, the

preference for the mother described above may be an

overestimation of judges' views in general.

However, Sorensen and Goldman (1989) found that there

were no discriminable groups of judges based on importance

ratings of specific criteria. While there was a large range

of variation in the importance ratings of criteria, there

did not appear to be distinct groups of judges who placed

more weight on some criteria than other groups. While this

finding does not indicate if young judges differ from older

judges on the use of the tender years presumption, it does

suggest that there is equal variability among young and old

judges in their evaluation of specific criteria.

While the above studies show that some specific case

variables are being considered by judges, they do not

specifically identify the relative amount of weight placed

on these and other criteria by judges. In an effort to

accomplish this task, several researchers have surveyed

judges requesting that they rate importance level for

individual criteria. Though limited in number, these

studies have suggested a continued reliance on the mother as

primary care giver though some specific criteria have been

identified as important to judges in their decision-making.

In addition to the issue of the psychological parent,

Pearson and Paul (1984) also found that children's wishes,

Guardian Ad Litem personnel recommendations, court ordered

investigations, and statements made by the parents

concerning their reason for seeking sole custody were used

in determining custody arrangements. And, though some

judges considered the input of mental health professionals,

determination of sole custody was not necessarily based on

following their recommendations.

Felner et al. (1985), Settle and Lowery (1982),

Sorensen and Goldman (1989), and Lowery (1981) all found

that specific factors were used in determining physical

custody without a strong preference for the mother.

Specific criteria in these studies that were most important

to judges in their decision-making included the emotional

stability of the parents, their ability to and concern for

providing care for the children, the amount of time they had

available for the children, the stability of the child's

living arrangements, the moral character of the parents, and

the parents affection for the children. Several other

factors were also identified as important, though somewhat

less important than the above issues. These factors

included the financial resources of the family, interparent

conflict, biological relationship of the parents and

children, the length of time each parent has had custody,

the parents' ability to provide contact with other

relatives, school, and other children of about the same age,

the child's wishes, and placing the child with the same sex

parent. Thus, judges do not appear to rely on a general

rule such as the tender years presumption.

Another factor that may be considered important to

judges is the effect of mental health experts' evaluations

of parents on custody determinations. Lowery (1981) found

that the level of importance rated by judges for their

consideration of psychologists' and psychiatrists'

evaluations of parents was high, averaging 7.56 on a scale

of 1 to 11. However, this finding was not supported by

Felner et al. (1985), as only 7% of the 43 judges sampled

considered mental health recommendations as important in

decision-making. Ash & Guyer (1984) also found that while

judges may utilize psychological evaluations of parents and

children, they considered other criteria more important than

mental health experts' evaluations.

The studies listed above have consistently found that

judges as a whole rate some criteria as having more

importance in their determination of custody than other

items. Thus, there is some general agreement among judges

about the relative importance of different groups of

criteria. It is important to note, however, that no

criteria have been identified yet that either override all

other criteria or are completely disregarded (Felner et al.,

1985; Lowery, 1981; Settle and Lowery, 1982; Sorensen &

Goldman, 1989). In fact, many items are rated as having

similar degrees of importance, suggesting that judges

consider a large number of issues in their decision-making

and are influenced by many case specific issues.

Although there were identifiable differences between

items in the amount of importance given to them by judges as

a whole, ratings also evidenced marked variability between

judges in the weight given to individual variables. Judges

did not make use of the criteria in a highly consistent

fashion (Sorensen & Goldman, 1989). Thus, judges continue

to differ in their evaluation of the relative importance of

specific criteria and to rely on their own evaluations when

weighing the many custody issues. These results suggest

that though a number of specific criteria likely to be used

by judges can be identified, how each of these items will

impact the decision-making process can not be predetermined.

The large variability in the importance ratings made by

judges make it difficult to describe judicial

decision-making in general.

The difficulty in describing generally judicial

decision-making may follow from several reasons other than a

simple difference between judges. One large reason is

simply that decision-making differs from state to state and

the studies that examine this process assess only one state

at a time. How decision-making changes from state to state,

and even the specific characteristics of decision-making

within a state remain unclear. Racusin, Albertini, Wishik,

Schnurr, and Mayberry (1989) did find that custody awards

differed among two states in accordance with the predominant

legal presumption of each state. Additionally, since there

were no studies on judicial decision-making in custody cases

prior to the advent of joint custody, it is impossible to

describe what changes have occurred since that time.

Consequently, a ranking of their relative importance in

decision-making in general may be impossible.

It is possible, however, to say that the courts and

legislature have responded to issues raised by legal and

other professionals regarding the importance of specific

case factors. For example, joint custody has been offered

as a custody arrangement that allows both parents to be

involved in the growth and development of their children

following a divorce. Similarly, some specific criteria have

been defined in state statutes, and clearly judges are

making use of these guidelines in their evaluation of

custody, at least by their own report. These events support

the conclusion of a continued concern about children and a

willingness to respond to information regarding what

constitutes the best interests of the child.

After identifying some of the criteria that judges use

in determining what constitutes the best interest of the

children following a divorce, it is important to recognize

how consistent these criteria are with literature on

post-divorce child adjustment. A comparison between these

two bodies of literature may help further delineate where

the best interests of the child lie.

Development of Children Following Divorce

To date, research on the factors that influence the

development of children after divorce has followed several

paths, the most common of which is the comparison of

divorced to intact families (Glover, 1989; Pfeffer, 1981;

Pringle, 1975; Roman & Haddad, 1978; Rothberg, B., 1983;

Shybunko, 1989; Wallerstein & Kelly, 1980; Warshak &

Santrock, 1983). The specific issues examined by these

studies have included the differences between parent-child

interactions, parent-to-parent interactions, school

achievement for the children, social adjustment, and

demographic differences between the two groups. Other

studies examining post-divorce child adjustment have also

assessed the influences of external support structures, life

events, relations between children and each of the parents,

parental style, parenting skills, working versus nonworking

parents, amount of conflict between the parents, and the

amount of time parents spend with their children (Hess &

Camara, 1979; Isaacs, 1988; Kline et al., 1989; Kurdek,

1988; Santrock & Warshak, 1979; Wallerstein & Kelly, 1980;

Walsh & Stolberg, 1989). Research has also examined the

differences between families in which mothers as compared to

fathers are the post-divorce single parent (Adams, Milner, &

Schrepf, 1984; Lamb, Pleck, & Levine, 1985).

It must be noted, however, that for most of the issues

listed above the vast majority of families studied have

included only a maternal sole custody arrangement, and

interviews have been the primary method of data collection.

Additionally, families studied are often middle class, and

their custody arrangements have been constructed and agreed

upon by both parents. These limitations in the research

make the evaluation of paternal care and its effect on

children difficult, and a clear statement of how the

elements of family structure result in stable growth for

children after divorce remains impossible. However, some

more recent research has involved the use of standardized

measures in the evaluation of child adjustment (Glover,

1989; Kline et al., 1989; Shybunko, 1989; Walsh & Stolberg,

1989). These studies have examined the effect of many

family and situational variables on the emotional adjustment

of children. While the more recent studies provide further

support for the validity of the findings on post-divorce

child adjustment, the following literature must be

interpreted with some caution.

Parental Characteristics

Felner's review of the literature on children of

divorced families (1987) and more recent research has

supported the conclusion that parental variables are the

greatest predictor of child development and general

emotional adjustment following a divorce. In particular,

the most salient predictors of general child adjustment

identified have included interparent conflict, the quality

of the parent-child relationship, the degree of instability

in the child's daily life, the emotional well being of the

parents, the level of parenting skill, and the degree of

economic stress present.

Of these factors, parental conflict emerged as a very

strong predictor of child adjustment, even two years after

parental divorce (Felner, 1987; Kline et al., 1989; Kurdek,

1988). However, hostility tends to decrease over time, at

least for low conflict couples, and thus becomes a less

effective predictor of child adjustment by five years post-

divorce (Walsh & Stolberg, 1989). Part of the impact of

interparental hostility on child adjustment may be effected

through its negative influence on caretakers' parenting

skills (Walsh & Stolberg, 1989). The relationship between

parental hostility and child adjustment has also been found

outside of divorced families. Stolberg, Camplair, Currier,

& Wells (1987) found that marital hostility was highly

related the number and severity of internalizing and

externalizing behavior problems, lower self-esteem and lower

prosocial behavior. Parental hostility appears to be a very

salient factor in predicting child adjustment generally,

further supporting the importance of examining this factor

when dealing with divorcing families.

Another important predictor of child development

includes the style of parenting within a family. Several

researchers found that authoritative (as opposed to

authoritarian) interaction which allows for verbal give and

take, warmth, and nonpunitive and consistent rule

enforcement were positively correlated with healthy peer

interactions, decreased aggression, improved relations with

parents, and improved attitudes towards school (Hess &

Camara, 1979; Santrock & Warshak, 1979; Wallerstein & Kelly,

1980; and Warshak and Santrock, 1983). Similarly, parenting

skills have also been found to be influential in post-

divorce child adjustment (Walsh & Stolberg, 1989). These

findings remained consistent regardless of which parent held

sole custody.

Another consistent and important predictor of post-

divorce child adjustment is maternal emotional adjustment.

(The effect of paternal emotional adjustment has not been

evaluated in part due to the infrequency of this

arrangement.) Kurdek (1988) and Kline et al. (1989) both

found a strong relationship between child behavior problems

and maternal mental health at one- and two-year follow-ups

respectively. Because parental mental health can impact

many aspects of the parent-child relationship, its influence

on child adjustment has probably been evident in a variety

of other studies that examine parent-child relationships.

In any case, the potential impact of this variable on child

adjustment is obvious and should thus be measured directly.

Social support or isolation has been studied

extensively in the literature regarding family problems. In

a review of the literature, Wood (1984) and Mitchell and

Trickett (1980) reported that symptomatology and self-esteem

are related to social support and social networks, and that

interpersonal conflict can mediate whether experiences are

supportive or stress inducing. These researchers suggest

that social support and social networks may be important

variables in mediating the impact of stressful experiences,

of which parental divorce is clearly one. The positive

effect of parental social support on children has been shown

in research on divorced families as well. Kurdek (1988)

found that at one year post-divorce follow-up, child

adjustment as evidenced by a limited number of behavior

problems, was predicted by a high degree of parental social

support. This study offers further support for the

importance of this variable to child adjustment. However,

it has been given little importance by judges in determining

custody arrangements (Lowery, 1981).

In contrast to the consistent similarities of effects

resulting from maternal and paternal sole custody, Warshak

and Santrock (1983) reported that the father's use of social

support networks increased and that the mother's standard of

living decreased after separation. This effect could result

from the different role expectations for mothers and fathers

and not be indicative of the parenting skills of mothers and

fathers (Warshak & Santrock, 1983). While the source of

this finding remains unknown, it clearly suggests that there

is a difference between maternal and paternal custody at

least in how the arrangements affects the families' lives.

Parent-Child Relationships

The most important variable found in predicting a

child's ability to interact warmly with peers and cope with

stress is the post-divorce parent-child relationship

(Felner, 1987; Hess & Camara, 1979; Santrock & Warshak,

1979; Shybunko, 1989). Some of studies evaluating this

issue have used structured observations of parent and child

behavior and independent raters. Santrock and Warshak

(1979) reported that boys having good relationships with

their fathers behaved more maturely and warmly, while those

having good relations with their mothers expressed better

self esteem. Girls showed a similar difference in their

behavior, demonstrating more self esteem and social

competence when relations with the mother were positive.

Relationships with each parent provided unique benefits to

child in social adjustment, with better adjustment being

related to free access to the noncustodial parent.

Hess and Camara (1979) also found that the best

predictor of positive social development and effective

coping behaviors in children following a divorce was the

quality of parent-child interactions, even over parental

disharmony. School work improved while aggressive behavior

and stress decreased when parent-child relationships were

largely devoid of conflict. Furthermore, children were more

likely to exhibit effective coping behaviors when both

parents maintained a positive relationship with them

following the divorce. In fact, the quality of a child's

relationship with each parent did not correlate

significantly. In other words, Hess and Camara (1979) found

that the child's involvement with each parent individually

after divorce is differentially related to a child's social


More recent research using standardized measures have

also evidenced the relationship between child adjustment and

the relationship with the noncustodial parent. Shybunko

(1989) found that in mother physical custody positive

father-child relationships post-divorce predicted child

social competence. Furthermore, Isaacs (1988) concluded

that the regularity of visits between fathers and children

buffered the child from interparent hostility. These

findings were consistent for joint custody and sole custody,

suggesting the greater influence of the child's relationship

to the noncustodial parent over the legal custody


Wallerstein and Kelly (1980) followed 60 families after

divorce, in part to describe the differential importance of

mother-child versus father-child relationships. The

mother-child relationship affected the level of stress in

the family, while the father-child interaction was most

important for adolescent development in both boys and girls.

They also found that if the father remarried, his

relationship with his children suffered. However, if the

mother remarried, the only change in the family was a

reduced level of stress for her.

The above studies suggest that post-divorce child

adjustment is more dependent on the quality of each

parent-child relationship, interparent conflict, and the

stability of the child's environment regardless of the

custody arrangement than maintaining a consistent contact

with one particular parent. In fact, none of the above

researchers have been able to find any consistent

differences in children's behaviors not explained by these

variables (Adams et al., 1984; Hess & Camara, 1979; Kline et

al., 1989; Santrock & Warshak, 1979; Shybunko, 1989). More

specifically, children's behavior were not found to be

related to parental employment status (Santrock & Warshak,

1979), or the number of visits made by a noncustodial

parent, when controlling for the duration of those visits

(Hess & Camara, 1979).

A preference for the mother or father as a single

custodian or the preferred primary residential custodian is

thus not supported by the above findings. Instead, these

studies suggest that the specific case variables of parental

conflict, parent-child relationships, and the style of

parenting will be most predictive of the child's adjustment

and hence most indicative of the best interests of the child

following a divorce. Furthermore, each of these variables

needs to be considered when attempting to predict child

adjustment, as they are all interrelated and influential.

Because of the limitations in the research cited above,

further research must be conducted to clarify which

variables are most important in predicting children's

post-divorce adjustment and how best to measure these

variables. Unfortunately, most of the articles already

published in this area have examined only small groups of

low conflict families, have only sampled one socio-economic

level, and have not matched samples for age of children or

parents, length of separation, or number of siblings.

Therefore, the effect of the variables studied on child

behavior may be exaggerated due the lack of consideration

for the above problems. Similarly, the proliferation of

position papers, which only describe an individual's

personal beliefs, add little to the fund of information

(e.g., Mnookin, 1975). Without the consideration of each

variable and its influence on behavior across a reasonably

large sample, results from research cannot be generalized to

other families.

There also remain several specific variables that have

not been examined in the literature with regard to their

influence on the normal development of children after

divorce, though they have been identified by judges as

important in the decision-making. The effect of remarriage,

for example, has only been given casual attention in the

literature on post-divorce child adjustment, though it has

been assumed by some judges to increase the consistency of a

child's environment and to fill the needs left by the

noncustodial parent (Sager et al., 1980). Wallerstein and

Kelly (1980) did state that the only benefit of remarriage

was to reduce maternal stress and that several problems

commonly occur in the families they studied that

counteracted this benefit. For instance, they found that

children feared the step-father would try to take the place

of their biological father and that divorce occurred more

often in the second versus first marriages. While this

study suggests that remarriage has less impact on the

children than the other variables discussed, only interview

data and maternal sole custody families have been used to

draw this conclusion. Further research is necessary to

delimit the effect of remarriage on several different types

of custody arrangements.

Another general issue that has also been considered by

judges as somewhat important in their decision-making but

has not been examined relevant to post-divorce child

adjustment is the moral influence of a parent upon children.

The issue of a custodial parent's sexual preference (homo-

versus heterosexual) is one example (Pagelow, 1980). Judges

also have considered several other variables that relate to

the moral influence of parents as well (Felner, 1987; Felner

et al., 1985). Furthermore, variables such as criminal

convictions, religious involvement, and the influence of

mental health experts have not even been addressed in

literature studying the factors that might influence a

child's development. Consequently, these variables and

their importance in predicting the post-divorce adjustment

of children need further evaluation.

In summary, present research supports increasing the

specificity of current legal guidelines over the simple

reliance on one type of custody arrangement. Though some

specific guidelines have been established and joint custody

has become more common following the recognition of the

importance of both parents to the child's adjustment, state

statutes remain incomplete and globally stated. The

statutes do not specify how to evaluate when one or both

parents are unfit as custodians or when to adjudicate a

different custody arrangement than joint custody.

Consequently, judges are still left to depend upon their own

perceptions of what variables affect children and upon court

precedence in following the statutes. As legal precedents

are further delineated, there will probably be a decreased

reliance on the use of simple social role definitions in

determining who will care for children. As judges become

more aware of social factors and modify their decision-

making regarding custody determinations, post-divorce child

adjustment can be further facilitated.

Further analysis of criteria defined by state statutes

and social science research will be important in

facilitating the shift to the use of specific criteria that

have been identified as important in promoting healthy

development for children after divorce. Further research

also is required to clarify these present findings.

Criteria based upon many specific variables that both

describe the family's characteristics and predict the future

development of the children could be developed that would

help delineate what custody arrangements best support the

child's interests. It is the authors' hope that a continued

relationship between social scientists, legislatures, and

court personnel will lead to improvements and further

specification of the decision process involved in supporting

the best interests of children.

Specific Aims

Previous research in the area of judicial

decision-making with regards to where a child will reside

following parental divorce has followed primarily one

methodology, asking judges to rate the importance of

specific criteria or asking lawyers what criteria they

believe judges use (e.g., Felner et al., 1985; Lowery, 1981;

Pearson & Ring, 1981; Sorensen & Goldman, 1989). While

these studies have offered some explanation of how the

physical residence of post-divorce children is determined,

they offer only general statements concerning the importance

of specific issues. These studies have only examined

ratings of the importance of criteria in general, rather

than in terms of specific types of cases. Additionally,

judges and lawyers may or may not be able to identify their

decision-making process clearly, and there is always the

potential for bias towards the socially desired response in

self report studies.

The purpose of the present study was to examine

judicial decision-making from the vantage point of

predicting individual case outcomes from the information

available to judges during court hearings. The importance

of each specific variable relative to all other information

collected was evaluated. The results of these analyses were

then compared with previous literature on judicial

decision-making regarding custody determinations, on child

adjustment following parental divorce, and with the Florida

legal statutes regarding custody determinations. Additional

comparisons were made of case outcomes from rural versus

urban and northern versus southern settings in order to

evaluate the dissimilarity in how custody is awarded within

different parts of the state. A secondary aim of this study

was to describe the characteristics of the families that get

referred to the Guardian Ad Litem.

Since this study is the first attempt at combining

research on judicial decision-making and post-divorce child

adjustment through an examination of case records, a

detailed analysis of the many specific issue relevant to

these cases could not be undertaken at the present time.

Rather, this project collected data covering a very large

range of psychological and sociological issues generally and

their interrelationships. The purpose of this project,

then, was to identify which of these sociological and

psychological issues are highly predictive of case outcomes.

Future research can make use of the findings from this

project to perform more detailed analyses on the specific

variables identified as important.


The literature on judicial decision-making and state

statutes regarding custody determinations suggest that the

present study should result in several findings. One

hypothesis was that judges will continue to rely, at least

in part, on a general preference for the mother as

residential custodian over the father regardless of some

specific case variables. It also was expected that this

preference for the mother will be stronger for children

under 12 years of age.

A second hypothesis was that there would be a few

issues that have a strong influence on judges'

decision-making in that these variables can determine if a

parent will be considered unfit as a custodian of the child.

It was expected that these issues would include

substantiated drug use, child or spousal physical,

emotional, or sexual abuse and neglect, emotional stability

of the parents, having kidnapped the child previously, and

parental handicaps. Similarly, it was expected that several

variables would have little importance relative to the above

issues. These variables were expected to include parental

education, employment, income and age, willingness to allow

visitation by the noncustodial parent, allegations only of

child or spousal abuse or neglect, alcohol or drug abuse,

child's wishes, placing the child with the same sex parent,

and having more time available for the children.



The subjects for this study were 60 families referred

to Florida's Guardian Ad Litem program (GAL) due to

involvement in contested custody disputes. In each of these

cases court intervention was necessitated because the

parties were unable to reach an agreement concerning the

custody of their childrenn. Data was not collected on any

cases resolved through mediation only. Families were

referred to the GAL by lawyers, judges, and family members.

These cases were adjudicated between September, 1988 and

October, 1990. A description of the GAL program is

contained in the procedures section of this document.

Of the sixty cases, 44% involved marital dissolution

and initial custody determinations, while 56% involved a

modification of one or more components of a previous custody

arrangements. Within the group of cases seeking a

modification, 20% involved a visitation modification, 85% a

residence modification, and 9% some other type of

modification (e.g., child support payments). In 58% of the

cases filing for a modification of a previous custody

arrangement, one or both of the parents had remarried.

Approximately two-thirds of the cases were referred to

circuits serving large metropolitan areas including Miami,

Jacksonville, Tampa, Orlando, and Gainesville. The

remainder of the cases were referred to a other circuits

that serve a number of urban and rural counties, but only

three cases involved families living in rural counties. The

sample was distributed over the state approximately equally,

with 61% from counties parallel to or north of Orlando and

39% from counties south of Orlando.

The families involved in these cases were predominantly

Caucasian (85%), although five were Black, one was Hispanic,

and three were of mixed race. Almost all of the families

included a child under the age of 12 (92%). The average age

of the children was 7.4, with a range from less than one

year to 20 years. About half of the families had only one

child (48%). There were approximately equal numbers of male

and female children in the sample. Demographic information

on the parents (see Table 1) indicates that the parents were

generally in their 30s and had little more than a high

school education. Mothers and step-mothers were frequently

unemployed and seldom earned more than $10,000 a year, while

fathers and step-fathers often held skilled jobs earning in

excess of $20,000. Only 18% of the families had a parent

who earned more than $30,000 a year. Additionally, one or

both parents has recently lost a job in 37% of the families.


Demographic Characteristics of the Parents


Step- Step-

Variable Mother Father Mother Father

Percent Single Parents 85 75 -- --

Average Age 31.3 34.8 32.3 39.6

Average Years of Education 12.3 13.3 13.3 13.7

Average Yearly Income* 8.4 26.5 6.9 23.7

Percent Unemployed 29 3 29 0

Percent Unskilled Labor 51 45 36 33

Percent Skilled Labor 12 30 21 67

Percent Professionals 8 22 14 0

* In units of $1000

Due to the large number of custody cases brought to the

courts each year and the scarcity of the GAL workers, it is

assumed that the cases referred to the GAL have a greater

than average level of family problems, most notably familial

conflict, when compared with the general population of

divorce cases.


A questionnaire was developed which included a total of

61 items (see Appendix). These items were derived from

previous research on judicial decision-making in custody

cases, literature on post-divorce child adjustment, and the

Florida statutes regarding custody determinations. Of these

items, 8 required the GAL representative to rate one or more

family members on a specific issue and to report their

confidence in making those ratings. (All ratings were made

on graphic scales and measured to generate a value from 1-

99.) There were also 2 questions which required the

guardian to make a judgement about which caretaker provided

for the children's needs. In addition to the ratings, there

were 10 items requiring a simple endorsement of the presence

or absence of specific problems made in a given case. The

remaining 41 questions required a simple report of

demographic and case information. Additionally, a variable

reflecting fitness of each parent was created with reference

to legal criteria in the Florida Statutes (Joint Legislative

Management Committee, 1985). These state that substantiated

child abuse/neglect or spouse abuse defines a parent as


Completion of this questionnaire required guardians to

evaluate many different characteristics of the family based

upon the data they collect during their interviews with

family members. This process of forming impressions about

families is also a central part of the Guardian Ad Litem's

court appointed role in custody cases. In order to evaluate

the stability both over time and across raters of the

ratings recorded on the questionnaire used in this study, a

pilot study was conducted.

Due to the large demand for the small number of

guardians working on custody cases and the sensitivity of

the material collected by the guardians, it was not possible

to generate reliability data using actual cases.

Consequently, a video tape of a mock interview between a

guardian and a family was developed. There are four

separate interviews included on this video tape, one with

the child alone, one with each parent individually, and one

with the all family members present. This video tape was

shown twice to a group of 19 experienced guardians who have

been assigned to at least one custody case previously, with

a two week interview between viewings. The video tape was

also shown to a group of 20 clinical psychologists licensed

in the state of Florida.

During each of these viewings, subjects completed the

section of the questionnaire that requires making

evaluations of family members. Of the 8 questions that

involve some rating on the family, 5 require separate

ratings for each parent's relationship with the childrenn.

All questions require separate ratings on the issue and the

person's confidence in the ratings. Consequently, 26 rating

variables were generated as part of this pilot study. The 2

questions which required the rater to make a judgement about

which caretaker provided for the children's needs were also

included in this pilot study.

Test-retest reliability was generated by calculating

the Pearson Product Moment correlation between each of the

guardian's ratings made after the first and second viewings.

Of the 26 correlations, 22 were significant at the .05

probability level or better, with 15 of those significant at

the .01 probability level or better. Of those 22, the mean

r (19) was .66, with a range from .46-.92. These results

suggest that there is moderate to good stability over time

for these 22 ratings.

Unfortunately, the relationship between the ratings

made during the first and second viewing on some of the

questions did not generate significant test-retest

correlations. These questions included the father's warm

feelings for the children, how hostile the mother was to the

interviewer, and the satisfaction with social support for

father. Ratings of the confidence guardians had in their

ratings of the mother's warm feelings for the children also

did not demonstrate significant test-retest correlations.

While a nonsignificant test-retest correlation suggests

that subjects' impressions of these issues change over time,

it does not provide any information either on the degree of

difference between the ratings or on any possible trend

within the ratings. In order to further examine how

dissimilar guardians' ratings on these variables were across

the two viewings, matched t-tests were calculated. None of

these variables exhibited a difference between the first and

second ratings, even at the .2 level of significance.

Consequently, there does not appear to be a large difference

between the ratings made across the two viewings.

Similarly, there was no evidence on any question of a

general trend for the rating made during the first viewing

to be consistently higher or lower than the one made during

the second viewing.

The test-retest stability for the two variables

requiring the guardian to determine which parent primarily

provided for the physical and emotional needs of the

children was calculated using a chi-square analysis and

percent agreement. Regarding physical needs, the same

parent was identified by 82% of the guardians across

viewings, indicating a significant reliability, X2 (1,17) =

5.24, R<.05. Identification of the parent providing for the

child's emotional needs was even more stable, X2 (1,17) =

18.00, R<.001, as 100% of the guardians identified the same

parent after the two viewings.

There remain several possible reasons why ratings made

during each of the viewings on these questions did not

demonstrate significant relationships. One possible

explanation is that there was insufficient variability in

the distribution of ratings made on these questions to allow

the Pearson Product Moment Correlation to calculate a

relationship. This possibility was evaluated by examining

the variance of ratings on these questions. None of the

distributions of ratings on these questions appeared

excessively limited, suggesting that variance was not an

explanation of the poor test-retest correlations.

A second possible reason why a relationship between the

ratings was not demonstrated is that guardians had

insufficient information on these issues to generate a

specific impression of the family and thus responded more

randomly to these items. It was assumed that guardians

would report low confidence in their ratings on a question

when they had little information on that topic.

Consequently, if low reliability was due to a lack of

information, there would be a relationship between

confidence ratings and the value of the test-retest

reliability estimate. Due to the small sample size and to

the fact that the reliability estimates were not normally

distributed, a Spearman Rank correlation was used to assess

the relationship between these variables. Results indicate

a significant positive relationship between these variables,

Zs (11) = .54, R<.01.

A third possible reason why stability over time was not

demonstrated for some ratings is that guardians' impressions

may be changing as they consider the information on the

family a second time. In fact, information salient to

guardians during the first viewing of the tape may not be

considered as important during the second viewing.

Unfortunately, information on how guardians form their

impressions of families is not available at this time. Thus

further statements about why these particular ratings do not

remain stable can not be made.

In general when the stability of a variable can not be

demonstrated, that variable is dropped from further

analyses. In this case, however, that course of action was

not taken because regardless of the stability of guardian's

impressions, these impressions are being reported to the

court in custody cases. Consequently, the ratings need to

be included in the analyses in order to evaluate their

relative influence in the judges' decision-making process.

Our findings of low test-retest reliability do indicate that

further study is necessary in order to evaluate why

guardians' ratings change over time.

In addition to evaluating the stability of guardians'

ratings over time, we also assessed the consistency in

ratings across raters. Viewers only rated one interview.

Consequently, it was not possible to calculate interrater

reliability of these questions using the standard procedure

of comparing a few raters across a series of ratings.

However, it was possible to examine the similarity of

ratings made across evaluators on individual questions by

examining the degree of variation within the distribution of

ratings. The coefficient of variation (CV=Std/g*100) is a

unitless measure of this variation that compares the size of

the mean to the size of the standard deviation. Although

there is no statistical test for the significance of this

value, values less than one hundred are generally considered

to indicate low interrater variability and thus high

consistency in the ratings. Values much greater than one

hundred suggest high interrater variability or poor

reliability between raters.

The values of the coefficient of variation for each

question was calculated for each of the two groups of

subjects, guardians and licensed psychologists, separately.

Guardians' ratings generated following the first viewing

were used. Ratings made by each group were found to have CV

values considerably less than 100 for all questions,

including ratings of confidence. In fact, the average CV

value for the guardians' ratings across questions was 36.32,

with a range from 9.59-60.90. The average CV value for the

psychologists' ratings across questions was 31.77, with a

range from 9.97-69.98. These values indicate considerable

similarity in the ratings made by guardians and

psychologists on all questions. We interpreted this high

degree of consistency to indicate that guardians as a group

have similar definitions for the issues queried in the

measure. Psychologists appear to similarly agree with one

another upon the definition of these issues.

In addition to evaluating the level of variation

between ratings made within these groups of subjects, we

also compared the guardians' ratings to those made by

psychologists on each question using t-tests. Results of

these comparisons indicated significant differences between

guardians' ratings and psychologists' ratings on only four

questions. Guardians rated the parents as less conflicted

(M = 65.32) than psychologists (M = 77.95), t (37) = -2.20,

R<.05. However, guardians rated the father-child conflict

as greater (I = 82.89) than psychologists (M = 71.00), 1

(37) = 3.46, R<.01. Guardians also rated each parent's

satisfaction with their social supports lower (M = 37.10 for

mothers and 35.21 for fathers) than psychologists (M = 49.90

for mothers and 49.70 for fathers), t (37) = -2.33, p<.01

and (37) = -2.04, R<.05 respectively. Although the

differences between these ratings were significant, the mean

values do not indicate radical differences between

guardians' and psychologists' ratings. Consequently,

guardians and psychologists appear to agree, at least in

general, upon the definition of the issues rated during this


Another question relevant to the use of the ratings is

what exactly are guardians using as concepts when evaluating

the families on the issues queried. For example, to what

are guardians referring when they report a high degree of

conflict within a family? Additionally, what degree of

conflict indicated on the rating form is consistent with a

normal versus clinical cutoff point? These questions are

beyond the scope of this research project. However, we did

attempt have some families complete some previously

validated measures so that we could compare guardians'

ratings to them, but we were unable to generate sufficient

data. Consequently, we will not be able to make statements

from these ratings about what information, other than

guardians' impressions, is being used by the court to form

custody determinations. Consequently, this project will

focus on the impact of guardians' impressions of families on

custody determinations.

Data were collected by GAL workers, who are generally

referred to as guardians. The GAL program is a public

agency operated under the direction of the Chief Judge of

the circuit court. The role of the GAL is to serve as an

advocate for the child and to provide both information about

the family and a recommendation regarding custody or

dependency to the court. Seven judicial circuits have GAL

programs serving in custody disputes; each of these

participated in the data collection.

The questionnaire was completed by guardians assigned

to each case after the court made its final determination.

The assistance of a supervisor in this process was provided

on occasion. GAL workers had access to all information

available to the court on these cases, except parental

criminal history.

The researchers were not provided with names of family

members or other identifying information; cases were

identified by a research number to protect confidentiality.

Due to the archival nature of the data, the Institutional

Review Board determined that provisions for receiving

informed consent were unnecessary.

Statistical Procedures

Much of the data collected for this study described

each individual parent and step-parent in the family

separately. For example, guardians' rated the amount of

conflict between each parent and the children.

Unfortunately, the sample size did not permit us to analyze

the influence of step-parents relative to natural parents on

judicial decision-making. However, we did not want to

simply drop the information on step-parents from our

analyses. Consequently, data on the step-parent were

combined with data on the natural parent to whom they were

married. This was achieved in two ways. Ratings made by

guardians were averaged. Reports of abuse or of a parent

taking a child secretly from the other parent were

considered present if they were made on either member of a

natural parent/step-parent pair.

GAL workers provided us with a number of ratings on

family relationship characteristics. In addition to rating

family relationship information, guardians also rated their

confidence in making those ratings. We hypothesized that

guardians confidence would affect their ratings' influence

on judicial decision-making. This hypothesis was tested by

including confidence ratings as main effects and as

interactions with the ratings in subsequent analyses.

Although we obtained confidence ratings for each

parent/step-parent pair, these ratings were highly related.

Correlations of confidence scores between parental dyads

averaged .80 and ranged from .67 to .93 (all R .0001). In

order to avoid multicollinearity among these confidence

scores, they were pooled across parental dyads for each


Although we received information on family relationship

characteristics and confidence for all cases, guardians

often reported only a subset of this information to the

court. Because we were using these ratings to predict

judicial decision-making, we needed to identify which

information was provided to the judges in each case. This

was accomplished by using a binary variable that indicated

whether the item was reported as a weighting variable. In

other words, rating scores were multiplied by this weighting

variable before they were included in the analyses. We also

examined whether guardians' decision to report information

on a rating was related to the value of the rating or to

their confidence in making the rating.

In order to determine which variables were predictive

of disposition, multiple logistic regression models were

calculated using the method of maximum likelihood. This

statistical procedure is appropriate when a variety of

discrete and continuous variables are being used to predict

a dichotomous independent variable. The test statistic for

this comparison is called the goodness-of-fit chi-square

(G2). This statistic compares the outcome predicted from a

regression equation with the observed outcome to determine

whether the model fits the data. In contrast to more

conventional analyses, in this procedure small values of the

statistic (or large p-values) indicate significant findings.

This statistic requires no assumptions of multivariate


In addition to testing the significance of a logistic

regression model, it is also possible to test the

significance of individual independent variables. This test

is also based on a chi-square, however interpretation of

this statistic is more conventional. Large chi-square

values (or small p-values) indicate that a variable

significantly improves prediction after all other variables

have been included in the model. Furthermore, the direction

of influence each significant variable exerts on the outcome

variable can be interpreted by the sign of the parameter

estimate. Positive estimate values indicate an increased

likelihood while negative values indicate a decreased


Due to the large number of independent variables

relative to the sample size, it was necessary to calculate a

substantial number of regression equations with a subset of

the independent variables included in each. The procedure

used to select variables for each regression is described

below in the Variable Selection section. After all

variables were included in a model, a backwards stepwise

procedure with replacements was used in order to exclude

variables that failed to improve prediction. The variables

remaining in these models were then combined into larger

models. Subsequent to this, variables that did not improve

prediction were again eliminated from the model in a

backwards stepwise fashion.

One of the consequences of running a large number of

models is an increase in the experimentwise error rate. In

order to control for this problem, a conservative criteria

was set for determining whether a variable improved

prediction. Variables were considered significant

predictors only when the overall model goodness-of-fit test

was nonsignificant and the variable chi-square test was


Before discussing variable selection, one problem in

the analyses must be discussed. Within all of the

regression equations, many of the independent variables were

highly correlated. This problem of multicollinearity

complicated both the analyses and the subsequent

interpretation. Although multicollinearity does not greatly

affect the model goodness-of-fit tests, it does bias

variable chi-squares tests towards nonsignificance.

Consequently, caution was used when excluding nonsignificant

independent variables from a regression model. Variables

that appeared to improve the predictability of the overall

model but did not have significant chi-square values were

retained in the model until the effect of other potentially

related variables could be evaluated.

Variable Selection

The independent variables were grouped into the

following categories: psychological, sociological, and

legal. Psychological variables primarily included items

describing family relationships that have been identified in

the literature as related in particular to post-divorce

child adjustment or in general to psychological well-being.

They include items addressing family conflict, social

support, and child(ren)'s emotional and physical needs.

Additionally, information on whether psychological

evaluations or treatment were recommended for any family

member and the parent's response to the guardian were

included in this group.

Variables were labeled sociological if they described

the demographic characteristics of the families. These

included race, occupation, income, education, which parent

had remarried, whether extended family members lived with

the parents, and the number of children in the family.

The final category, termed legal variables, included

items identified either in the Florida Statutes or in legal

journals as relevant to custody determinations. The

following information was included in this group: where the

parents planned to live, the child's wishes, who held

custody following parental separation, the emotional bond

between the parents and children, each parent's capacity to

care for the childrenn, whether any child is of "tender

years", reports of abuse or kidnapping, and the guardian's



Case Characteristics

One of the aims of this project was to describe the

cases referred to the GAL custody program. A description of

these cases should provide some reference points for

interpreting the predictability of judicial decision-making

regarding these families. Table 2 presents information on

the parent/step-parent pairs. Almost all parents are

represented by counsel and plan to live in the state.

Consequently, these issues do not differentiate between

parental dyads, and thus this information can have little

impact on decision-making. In contrast to the above items,

information on who held custody after separation, who

provided for the children's physical and emotional needs

prior to separation, and to a lesser degree who has more

time available to care for the children all point towards

one parent as a preferred caretaker, the mother.

Florida Statutes list likelihood of allowing the child

frequent and continuing contact with the nonresidential

parent as one of the guidelines for custody determinations.

Literature on post divorce child adjustment also support the

importance of continued contact with the nonresidental

parent (Hess & Camara, 1979; Isaacs, 1988; Shybunko, 1989;

Wallerstein & Kelly, 1980). In light of these points, the



Characteristics of Parent/Step-Parent Pairs*

Mother/ Father/

Step-Father Step-Mother
Represented by Counsel 92 93

Plans to Live in Florida 92 95

Provided for the Physical Needs of 70 27
the Children Prior to Separation

Provided for the Emotional Needs of 70 32
the Children Prior to Separation

Held Custody After Separation 73 15

Has More Time for Caretaking than 54 17
the Other Parental Pair

Can Be Labeled Unfit 28 18

Physical Problems Limit Caretaking 12 3

Secretly Taken the Child 18 13

More Willing to Allow the Other 33 35
Parental Pair Visitation

Recently Moved 37 25

falling into

* Table values represent percentage of pairs
each category.

item addressing willingness to allow parental visitation and

that which refers to a parent secretly taking a child from a

custodial parent are of interest. In our sample, concerns

about willingness to allow visitation arose in almost 70% of

the cases while kidnapping occurred in 31% of the cases.

Consequently, willingness to allow contact between the

children and the nonresidential caretaker appears to be an

issue of concern in many of these contested custody cases.

The information in the table also suggests that in a

few cases one or both parents may have limited caretaking

abilities. The difficulty caring for the children may arise

from some physical limitation due to illness, injury, or

handicap. This occurred in very few cases and only one

parent in a family was ever handicapped. Additionally, a

parent may be unfit as a caretaker due to a history of child

abuse/neglect or spousal abuse. Although at least one

parent was determined to be unfit in 40% of the cases, both

parents were unfit in only four cases (6%). Consequently,

these issues clearly discriminated between parental pairs.

Due to the impact abuse and neglect have on children,

further description of our sample with regard to these

issues is warranted. Figure 1 presents the relative

frequencies of the different types of abuse allegations and

substantiated reports. For all types of abuse and neglect,

allegations arose much more frequently than substantiated

reports, especially for sexual and emotional abuse. In




- 0




fact, 83% of the cases included at least one allegation of

abuse or neglect. Surprisingly, in 47% of the cases, these

allegations involved only one parental pair, and thus could

be used to discriminate custodial options for some cases.

With regard to the type of problem, more cases involved

allegations or substantiated reports or emotional abuse than

any other abuse, while physical and sexual abuse were

alleged or substantiated least often. This pattern was not

replicated when substantiated reports are considered alone.

Neglect was substantiated in the largest number of cases,

while sexual abuse occurred least frequently. In addition

to reports of current abuse/neglect, substantiated reports

of previous child abuse/neglect were present in 13% of the

cases. An additional 10% of the cases involved allegations

of previous abuse/neglect.

Reports of spousal physical and emotional abuse were

also prevalent in our sample. Spousal physical abuse was

alleged in 29% of the cases, while spousal emotional abuse

was alleged in 35% of the cases. Substantiated physical and

emotional abuse occurred in 11% and 16% of the cases


Reports of alcohol and drug abuse have also been given

attention in literature describing judicial decision-making

and post divorce child adjustment. In our sample,

allegations of substance abuse occurred in 60% of the cases,

while substantiated reports were made in an additional 25%

of the cases. Of the cases involving allegations or

substantiated reports of substance abuse, the reports were

made against one parent in 47% of them. Alcohol abuse was

listed as a problem in 52% of the cases, while drug abuse

was reported in 43%. Both were alleged in 37% of the cases.

Reports of alcohol and drug abuse were made approximately

equally against mothers and fathers. These figures indicate

that substance abuse reports are relatively common in these

contested cases.

The number of substantiated reports of abuse/neglect

and the number of parents involved with alcohol or drug

abuse indicate the incidence and degree of psychological

problems within these families. In fact, guardians

recommended treatment for a family member in 60% of the

cases. Additionally, guardians recommended that someone

undergo a mental health evaluation in 49% of the cases.

Evaluations, however, were only completed in 28% of the

cases. The difference in these frequencies are due in part

to the difficulty of getting a parent to consent to the

evaluation for either themselves or their children.

Frequency of Custody Awards and Recommendations

Before describing the result of the regressions, we

will discuss GAL recommendations for custody awards as well

as the actual custody determinations awarded by the court.

This information is reported in Table 3. Consistent with a

presumption for shared custody, this arrangement was awarded


Custody Awards and Recommendations

Final Determinations

Shared Custody with Mother as Primary Custodian

Shared Custody with Father as Primary Custodian

Shared Custody with Split Primary Custody

Maternal Sole Custody

Paternal Sole Custody

Other Custody Arrangement

Guardian Ad Litem Recommendations

Mother as Primary Custodian

Father as Primary Custodian

Alternative Arrangement Recommended

Children's Wishes

Preference for the Mother as Primary Custodian

Preference for the Father as Primary Custodian

No Preference was Stated








much more frequently than sole custody or any other

arrangement. Additionally, primary residential custody was

almost universally awarded to one parent under shared

custody. An interesting item evident in the table is the

number of cases in which physical residence was awarded to

the father. These results suggest that, at least in our

sample, there is little apparent tendency for either

guardians to recommend physical residence to mothers or for

the court to award custody to mothers over fathers. In

fact, a preference for the mother to be the primary

caretaker was evident only in the children's wishes.

Prediction of Custody Awards

The primary aim of this study was to describe the

factors that are most influential in child custody

determinations. A series of logistic regression equations

were calculated in order to address this issue. In

particular, two aspects of custody determinations were

addressed: whether shared or sole custody was awarded and

which parent would serve as primary residential custodian.

Awards of Shared Versus Sole Custody

Prior to evaluating the importance of legal,

sociological, and psychological issues, we assessed whether

custody awards might be influenced by factors unrelated to

the information provided to the court. For instance, since

data was collected on both initial disposition and

disposition modification cases, we were concerned that

custody awards might differ between these groups of cases.

We were also interested in testing whether the location of a

judicial circuit affected custody awards. Thirdly, in a

subset of the case, guardians did not have the opportunity

to interview all family members; the influence of this

procedural variable was also of concern. When these

variables were included in a prediction equation, none of

them surfaced as significantly related to an award of shared

versus sole custody. Consequently, custody appears to be

awarded similarly across the state, among disposition and

modification cases, and when guardians did not interview all

family members.

The legal variables were then entered into a series of

logistic regression equations. A few variables emerged as

significantly related to the determination of shared versus

sole custody within several of these regressions. As

presented in Table 4, both the model goodness-of-fit chi-

square and the variable chi-square were significant in each

case. In this table, positive parameter values indicate

that higher values of that variable resulted in a increased

likelihood of sole custody awards.

Relationships between fathers and their children,

specifically the degree of warm feelings held by the fathers

for the children, also affected custody awards. However, it

is difficulty to interpret the effect this variable has on

custody awards. The parameter estimate for this variable is



Logistic Regressions Using Legal Variables
To Predict Shared Versus Sole Custody

Parameter Chi-
Df Predictor Variables Estimate Df Square

Initial Legal Models
11.66a 9 Physical Problems Limit
Mothers Caretaking

45.36a 52 GAL Rating of Father's
Warm Feelings for
the Children

GAL's Confidence in Rating
Warm Feelings

23.66a 21 Mother Determined Unfit

Presence of a Young Child

Interaction Between Child's
Age and Child's Wish to
Live with their Mother

13.95a 14 Substantiated Physical Abuse
by the Father

Combined Legal Model
33.28a 51 Physical Problems Limit
Mothers Caretaking

GAL Rating of Father's
Warm Feelings for
the Children

GAL's Confidence in
Rating Warm Feelings

Mother Determined Unfit












1 3.76b

1 3.75b

1 3.85b

1 4.19b

1 4.37b

1 4.78b

1 4.65b

1 5.81b

a R>.05
b p<.05
c P<.01
d p<.001




negative, suggesting that shared custody was more likely to

be awarded when this variable was rated high. However, the

parameter estimate for GAL confidence in this rating is

positive, suggesting that confidence mediates the

relationship between ratings of father-child relationships

and an award of shared versus sole custody. Consequently,

in cases where confidence in this rating is low, high

ratings of paternal warm feelings increase the likelihood of

an award of shared custody, while low ratings of paternal

warm feelings have little effect on custody determinations.

However, in cases where confidence is low, high ratings of

paternal warm feelings have little effect on custody

determinations, while low ratings of paternal warm feelings

increase the likelihood of an award of sole custody.

Children's wishes and their age both influenced the

likelihood of sole versus shared custody awards. Taken

together, the significant effect of age and the interaction

between children's wishes and age suggest a nonlinear

relationship between these variables and custody awards.

When children were young and wished to live with their

mothers, shared custody was the most likely award. In fact,

primary residence was usually awarded to the mother in these

cases. However, when children were young and they did not

wish to live with their mothers, the likelihood of sole

custody being awarded to the father was increased. The

wishes of older children had little impact on awards of sole

versus shared custody; as there were only four families

without young children, this effect could not be adequately

tested within this sample.

Sole custody was more likely to be awarded when mothers

were either physically handicapped or perpetrators of any

form of child abuse or neglect. Further examination of the

data indicated that in these cases, judges awarded sole

custody to the father. Similarly, cases involving

substantiated reports of physical child abuse perpetrated by

the father also resulted in increased awards of sole

custody. In these cases, sole custody was typically awarded

to the mother. It is striking in this context, however,

that substantiated reports of child sexual abuse, child

emotional abuse, neglect, or spousal abuse do not

significantly increase the likelihood of judges awarding

sole custody when the father was the perpetrator.

Legal variables identified as significantly predictive

of outcome in the aforementioned regressions were then

combined into another logistic regression model. This model

was significant (see Table 4), and the following variables

were shown to be related to the custody arrangement awarded:

presence of maternal physical problems limiting caretaking

ability; the guardian's rating of the father's warm feelings

for the childrenn; the guardian's confidence in that

rating; and a determination that the mother was unfit as a

caretaker. The presence of a young child in the family,

substantiated reports of paternal physical abuse, and the

interaction between children's wishes and their age were

dropped from the combined model. This does not indicate

that these variables were unrelated to outcome, but that

they did not add significantly to the prediction of custody

awards over the other variables.

The eleven sociological variables were entered into two

regression equations. Although both of the two models were

significant, G2 (24, N=57) = 30.03, p_.05 and G2 (23, N=57)

= 21.47, p>.05, only mothers' occupational level was shown

to be predictive of outcome. For the purposes of this

study, occupation was classified into four categories:

unemployed, unskilled labor, skilled labor, and

professional. Within our sample, sole custody was more

likely to be awarded when mothers had skilled jobs than

otherwise (parameter value=1.30), X2 (1, N=57) = 3.95,

R<.05. Upon further examination of the data, the awards of

sole custody in these cases usually went to the mothers.

Given the small number of cases (N=4) in which mothers held

professional jobs, our sample may not have adequately tested

the relationship between custody determinations and this

occupational level.

Psychological variables, entered into two regression

models, were highly predictive of whether shared or sole

custody was awarded. Model goodness-of-fit chi-squares and

at least one variable chi-square were significant (see Table


As can be seen in Table 5, sole custody is more likely

to be awarded in those cases in which guardians rate mothers

as being hostile towards them. Fathers' relationships with

people outside the immediate family are also important

predictors of shared versus sole custody. The significant

interaction and main effects for ratings of fathers' social

support and guardians' confidence in those ratings together

indicate a nonlinear relationship between these variables

and shared versus sole custody awards. A judge is more

likely to award sole custody when guardians are confident

that fathers are satisfied with their social supports than

when the guardians are uncertain of their ratings on this

issue. Conversely, shared custody is a more likely outcome

when guardians rate fathers as being only moderately

satisfied with their social supports as well as when the

guardians are less confident of their ratings. This

relationship suggests that in the absence of clear

information on fathers' social support, judges favor an

award of shared custody.

The above paragraphs describe how each of the three

groups of variables, sociological, psychological, and legal,

individually predict the award of shared versus sole

custody. After the variables most predictive of custody

awards within these groups were specified, a combined model


Logistic Regressions Using Psychological Variables
To Predict Shared Versus Sole Custody

Predictor Variables


Initial Psychological Models

30.76a 50 Hostility of Mothers
Towards GAL

GAL Rating of Father's
Satisfaction with
Social Support

GAL's Confidence in Rating
Social Support

Interaction Between
Father's Social Support
Rating and GAL Confidence




Df Square

1 8.47C

1 3.05a

1 4.21b

.001 1 5.79b

3.90a 21

Psychological Evaluation
was Recommended for
Some Family Member


1 4.82a

Combined Psychological Model

30.76a 50

Hostility of Mothers
Towards GAL

GAL Rating of Father's
Satisfaction with
Social Support

GAL's Confidence in Rating
Social Support

Interaction Between
Father's Social Support
Rating and GAL Confidence




1 8.47c

1 3.05a

1 4.21b

.001 1 5.79b

a R>.05
b p<.05
C p_.01

was constructed to identify those variables most predictive

of custody awards. Nonsignificant variables were excluded

from the equation, resulting in the model presented in Table


In particular, three variables out of the nine entered

into this final model were dropped. These variables

included mothers' occupational level, presence of maternal

physical problems that limited caretaking ability, and

determinations that mothers were unfit as caretakers.

These variable were dropped because they failed to increase

the predictability of the outcome over the remaining

variables. Thus the remaining variables are the strongest

predictors of judicial determinations of sole versus shared


As Table 6 shows, mothers' and fathers' responses to

and involvement with people outside of the immediate family

are strong predictors of custody determinations. In

particular, judges are more likely to award sole custody

given either the presence of maternal hostility towards

guardians or confident GAL ratings of paternal satisfaction

with available social supports.

Relationships within the family are also predictive of

custody awards. As discussed earlier, the relationship

between ratings of paternal warm feelings for their children

and custody awards is contingent on the guardians'

confidence in their ratings of this variable. The reader


Logistic Regressions Using Combined Psychological,
Sociological, and Legal Variables
To Predict Shared Versus Sole Custody

Df Predictor Variables


24.10a 48 Hostility of Mothers
Towards GAL

GAL Rating of Father's
Satisfaction with
Social Support

GAL's Confidence in Rating
Social Support

Interaction Between
Father's Social Support
Rating and GAL Confidence

GAL Rating of Father's
Warm Feelings for
the Children

GAL's Confidence in
Rating Warm Feelings




Df Square

1 7.39c

1 2.34a

1 3.35a

.001 1 4.42b



1 4.28b

1 4.19b

a R>.05
b p<.05
c p<.01

should note that as with the issue of paternal social

support, judges are more likely to award shared custody when

there is an absence of clear information on these issues.

Awards of Maternal Versus Paternal Primary Physical

As with prediction of shared versus sole custody, we

assessed whether the purpose of the case (initial

disposition or disposition modification), a change in GAL

procedures (i.e., a family member was not interviewed), or

the location of the circuit influenced the award of primary

physical residence (PPR). These variables were entered into

a regression model before psychological, legal, and

sociological data were considered. Both the model chi-

square and the variable chi-squares indicated no

relationship between these variables and primary physical

residence awards. Consequently, PPR appears to be awarded

similarly across the state as well as among disposition and

modification cases. Additionally, PPR is not influenced by

the unavailability of a subset of family members for

interview by guardians.

Next, legal variables were entered into a series of

regression equations. Variables emerging as significant

predictors of PPR and with significant goodness-of-fit model

chi-squares are listed in Table 7. In this table, positive

parameter values indicate an increased likelihood of awards

for maternal primary residence.


Logistic Regressions Using Legal Variables
To Predict Primary Physical Residence Award

Df Predictor Variables


Initial Legal Models

40.86a 30 Mother has More Time
to Care for Children

64.58a 52 GAL Rating of Father's
Daily Involvement with
the Children

8.90a 11 Mother Held Residence
Following Parental

60.91a 49 Confidence in Rating
Parental Warm Feelings
for the Children

Interaction Between Ratings
of Mother's Warm Feelings
and Confidence

Interaction Between Ratings
of Father's Warm Feelings,
Mother's Warm Feelings, and





Df Square

1 13.12d

1 6.58c

1 5.42b

1 4.37b

.004 1 3.78b

.00002 1 3.84b

25.79a 22

Child Wishes to Live
with the Mother

Mother Determined Unfit

26.21 19 Substantiated Report
of Maternal Neglect




1 12.75d

1 4.93b

1 12.74d

TABLE 7--continued

Df Predictor Variables

Parameter Chi-
Estimate Df Square

Combined Legal Model

35.55a 49 Child Wishes to Live
with the Mother

Mother Determined Unfit

Mother Held Residence
Following Parental

GAL Rating of Father'
Daily Involvement with
the Children





1 6.94c

1 7.30c

1 4.77b

1 4.81b

a R>.05
b <. 05
C D<.001
d D<.001

Maternal PPR was more likely in cases where the mother

had more time to care for the children than the father and

when the mother held residential custody of the children)

following parental separation. Maternal PPR was also more

likely when children expressed a desire to live with their


Paternal PPR was more likely to be awarded in cases

where fathers had been involved in providing for the child's

daily living needs and when substantiated reports of

maternal child abuse/neglect or spouse abuse were present.

Within the set of variables regarding abuse, reports of

maternal neglect were most predictive of paternal PPR. In

fact, PPR was never awarded to mothers in cases where

reports of neglect had been substantiated.

The significant interaction effects which emerged

between ratings of parental warm feelings for the children

and guardians' confidence in these ratings describe an

interesting relationship between these variables and PPR

awards. When GAL confidence is high, ratings of maternal

warm feelings strongly influence the likelihood of PPR

awards. High ratings of maternal warm feelings predict an

award of maternal PPR, while low ratings on this variable

predict paternal PPR. However, when confidence is low,

ratings of maternal warm feelings have little influence on

PPR awards regardless of the value of this variable. In any

case, ratings of paternal warm feelings have only a moderate

impact on the likelihood of PPR awards.

The legal variables identified as significant

descriptors of judicial decision-making in one of the

several regression models were then entered into a combined

model. The purpose of this model was to identify the

strongest descriptors among all the legal variables. This

combined model (see Table 7) was significant, and the

following variables were shown to be related to the award of

PPR: children's wishes to live with their mother, the

determination that a mother was unfit, the child's residing

with the mother following parental separation, and GAL

rating of the father's involvement in providing for the

child's daily living needs. Variables including whether the

mother would have more time available for the child and

those involving ratings of parental warm feelings were

dropped because they did not add significantly to the

predictability of the model.

Sociological variables were entered into the next

series of regression equations. Out of the eleven variables

in this group, only one emerged as significantly predictive

of PPR awards. Paternal PPR was more likely to be awarded

when there was only one child in the family than otherwise,
G2 (11, N=55) = 9.59, p>.05 and X2 (1, N=55) = 5.07, p_.05.

Psychological variables were entered next. Of the four

regression equations calculated, only one was significant,

G2 (4, N=55) = 5.80, R>.05. Within this model, however, no

variable obtained significance. Consequently, these

variables, including family conflict, parents' relationships

with nonfamily members, and the presence of a recommendation

for either a mental health evaluation or psychotherapy do

not appear to be very influential in the decision-making


The previous paragraphs in this section describe how

variables within each of the three groups influence

decision-making regarding awards of PPR. With the exception

of the item indicating the number of children in the family,

all variables identified as influential fell within the

legal domain. All of these variables were combined in

another regression in order to determine which were most

influential. The goodness-of-fit chi-square for this model
was significant, G2 (47, N=55) = 30.83, p>.05.

Additionally, the individual chi-squares for all of the

legal variables were also significant. The number of

children, however, did not add anything to the prediction

equation. Consequently, the model combining predictors from

all three groups of variables was identical to the combined

legal model presented in Table 7. Thus, four legal items,

child(ren)'s wishes for a residential parent, maternal

fitness, the child's residing with the mother following

parental separation, and the degree to which fathers were

involved in providing for the daily living needs of the

children, were the most influential in determinations of

parental PPR.

An additional area of interest was the relative impact

of GAL recommendations regarding PPR awards in comparison to

the other predictors discussed. This variable was not

entered into the model earlier because it is based on the

same information described above and would be largely

collinear with the other variables. Consequently, a model

was constructed that included this variable and the four

legal variables found to best predict PPR awards. The

resulting model was significant, G2 (43, N=55) = 22.63,

p>.05, but no variables other that GAL recommendation, X2

(1, N=55) = 5.73, ps.05, were significant. Further

examination of the data showed that judicial determinations

were consistent with GAL recommendations in 86% of the


Guardians' Decisions Regarding When to Report Ratings

With regard to custody cases, the importance of

guardians' opinions about family members is evident in the

previous sections. Both their characterizations of family

relationships and their recommendations regarding custody

awards were shown to be influential. Consequently,

guardians' decisions regarding when to report information

significantly affected the decision-making process. One

reason for reporting a given characteristic could be that a

family exhibited it to an unusually high or low degree. A

second possible reason is that guardians' felt confident

about their evaluation of the family with regard to a given


The two hypotheses listed above were tested by using

the guardians' ratings of a family characteristic, their

confidence in making those ratings, and the interaction

between these to predict whether the issue were reported to

the court. Some issues required separate ratings for each

parent. For these issues, ratings on each parent separately

and on the interaction between them were included in the

analyses. As with the analyses on custody awards, the

outcome variable here was binary. Consequently, a logistic

regression procedure was used. Although some G2 values were

significant, no variable chi-squares were. Thus, in every

case, the results indicated no significant relationship

between either variable and the decision to report the



The results of this study provide descriptive

information about the contested custody cases which are

served by the GAL program in Florida. These cases involve

predominantly Caucasian families with young children. The

parents tend to be in their 30s, have a high school

education, and have a joint family income of between $25,000

and $35,000 a year. As mothers were frequently unemployed,

fathers tended to be the principle wage earners. Almost all

of the cases were referred from urban settings.

Additionally, the vast majority of parents were represented

by counsel in these cases.

These families have experienced considerable stress.

In addition their involvement with the court, many parents

have recently lost their jobs and the custodial parent is

often single. In a few of these cases, the conflict between

the parents resulted in the kidnapping of the child by one

of the parents. It is interesting to note, however, that

very few parents had either moved out of the state or had

plans to do so.

Additionally, many of these families exhibited signs of

more serious psychological problems. Substantiated reports

of abuse by at least one of the parents were present in 40%

of the cases, and allegations of abuse had been made in 83%.

The extent to which these allegations reflect actual beliefs

that abuse has occurred versus an attempt by the parents to

gain leverage in the custody dispute is undetermined.

Reports of alcohol and drug abuse were also common in these

families. Furthermore, guardians were sufficiently

concerned about the psychological functioning of these

families to recommend a mental health evaluation in 49% of

the cases and treatment in 60% of them.

It appears, therefore, that divorce is probably only

one part of a larger picture of personal and familial

problems experienced by these individuals. It is clear from

this description that the judges seldom have an

unequivocally safe and supportive placement available for

the children.

Before discussing specific case variables, it is

important to note that, in general, judges favored shared

custody over sole custody. The degree to which this was the

case suggests that judges approached cases with the

presumption that they would award shared custody. Such a

presumption is consistent with the legal statutes (Joint

Legislative Management Committee, 1985). An interesting

finding in this regard, was that the strength of the

information presented weighed heavily in the decision-

making. For example, ratings indicating that a father had

warm feelings for his children predicted sole custody awards

only when the guardian was confident in making this rating.

In the absence of confidence, this rating had only a

marginal impact on custody awards. Taken in conjunction

with the data indicating that judges ruled in accordance

with GAL recommendations in 86% of the cases, this suggests

that judges relied heavily on the judgement of the


In determinations of primary physical residence, judges

did not demonstrate a bias toward either parent. This is

consistent with some recent literature on judicial decision-

making suggesting a general decrease in the degree to which

judges favor maternal PPR even when there are young children

in the home (Felner et al., 1985; Sorensen & Goldman, 1989;

Racusin et al., 1989). Thus, the tender years presumption

no longer appears to hold in Florida. Although awards of

maternal PPR did occur somewhat more often, this appeared to

reflect characteristics of the case relevant to the legal

statutes. In particular, awards of maternal PPR were more

likely to occur in cases where mothers held custody

following parental separation, and paternal PPR was more

likely to occur in cases where the father has been more

involved in child care prior to marital dissolution.

Additionally, the slightly higher frequency of maternal PPR

awards is consistent with the greater caretaking role of the

mother in most of the families prior to separation. In

these cases, an award of maternal PPR maintained continuity

in the children's caretaking environment.

According to Goldstein et al. (1973), awarding PPR to

the mothers here supports the best interests of the children

by placing them with the parent to whom they are most

bonded. Literature on post divorce child adjustment also

suggests that consistency in a child's living environment is

important. In his review of recent literature, Felner

(1987) noted that consistency in the child's living

environment predicted adjustment. Consequently, maternal

PPR awards are consistent with the goal of protecting the

interests of children and do not indicate a general

preference for the mother as caretaker.

Issues relevant to parenting capacity were also

influential in the decision-making process. Substantiated

reports of maternal child abuse/neglect or spouse abuse

generally resulted in an award of sole paternal custody.

Additionally, sole paternal custody was awarded in cases

where a mother had physical problems that limited her

ability to care for her children. Similarly, in cases that

involved substantiated reports of paternal physical abuse

against the child, sole maternal custody was often awarded.

Consistent with our results, reports of abuse and neglect

have influenced custody awards for many years. This finding

dates back to the initiation of the best interests of the

child doctrine (Derdeyn, 1976), and has been supported

consistently in the literature examining judicial decision-

making (e.g., Folberg, 1984; Hoorwitz & Burchardt, 1984;

Pearson & Ring, 1981) in spite of the changes that have

occurred in the legal presumptions and statutes.

However, although some abuse reports resulted in sole

custody awards to the nonabusing parent, many did not. With

regard to fathers, only child physical abuse appeared to

affect judicial custody awards. Child sexual abuse,

emotional abuse, neglect and spousal abuse did not appear to

influence case awards significantly. This occurred even

when abuse reports were substantiated. This is in contrast

to the situation for mothers, wherein all forms of

substantiated abuse impacted judicial decision-making,

although neglect was most influential. It is difficult to

understand the finding that substantiated abuse has

divergent functional outcomes contingent upon which parent

perpetrated the abuse. The judges' decision-making would

appear to suggest that abuse, other than physical, renders

mothers but not fathers unfit as parents.

It is noteworthy that in comparison to substantiated

abuse, allegations of abuse did not emerge as predictive of

custody awards. One possible explanation for these findings

is that judges may consider abuse allegations to be simply a

product of the divorce process and not an accurate gauge of

parental competence. Secondly, given the frequency of

allegations in this sample, it is possible that they no

longer communicate anything about a case. However, it is

important to note that the lack of a significant

relationship between abuse reports and custody awards does

not indicate that this information was never utilized by

judges, only that it was not done so consistently.

Another issue considered by judges in contested custody

cases is the parent-child relationship. When fathers are

rated as having strong warm feelings for their children,

shared custody is more likely. Conversely, sole custody is

more likely when fathers are not rated as concerned about

their children. Although ratings of this variable have

little impact on the determination of parental PPR, ratings

of maternal warm feelings lead to maternal PPR when high and

paternal PPR when low. Consequently, judges make awards

that facilitate involvement with the children for any parent

known to have affection for them. Modifying custody awards

based on the parent-child relationships is consistent with

child adjustment literature, as warm parent-child

relationships have consistently predicted increased social

adjustment (Hess & Camara, 1979; Santrock & Warshak, 1979;

Shybunko, 1989).

A relationship between parental affection for children

and custody awards has been demonstrated in past research

(Felner et al., 1985; Settle & Lowery, 1982; Sorensen &

Goldman, 1989) as well as in this study. However, this

study found that judges considered this issue differently

depending on the parent. Fathers demonstrating a lack of

warm feelings were removed from a position of responsibility

over the children; sole maternal custody was awarded.

Mothers demonstrating a lack of warm feelings were denied

PPR but continued to be involved in child care through

shared custody.

Parental relationships with people outside of their

immediate family was also influential. High ratings of

fathers' satisfaction with social supports and maternal

hostility towards a guardian both predict an award of sole

custody. One possible reason for this finding is that

maternal expressions of hostility may lead judges to

conclude that the parents would not be able to work together

in a shared custody situation. The rational behind awarding

sole custody in cases of high paternal social support is

unclear, especially in light of the fact that this issue is

not significantly related to an award of PPR. This issue

has been given little attention in previous literature on

judicial decision-making. It is possible that parental

social support is related to other case variables that do

predict PPR awards. Unfortunately, the size of our sample

did not permit the type of exploratory analyses that might

uncover this kind of relationship.

Children's wishes also predicted custody awards.

Primarily they predicted PPR awards, but for young children

they also predicted sole versus shared custody. Cases

involving young children who did not want to live with their

mothers generally resulted in awards of sole paternal

custody. Maternal shared custody was awarded when children

wanted to live with their mothers. This result is

consistent with Florida Statutes which list children's

wishes as one of the guidelines for determining custody.

The reason for the sole versus shared custody awards in

these cases is unclear. However, since more children wished

to live with their mothers than fathers, it is possible that

a desire to live with one's father often arose out of

circumstances that would warrant restriction of the mother's

involvement in child care. Again, our sample size

restricted us from testing this hypothesis.

Another interesting finding was that cases in which a

guardian recommended that a family member receive a

psychological evaluation resulted in awards of sole custody

more often than when this was not the case. Previous

research has also found that judges consider parental mental

health in these cases (Lowery, 1981; Sorensen & Goldman,

1989). It appears that in the context of custody cases,

recommendations for psychological evaluations augur problems

sufficient to warrant restricting a parent from child care

responsibility. However, as with earlier research (Ash &

Guyer, 1984), this variable did not appear to be as

important to judges as other issues. When this variable was

included in a model with other variables found to influence

custody awards, it failed to add to the model. Research on

child adjustment has identified parental mental health as

strongly related to child adjustment. Specifically, mental

health problems have been related to child behavior problems

even at two-year follow-ups (Kline et al., 1989; Kurdek,

1988). This finding suggests that children might benefit if

parental mental health is more strongly considered by


Only two of the sociological variables were shown to

relate to custody awards. Higher maternal occupational

status increased the likelihood that judges would award sole

custody. It is possible that greater occupational status

indicates that a mother is less dependent on paternal

resources and thus may make sole custody a more viable

option. Discovery of a relationship between

occupational/financial status and custody awards is not new

(Felner et al., 1985; Lowery, 1982).

The second significant sociological variable, whether

there was one or more than one child, was related to PPR

awards. Paternal PPR occurred more frequently in cases in

which there was only one child. It appears that judges may

perceive fathers as less capable of caring for more than one

child. The rational for this decision remains unclear, and

it has not been studied previously.

In contrast with previous literature (Felner et al.,

1985; Lowery, 1982), our study did not demonstrate a

Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd