Interparental Conflict and Children of Discord and Divorce

See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/16053310

Interparental Conﬂict and Children of Discord and Divorce

ArticleinPsychological Bulletin · October 1982

DOI: 10.1037/0033-2909.92.2.310·Source: PubMed

CITATIONS

1,125

READS

11,779

1 author:

Some of the authors of this publication are also working on these related projects:

Child effects View project

Robert Emery

University of Virginia

191 PUBLICATIONS10,321 CITATIONS

SEE PROFILE

All content following this page was uploaded by Robert Emery on 14 September 2015.

The user has requested enhancement of the downloaded file.

Psychological

Bulletin

1982,

Vol.

92, No. 2,

310-330

Copyright

1982

by the

American Psychological Association, Inc.

0033-2909/82/9202-0310S00.75

Interparental Conflict

and

the

Children

of

Discord

and

Divorce

Robert

E.

Emery

University

of

Virginia

Data

on the

relation between marital turmoil (i.e., discord

and

divorce)

and

behavior

problems

in

children

are

reviewed.

It is

concluded that

a

relation

be-

tween

the two

domains docs exist. Several parameters

of

this relation

are

outlined,

including

type

of

marital turmoil, form

of the

child's behavioral response,

sex

differences,

age

effects, parental

buffering,

and

effects

of

parental psychopa-

thology.

Conclusions drawn

from

this review

are

used

to

evaluate several broad

etiological

hypotheses about

the

effect

of

marital turmoil

on

children,

and im-

plications

for the

treatment

of

behavior problems

in

children

from

these families

are

discussed. Finally, interpretative

and

methodological refinements

are

sug-

gested

for

future research.

The

idea that marital

turmoil

1

is the

cause

of

a

variety

of

behavior problems

in

children

is

widely held both

in the

public

and in the

professional

domain. During

the

first

6

months

of

1980, such prominent media

as

Newsweek

and the New

York

Times

Mag-

azine

ran

cover stories

on

children

of

divorce,

and

the

National Institute

of

Mental Health

solicited research proposals

on the

topic.

Recent interest

has

been spurred

by

popular

movies

and

books, which themselves reflect

the

turmoil

found

in an

increasing number

of

American families.

In

fact,

the

Academy

Award

winner

for

best motion picture

in

1980

depicted

the

problems

of

marital sep-

aration

and

child custody

(Kramer

vs.

Kra-

mer),

and the

1981

winner portrayed family

conflict

in a

two-parent household (Ordinary

People).

It has

been estimated

that

38% of the first

marriages

of

women

in

their

late

twenties

will

end in

divorce

(Glick

&

Norton, 1978).

Although

divorce

is

slightly less frequent

among

couples

who

have children under

5

years

of

age,

an

estimated

45% of the

chil-

dren born

in

1977

will

live

in a

one-parent

family

for at

least

several months, primarily

as a

result

of

divorce

(Cherlin,

1977; Glick

&

Norton,

1978).

The

census bureau

has

reported

a 79%

increase

in the

number

of

single-parent families between 1970

and

1980,

so the

current prevalence

is one in five

(U.S. Census Bureau, 1980). Because

not

every

discordant marriage

is

dissolved, these

astronomical

figures on the

number

of

chil-

dren

of

divorce omit

a

significant

and

largely

unknown

number

of

children

who are ex-

posed

to

serious marital conflict.

Psychologists have been called

on to

pro-

vide

both explanations

and

treatments

for

families

who are

affected

by

marital turmoil.

The

response

has

encompassed almost

the

entire range

of

possibilities.

Some

psychol-

ogists have suggested that whenever there

is

a

problem child, there

is a

problem mar-

riage

(Framo,

1975); others have concluded

that

a

relation between divorce

and

child

problems

has yet to be

demonstrated

(Her-

zog

&

Sudia, 1973).

The

relative lack

of

sophisticated

research regarding

the

relation

between

marital

turmoil

and

child

problems

encourages

such

a

diversity

of

opinion. Nev-

I

would like

to

thank

K.

Daniel

O'Leary,

Alan

O.

Ross,

and

Everett Waters,

as

well

as

John Darley,

as-

sociate editor,

and

several anonymous reviewers

for

their

comments

on

earlier versions

of

this article.

Requests

for

reprints should

be

sent

to

Robert Emery,

Department

of

Psychology,

Gilmer

Hall,

University

of

Virginia,

Charlottesville,

Virginia 22901.

1

The

term

turmoil

refers

to

families

characterized

by

discord

in

marriage, separation,

or

divorce

as a

group.

The

term

marital

problems

is

occasionally used

as a

synonym.

Interparental conflict

is

used

to

denote open

hostility

between married, separated,

or

divorced par-

ents.

Marital

discord

refers

to

problems

in

intact mar-

riages

only.

310

CHILDREN

OF

DISCORD

AND

DIVORCE

311

ertheless,

careful examination

of the

litera-

ture reveals that

a

considerable body

of in-

formation

is

available.

Covariation Between Marital

and

Child

Problems

A

number

of

investigators have

found

a

relation between discord

in

intact marriages

and the

severity

or

frequency

of

behavior

problems

in

children.

This

rinding

remains

consistent

across such countries

as the

United

States

(Emery

&

O'Leary,

in

press;

Olt-

manns,

Broderick,

&

O'Leary,

1977;

Porter

&

O'Leary,

1980), England (Rutter, 1971,

1979),

and

India (Chawla

&

Gupt, 1979).

Similarly, divorce

and

child problems have

often

been

found

to be

related (Anthony,

1974; Hetherington, 1979;

McDermott,

1968,

1970; Wallerstein

&

Kelly,

1980;

Westman,

Cline,

Swift,

&

Kramer, 1970).

Many

studies

of

marital

and

child

problems,

however,

suffer

from

one or

more

of a

variety

of

methodological

flaws. The

three most

common

problems

are (a)

biased

sampling—

usually

from

a

clinic population;

(b)

non-

independent

data—that

is, the

same judges

rate both

the

marriage

and the

child,

or,

in

the

case

of

divorce,

the

judges

of

child

be-

havior

are

aware

of the

marital status;

and

(c) the use of

measures lacking

in

reliability

and

validity.

Each

of

these problems

may

affect

estimates

of the

magnitude

of the as-

sociation between marital

and

child prob-

lems.

Stronger

associations between marital dis-

cord

and

child adjustment seem

to be

found

in

clinic

rather

than

in

nonclinic samples

(O'Leary

&

Emery,

in

press).

In a

series

of

studies

by the

Stony Brook group,

all

of

which used

parents'

ratings

on

well-estab-

lished

measures

of

child behavior (Behavior

Problem

Checklist,

Quay

&

Peterson,

Note

1)

and

marital

adjustment

(Short

Marital

Adjustment

Test,

Locke

&

Wallace, 1959),

stronger associations were

found

in

clinic

(Emery

&

O'Leary,

in

press; Oltmanns

et

al,

1977; Porter

&

O'Leary,

1980) than

in

nonclinic

samples (Oltmanns

et

al.,

1977;

Emery,

Note

2).

Significant relations were

found

in

both sets

of

samples;

the

clinic sam-

ples,

however,

may

have differed from

the

nonclinic

ones

on

some

third

variable

that

increased estimates

of the

association

be-

tween marital

and

child problems

(O'Leary

&

Emery,

in

press).

Some

data

suggest

that

the

expectation

bias

2

(Rosenthal,

1966)

of

raters

of

child

behavior

who are not

blind

to the

marital

relationship

can

also lead

to

higher estimates

of

the

relation between marital

and

child

problems. Robinson

and

Anderson

(Note

3)

found

that

a

significant correlation between

mothers'

checklist

ratings

of

their

marriage

and

their children's adjustment

was

reduced

to a

chance level when social desirability

was

partialed out. Emery

(Note

2)

found stron-

ger

correlations between mothers' marital

ratings

and

their

own

judgments

of

distur-

bance

in

their children than between moth-

ers' marital ratings

and

teachers' child

ad-

justment

scores.

Because

mothers

and

teach-

ers

rate behavior

in

different

settings,

however,

situation-specific behavior could

explain

this

result.

In

fact,

one

might

expect

child

behavior

to be

most

affected

in the

set-

ting

where

the

marital turmoil

is

present.

Thus, independent measures

of

marital dis-

cord

and

child problems

in the

home setting

may be

exceedingly

difficult,

but

important,

to

obtain.

Unlike

the

effects

of

using

clinic samples

and

nonindependent

ratings,

the use of un-

reliable measures would tend

to

attenuate

rather than increase estimates

of the

mag-

nitude

of the

association

between

marital

and

child problems. Researchers

are not

fully

satisfied

with

the

reliability

and

validity

of

the

available

measures

of

either childhood

adjustment

(Achenbach, 1978)

or

marital

relations

(O'Leary

&

Turkewitz, 1978).

It

is

not

surprising,

then, that many

of the re-

ports

reviewed below used measures

of

ques-

2

The

expectation—held

by

parents,

teachers,

and

mental health

professionals—that

marital

turmoil

causes

behavior

problems

in

children

not

only

may

create

biased

rating data

but

also

may be

problematic

in

that

a

self-fulfilling

prophecy

may

result.

For

example, par-

ents

who

attribute

a

child's (normal) misbehavior

to an

emotional

reaction

to

marital turmoil

may not

respond

to

that

child

with

their usual discipline. Thus,

in

their

attempt

to

understand

the

child, some parents

may set

limits

that

are

inconsistent

and

confusing

and

thereby

accidentally

create

the

problems that they

are

trying

to

avoid.

312

ROBERT

E.

EMERY

tionable adequacy. Unreliability

is

especially

important

to

consider when comparing cor-

relations

between

a

given

measure

and

sev-

eral other measures

of

different

reliabilities.

In

this

case,

differences between

the

correla-

tions

that

appear

to be

substantive may,

in

fact,

only

reflect error variance.

Given

the

above concerns,

one

could choose

to

attend

to

only

the

very small number

of

reports

that

contain none

of

these

problems.

This option

was

rejected

in the

present

re-

view

for

four

reasons. First,

the

topic poses

important

questions

that

must

be

answered

by

clinicians

on the

basis

of

existing data.

It

seems

worthwhile, therefore,

to

attend

to

whatever

evidence

is

available, albeit with

appropriate caution toward methodological

limitations. Second, close inspection

of the

data across studies

often

reveals

a

conver-

gence

of

results

from

studies

containing dif-

ferent

flaws. It is

unlikely, though possible,

that

different

flaws

would consistently

lead

to

similar erroneous conclusions. Third, some

of

the

results that

may be

attributable

to

methodological differences

are of

interest

in

their

own

right.

For

example,

the

difference

between

clinic

and

nonclinic

samples

is an

intriguing

one. Finally,

in

those

few

studies

that contain none

of the

three

problems,

marital

and

child problems

are

related.

That

is,

researchers

who

used well-established

measures

while studying

intact,

nonclinic

families

have

found

that

teachers'

ratings

of

children's behavioral adjustment were

re-

lated

to

independently obtained parental

re-

ports

of

discord (Block, Block,

&

Morrison,

1981;

Emery,

Weintraub,

&

Neale,

in

press;

Rutter, 1971; Emery, Note

2).

Similarly,

divorce

and

child adjustment problems have

been

found

to be

nonclinic samples

where child behavior

was

indexed

by

such

independent measures

as

behavioral obser-

vation,

teacher ratings,

and

peer nomina-

tions

(Hetherington,

1972;

Hetherington,

Cox,

&

Cox, 1978). Thus, although meth-

odology

can

affect

estimates

of the

magni-

tude

of the

relation between marital

and

child

problems,

the

relation

is

nevertheless

a

real

and

important one.

One can be

con-

fident

that

the

results

of

investigations

that

used weaker methods

are not

wholly attrib-

utable

to

methodological third variables.

Pathogenic Dimensions

of

Marital Turmoil

A

necessary

step

in

further specifying

the

covariation between marital

and

child prob-

lems

is to

define marital turmoil

and

child

behavior problems more precisely. This task,

however,

is

extremely

difficult.

Neither

the

field

of

abnormal

child

psychology (Achen-

bach,

1978)

nor the

study

of

marital

rela-

tions (Gurman

&

Kniskern, 1978;

O'Leary

&

Turkewitz, 1978)

has

produced

a

widely

accepted classification system.

The

present

review

takes

a

cautious approach

to

child

adjustment

by

limiting discussions primarily

to two

broad

and

well-accepted

dimensions

of

child disturbance: problems

of

overcontrol

(e.g.,

anxiety,

withdrawal)

and

problems

of

undercontrol

(e.g.,

aggression, conduct dis-

orders;

Achenbach

&

Edelbroch, 1978).

more anxiety-related problems.

McDermott

(1968,

1970) reported that con-

duct problems were most prevalent

in his

divorce

sample,

but he

also

found

signs

of

depression

in

these

children.

Whereas

the

above

two

studies were based

on

noninde-

pendent, clinical impressions,

a

study

by

Hetherington

et

al.

(1978)

showed that non-

clinic

children

of

divorce were more depen-

dent, disobedient, aggressive, whining,

de-

manding,

and

unaffectionate

than were chil-

dren

from

intact marriages

as

indexed

by

reliable

observational measures.

In

studies

of

intact marriages,

McCord

et

al.

(1962)

found

that discord

was

feminine-aggressive behavior, antisocial

be-

havior,

and sex

anxiety

but not to

abnormal

fears

in a

nonclinic sample.

In a

clinic group,

Pemberton

and

Benady (1973) found that

marital

discord

was

childhood

aggression

and

encopresis

but not to

enu-

resis. Porter

and

O'Leary (1980)

found

mar-

ital discord

to be

related both

to

problems

of

conduct

and to

anxiety

in

clinic children,

but two

similar investigations found signif-

icant

relations

only

for

conduct problems

(Emery

&

O'Leary, 1982; Oltmanns

et

al.,

1977).

Also

in

direct

conflict,

Rutter (1971;

Wolkind

&

Rutter, 1973) found significant

relations

for

conduct problems

but not for

neurotic problems

in

nonclinic samples,

whereas

another British general population

report

found

significant

relations

for

both

conduct

and

neurotic problems

(Whitehead,

1979).

Finally,

in a

study that used exem-

plary

methodology, Block

et al.

(1981)

found

that

a

measure

of

parental disagreement

about

child rearing

was

more strongly

re-

lated

to

teachers' ratings

of

undercontrol

of

impulses than

to

their

ratings

of

overcontrol

of

impulses

in

nonclinic

children

who

were

evaluated

1,

2, and 4

years

after

the

marital

index

was

completed.

316

ROBERT

E.

EMERY

Although

these studies present

a

some-

what

conflicting

pattern

of

results, every

in-

vestigation, whether

of

questionable

or

sound

methodology,

found

marital turmoil

to be

some

form

of

undercontrolled

be-

havior.

The

results

for

overcontrol were

in-

consistent.

The

fact that

undercontrol

is

typ-

ically rated

more

reliably than overcontrol

(Cantwell,

Russell, Mattison,

&

Will, 1979;

Quay

&

Peterson,

Note

1;

Emery,

Note

2)

may

partially account

for

this

difference.

Still,

it

seems

safe

to

conclude that inter-

parental

conflict—in

broken

and in

intact

families—appears

to be

related more strongly

to

children's problems

of

undercontrol than

of

overcontrol.

Different Effects

on

Boys

and

Girls

Considerable

recent evidence suggests that

marital turmoil

has a

greater

effect

on

boys

than

on

girls from both divorced (Cadoret

&

Cain, 1980;

Hess

&

Camara, 1979; Heth-

erington

et

al,

1978;

McDermott,

1968;

Wallerstein

&

Kelly, 1980)

and

intact, dis-

cordant

marriages

(Block

et

al.,

1981;

Emery

&

O'Leary,

1982; Porter

&

O'Leary, 1980;

Rutter,

1971; Wolkind

&

Rutter,

1973).

Nevertheless,

a

careful examination

of the

literature suggests that

one

must consider

a

number

of

factors when interpreting this

finding.

On the

basis

of

data obtained

in

interviews

with nonclinic families,

Rutter

(1971)

found

that discord

in

intact marriages

was

asso-

ciated with school problems

in

boys

but not

in

girls.

Whitehead

(1979)

questioned

this

result.

Citing Becker's (1964) work docu-

menting

that

parental

punitiveness

was re-

lated

only

to

boys' behavior

in

school

but to

both boys'

and

girls' behavior

at

home,

Whitehead

argued

that

the

relation

with

marital turmoil

was

obscured because girls

inhibit aggression

in

school.

She

presented

data from

a

nonclinic sample showing sig-

nificant

associations between mothers'

re-

ports

of the

marriage

and

both

boys'

and

girls' problems

at

home.

The

results reported

by

other investigators (Emery

&

O'Leary,

1982;

Porter

&

O'Leary,

1980),

however,

are

contrary

to

Whitehead's.

In

samples

of

clinic children, significant relations were

found

between parents' ratings

of

marital

discord

and

their ratings

of

behavior prob-

lems

at

home

for

boys, whereas

no

signifi-

cant

relations

were found

for

girls. Similarly,

Gassner

and

Murray (1969) reported

find-

ing

a

greater

difference

in the

degree

of ob-

served

parental conflict between groups

of

clinic

and

nonclinic boys than between sim-

ilar groups

of

girls.

A

factor that

affects

the sex

difference

more

than

the

setting

may be the

type

of

sample

used.

In

nonclinic samples

of

intact

marriages, problems

in

both boys

and

girls

have been found

to be

discord

(Block

et

al.,

1981; Whitehead, 1979; Emery,

Note

2),

whereas

in

clinic samples, relations

have

been

found only

for

boys

(Emery

&

O'Leary, 1982; Porter

&

O'Leary, 1980).

When

it is

noted

that

Rutter's

(1971)

non-

clinic

sample

was

preselected

to

contain

a

high proportion

of

disturbed children

and

children

of

parents

with individual psycho-

pathology—thus

making

it

comparable

in

many

ways

to a

clinic

group—this

study, too,

is

consistent with

the

clinic/nonclinic

sex

differences.

How

is

this

pattern

of

results

to be ex-

plained?

An

examination

of how

each

sex

may

respond

to

marital turmoil

may

hold

an

answer.

Two

nonclinic

studies

of

intact

mar-

riages (Block

et

al.,

1981;

Whitehead,

1979)

and

one of

divorce

(Hess

&

Camara, 1979)

found

marital turmoil

to be

related directly

to

measures

of

undercontrol only

for

boys,

whereas

those

weaker

associations

that

were

found

for

girls were with overcontrolled

be-

havior.

(In

fact,

one

study [Block

et

al.,

1981]

found

that

discord

was

inversely

re-

lated

to

undercontrol

in

girls.) Because chil-

dren

are

more likely

to be

referred

to

clinics

because undercontrolled behavior

is

more

readily

identified

as a

problem than

is

over-

controlled behavior

(Ross,

1980),

it may be

the

case

that

boys respond

to

interparental

conflict

in a

more pervasive,

maladaptive

manner

that

is

more likely

to

lead

to a

clinic

referral.

Thus,

it may be

that

there

is a sex

difference

in

response

to

marital

discord,

but

the

difference

is in how and how

much

the

sexes

respond,

not

whether they

do.

A

further

complication

exists

in

inter-

preting

sex

differences among

the

children

of

divorce.

These

differences appear

to be a

result

of at

least

a

two-fold process. First,

CHILDREN

OF

DISCORD

AND

DIVORCE

317

as

in

studies

of

conflict

in

intact

families,

differences

betwen

the

sexes

in

terms

of

con-

flict

responses

are

expected.

In

addition,

there

may be a

second

effect

being

in

the

custody

of the

opposite-sex parent.

Santrock

and

Warshak

(1979),

in a

study

that

used observational

and

self-report mea-

sures

of a

nonclinic sample,

found

that

both

boys

and

girls

of

divorce

had

more behavior

problems

when

the

opposite-sex parent

had

custody

than when

the

same-sex parent

had

custody.

Because divorced mothers currently

have

child custody over

90% of the

time

(Glick,

1979), boys would

be

expected

to

have more postdivorce adjustment problems

than would girls.

Thus,

it

appears

that

boys

whose parents divorce

and

whose mothers

have

custody have

a

dual risk

for

developing

behavior

problems.

In

sum,

it can be

concluded that marital

turmoil,

in

several studies

of

adequate meth-

odology,

is

more strongly related

to

boys'

than

to

girls' obviously maladaptive behav-

ior.

One

needs

to

consider, however, setting

effects,

sampling,

form

of the

response,

and

custody arrangements

in

interpreting

sex

differences.

Further,

it is

possible

that

the

effects

on

girls

may be

delayed and,

as was

found

in one

study, appear prominently

in

subsequent

relations with

the

opposite

sex

(Hetherington,

1972). Girls

are

likely

to be

just

as

troubled

by

marital turmoil

as

boys

are,

but

they

may

demonstrate their feelings

in

a

manner

that

is

more appropriate

to

their

sex

role, namely,

by

becoming anxious, with-

drawn,

or

perhaps very well behaved.

Child's

Age

Various

arguments suggest that either

older

or

younger children should

be

more

affected

by

marital

turmoil.

For

example,

it

could

be

argued that younger children

are

more susceptible because

of

their greater

dependence

on

their parents. Children

as

young

as 1

year

old

have been shown

to re-

spond

to

interparental

conflict

with upset

and

anger (Cummings,

Zahn-Waxier,

&

Radke-Yarrow,

1981).

On the

other

hand,

older children

are

more sensitive

to

emotion

and

may

feel

pressure

to

become

involved

in

interparental

conflict,

thus making them-

selves more vulnerable

to its

effects.

Although

such speculations

are

common,

surprisingly

little evidence

is

available

on

differential

age

effects.

Two

investigations

of

discord

in

intact marriages

that

controlled

for

age

found

no

major

effect

(Porter

&

O'Leary, 1980;

Rutter,

Graham,

Chadwick,

&

Yule,

1976).

These

studies,

however,

ob-

tained nonindependent

data

and

were lim-

ited

to

children

6

years

of age or

older.

Wallerstein

and

Kelly (1974, 1975; Kelly

&

Wallerstein, 1975, 1976),

in

contrast, sug-

gested

that

divorce

has

unique

effects

on

children

of

different

ages, from

2

years

to

adolescence. Their results were based solely

on

clinical impressions

and

small samples,

however,

and

therefore

cannot

be

viewed

as

reliable. Further, although these investiga-

tors reported subtly

different

effects

at

dif-

ferent

ages,

all age

groups were

found

to be

affected

substantially

by

divorce.

Results

from

studies examining

the

effects

of

a

child's

age at the

time

of

divorce have

been

inconsistent. Findings

that

suggest

more detrimental

effects

of

separations

oc-

curring

before

the age of five or six

(Heth-

erington, 1972), after this

age

(Gibson,

1969;

McCord

et

al,

1962),

and no

effect

of

age at the

time

of

separation (Power

et

al.,

1974), have

all

been reported. Thus,

it

has

not

been demonstrated that

a

child's

age

is

an

important determinant

of the

effects

of

marital turmoil. Nevertheless,

age

contin-

ues to be of

interest

and

merits careful con-

sideration

in

future

investigations.

Good

Parent-Child

Relationships

Some recent empirical investigations have

supported

the

clinical impression that

the

parent-child relationship deteriorates

as a

result

of

marital

turmoil

(Hess

&

Camara,

1979;

Hetherington

et

al.,

1979;

Rutter,

1971).

None

of

these

investigators,

however,

suggest that deterioration

is

inevitable,

and,

in

fact,

it is

argued that

a

1

good relationship

with

one

parent

can

"buffer"

the

child

from

some

of the

negative

effects

of

marital tur-

moil

(Hetherington

et

al.,

1979).

In

a

study

of

intact,

nonclinic families,

Rutter

(1971)

reported

that

a

good relation-

ship with

at

least

one

parent

significantly

reduced

the

likelihood

that boys would

be

judged antisocial

by

their teachers.

It is

318

ROBERT

E.

EMERY

worth

noting, however,

that

boys from dis-

cordant

marriages

still

had

more

problems

than

did

boys

from

happy marriages, despite

the

good parent-child relationship. Further,

a

good

parent-child

relationship

was

strictly

defined

as

consisting

of

both positive warmth

and

lack

of

negative feelings

as

determined

by

parental interview.

Similarly,

only

particularly good

parent-

child

relationships produced

buffering

ef-

fects

in a

study

of

children

of

divorce

(Heth-

erington

et

al.,

1979).

Children's

adjust-

ment,

as

measured

by

multiple methods,

was

better only when, based

on

several parenting

indices,

the

parent-child relationship

was

classified

as

very good. Relationships judged

to be

moderate

had no

buffering

effect.

Fur-

ther,

the

very good relationship

had to be

with

the

mother.

Positive father-child rela-

tions

did not

mitigate

the

detrimental

effects

of

marital turmoil.

In

another nonclinic study

of

divorce,

Hess

and

Camara (1979) also

found

that

measures

of the

quality

of the

parent-child

relationship were important predictors

of

child

adjustment

as

indexed

by

multiple

in-

dependent

measures.

In

fact,

statistical

anal-

ysis

revealed that parent-child relations pre-

dicted

a

greater proportion

of the

variance

in

child adjustment than

did

interparental

conflict.

As

with

Rutter's

(1971)

study, how-

ever, conflict

was

children's

ad-

justment even when there

was a

good par-

ent-child relationship. Finally,

in

contrast

to

Hetherington's

findings,

good relation-

ships with either parent appeared

to

have

a

buffering

effect.

In

sum,

there

is

emerging evidence based

on

adequate methodology that

a

particularly

warm

relationship with

at

least

one

parent

can

mitigate,

but not

eliminate,

the

effects

of

marital turmoil

on

children. Without

doubt,

the

quality

of

parent-child

relation-

ships

deserves further attention

in

studies

of

both discord

and

divorce.

Individual

Parental

Psychopathology

It is

well documented

that

children from

families

where

a

parent

has a

psychological

disturbance

are at an

increased risk

for a

variety

of

behavioral

problems

(Mednick

&

McNeil, 1968). Further, disordered

in-

dividuals

are

more likely

to

have discordant

marriages

and to

divorce (Molholm

&

Din-

itz, 1972); thus,

it is

possible that marital

turmoil

may

partially explain

the

increased

problems among

the

children

of

these

indi-

viduals.

Data supporting this possibility come

from

Rutter

(1971).

He

found

that

discord

in in-

tact

marriages

was

antisocial

be-

havior

in the

children both

of

normal parents

and

of

parents with

a

personality disorder.

On the

other hand, when

the

marriage

was

harmonious,

there

was no

increased antiso-

cial behavior associated with parental per-

sonality

disturbance. There was, however,

a

trend toward

an

even greater

risk

for

anti-

social problems when both discord

and

per-

sonality

disturbance were present.

Emery

et al. (in

press) reported similar

findings

among

the

children

of

parents with

other diagnoses.

In

intact marriages discord

explained most

of the

association between

parents'

affective

disorder

(unipolar

depres-

sion

or

bipolar disorder)

and

children's dis-

turbed

school behavior. When

the

effect

of

marital

discord

was

taken

into

account,

little

association

was

found

between these disor-

ders

and

children's behavior.

But

when

the

diagnosis

was

parental

schizophrenia, mar-

ital

discord

did not

explain

the

children's

problems

in

school.

In

fact, little association

was

found

between discord

and

child behav-

ior

problems

in the

schizophrenic

group,

per-

haps

because

the

children

of

schizophrenics

were already disturbed

for

other reasons.

Another possibility

is

that, although

the

global

satisfaction ratings were similar, mar-

ital discord

may be

qualitatively

different

in

schizophrenic families. Conflict

may be re-

duced because

one

partner

is

acknowledged

as

sick.

These

two

reports—both

based

on

non-

clinic

samples, independent raters,

and re-

liable

measures—suggest

that, except

in the

case

of

schizophrenia, concomitant marital

turmoil

may

explain

a

good part

of the in-

creased problems among

the

children

of be-

haviorally

disordered

parents.

Researchers

need

to

examine more closely

the

dual, non-

independent

effects

of

interparental

conflict

and

parental

psychopathology

on

children.

CHILDREN

OF

DISCORD

AND

DIVORCE

319

Mechanisms

By

Which

Marital

Turmoil

Affects

Children

Research

on

marital

and

child problems

has not

generally been guided

by

broad eti-

ological rationales.

Rather,

mini-theories

that

offer

only

a few

predictions

or

make

no

allowance

for

critical evaluation

are

more

common. Because

of

such theoretical

defi-

ciencies, this review evaluates several

hy-

potheses about

how

marital turmoil

may

produce childhood disorders:

(a)

disruption

of

attachment

bonds,

(b)

modeling,

(c) al-

tered discipline practices,

and (d)

other

mod-

els, including stress, taking

on the

symptom,

and

child

effects.

Each

hypothesis

is

evaluated

in

terms

of

the

parameters

of the

relation between mar-

ital

and

child

problems.

Where appropriate,

theoretical assumptions

are

weighed against

the

conclusions

that

(a)

concomitant con-

flict,

not

separation

per se,

appears

to be re-

sponsible

for

many

of

divorce's serious, long-

term

pathogenic effects;

(b)

openly hostile

and

continued

conflict

has a

great

effect;

(c)

the

most prominent behavior disorder

is

a

problem

of

undercontrol;

(d)

boys demon-

strate

a

greater observable response than

do

girls;

(e) age has not

been shown

to

alter

the

child's reaction;

(f) a

good relationship with

at

least

one

parent

can

partially

buffer

the

negative

effects;

and (g)

increased

interpar-

ental

conflict

may

explain many

of the

nega-

tive

effects

found

among children

of

parents

with

individual psychopathology. Other

di-

rect

evidence

is

discussed where

it is

avail-

able

and

appropriate.

Attachment

and

Separation

One of the few

detailed hypotheses about

the

effect

of

marital turmoil

on

children

is

the

maternal deprivation hypothesis

(Bowlby,

1969).

According

to

this account,

the

loss

of

a

figure

to

whom

one is

attached (usually

the

mother) produces

an

instinctive

fear

re-

sponse.

The

survival

value

of the

fear

re-

sponse

accounts

for its

instinctive nature.

Fear

keeps

the

infant

proximal

to the

care-

taker, thus increasing

the

chances

for

sur-

vival.

On the

basis

of

this rationale, Bowlby

(1973) concluded that whenever

an

attach-

ment

figure

is

lost

for any

reason, anxiety

results.

The

specificity

of

this hypothesis makes

it

easily

testable,

but

theoretical assumptions

in

relation

to

marital turmoil

may

have been

overstated.

The

strongest prediction

from

the

attachment hypothesis

is

that

divorce—

because

it

involves separation

from

a

par-

ent—has

a

direct,

negative

effect

on the

child.

Although separation responses have

clearly been demonstrated, evidence

re-

viewed

earlier suggests

that

concomitant

in-

terparental conflict, rather than separation

per

se, is the

more salient factor

in

creating

behavior

problems

in the

children

of

divorce.

Further,

the

hypothesis predicts

that

mater-

nal

deprivation

is the

most damaging

be-

cause

the

mother

is

most commonly

the

pri-

mary

attachment

figure.

Divorce usually

re-

sults

in

paternal deprivation, thus

the

hypothesized

effects

on

these

children

are

unclear.

Other parameters discussed earlier

are

also

in

conflict

with

the

attachment hypoth-

esis. Separations before

the age of

five

are

predicted

to

be the

most damaging (Bowlby,

1973),

but no

consistent

effects

of age at the

time

of

divorce were

found.

Problems dem-

onstrated

in the

animal literature

as

result-

ing

from

maternal deprivation

are

frequently

fears

and

poor

social

development

(Rutter, 1972),

yet

marital turmoil

was

found

to be

most strongly associated

with

conduct disorders. Finally,

sex

differences

are not

explicitly predicted

by the

theory.

Variations

on the

attachment hypothesis

(e.g., Goldstein, Freud,

&

Solnit, 1973) have

suggested that loss

of

love (emotional

de-

privation)

may be the

central cause

of

prob-

lems.

In

support

of

this hypothesis, some

studies comparing clinic

and

nonclinic fam-

ilies

found

that

parents

of

clinic children

were

more negative with their children (Bug-

ental,

Love,

&

Kaswan, 1972; Lobitz

&

Johnson,

1975a; Love

&

Kaswan, 1974).

Further, review

of the

buffering

effect

sug-

gested that

a

good parent-child relationship

may

lessen

the

detrimental

effect

of

marital

turmoil.

Nevertheless,

it is

unclear whether loss

of

love

accounts

for

these

findings

or

whether

some other aspect

of the

parent-child

rela-

320

ROBERT

E.

EMERY

tionship

produces

the

effect.

Some

direct

evidence

supports

the

latter

alternative.

Us-

ing

a

measure

of

felt

acceptance

rated

by

the

child

(Schaefer,

1965), Emery

and

O'Leary

(1982),

in a

clinic sample,

found

that children's perceptions

of

being accepted

by

their parents were

not

significantly

re-

lated

to

parental ratings

of

satisfaction

with

their

intact marriages. Further, although

the

children

were generally aware

of

marital

discord,

their ratings

of

felt

acceptance

did

not

correlate

significantly

with

their

own

rat-

ings

of

parental discord. Thus, marital dis-

cord

had

little direct

effect

on the

children's

feelings

of

being loved, perhaps because par-

ents were particularly concerned about this

possibility

and

explicitly guarded against

it,

In

sum, although

it is

unlikely

that

de-

privation

from

a

parent's love

is the

principal

explanation

for the

association between mar-

ital turmoil

and

ongoing behavior problems

in

children,

the

attachment hypothesis

has

made

a

valuable contribution

in

identifying

the

acute

distress syndrome. Researchers

would

do

well

to

attempt

to

identify

more

fully

separation responses

in the

children

of

divorce.

This

could lead

to

distinguishing

children's

reactions

to the

difficult,

but ap-

parently

temporary,

1-2-year

separation

pe-

riod

identified

by

divorce researchers (Heth-

erington,

1979)

from

their ongoing behavior

problems

that appear

to

have

a

different

cause.

Such

information

would

be of

great

value

to

clinicians

and to

parents

who

could

be

educated

as to

what

to

expect

from

chil-

dren

as a

natural course

of

adaptation during

a

trying period

in

family

life.

Before

we

leave

the

topic

of

attachment

and

separation,

two

additional points should

be

made. First, although loss

of

love

is

prob-

ably

not the

primary mediating variable

be-

tween

marital

and

child problems,

it

does

account

for

important variance

in

explaining

behavior

problems

of

childhood

(e.g.,

Ar-

mentrout,

1971;

Emery

&

O'Leary,

1982).

The

attachment hypothesis therefore merits

further

consideration.

Second,

the

specificity

of

the

hypothesis

has

stimulated interest,

offered

testable predictions,

and

generated

research that

states

clear objectives.

The

area would

benefit

from

similarly

precise

hypotheses.

Modeling

is a

second major

etiological

mechanism

by

which marital turmoil

may

affect

children. Schwarz (1979) suggested

that

interparental conflict

may

interfere with

imitation

of the

same-sex parent

or may

lead

to

rejection

of

both parents

as

models.

Either

process

would disrupt normal socialization

in

that

appropriate parental behavior might

not

be

imitated,

and

other,

more

deviant,

models might

be

found.

Other

modeling

hy-

potheses

are

also applicable.

For

instance,

parents

in an

unhappy marriage

may

exhibit

more

hostile

and

aggressive behavior than

do

happily married couples,

and

this behav-

ior

could

be

imitated

by

their children.

Un-

happy

couples show more

conflict

than

do

happy couples (Jacobson

&

Martin, 1976;

Patterson, Weiss,

&

Hops, 1976),

and

much

of

this behavior might

occur

in the

presence

of

their children.

The

increase

in

negativity

toward children

that

corresponds with mar-

ital turmoil (Johnson

&

Lobitz,

1974) could

be

interpreted

as

evidence

in

support

of

this

latter hypothesis.

Considerable research

on the

parameters

of

marital turmoil

is

consistent with mod-

eling

hypotheses. Interparental conflict,

as

opposed

to

separation

from

a

parent, would

be

expected

to be

strongly

to

child

behavior problems.

The

stronger

rela-

tion

between interparental

conflict

and

con-

duct

problems

is

also congruent with

a

mod-

eling

hypothesis. Disagreements

and

argu-

ments

are

often

features

of

marital turmoil,

and

aggression

is

readily imitated

by

chil-

dren

(Bandura, 1973). Because

the

specifics

of

the

interparental

conflict

and

children's

conduct problems

may be

quite

different,

however,

research

that

details

the

corre-

spondence between

how

parents express hos-

tility

and how

children respond

to

witnessing

it is

needed.

Modeling

explanations

of the sex

differ-

ence

in

response

to

marital turmoil

are

per-

haps

most interesting.

The

modeling litera-

ture suggests

that

boys are,

in

general, more

likely

to

imitate aggressive behavior than

are

girls

(Flanders, 1968).

The

observed

sex

dif-

ference

in

response

to

marital turmoil

may

simply

be a

subset

of

this general

sex

dif-

CHILDREN

OF

DISCORD

AND

DIVORCE

321

ference.

An

alternative

modeling explana-

tion

is

that

fathers

in an

unhappy

marriage

are

more aggressive

and

uncooperative

and

that boys imitate fathers more than girls

do.

There

is

evidence that children

are

more

likely

to

imitate

a

same-sex model (Bandura,

1969;

Margolin

&

Patterson,

1975).

The

possibility

that

husbands

in

disturbed mar-

riages

act

more aggressively, whereas wives

react with more anxiety

and

withdrawal,

is

readily translated into testable research.

Modeling hypotheses

may

also explain

the

sex

differences

in

children's separation

and

life-change

responses

to

different

custody

arrangements. Because mothers have cus-

tody

in

most divorces

(Glick,

1979),

divorce

may

have

a

more noticeable

effect

on

boys

because

the

male model

is

missing.

Santrock

and

Warshak's (1979)

finding

of

better

ad-

justment when

the

same-sex parent

has

cus-

tody

is

congruent with

the

hypothesis,

as is

evidence suggesting that contact with adults

other than

the

custodial parent

is

beneficial

to

the

child's development (Jacobson, 1978).

Its

apparent

fit

with data

on the

various

parameters

of

marital

and

child problems

is

a

strength

of the

modeling hypothesis. Nev-

ertheless, modeling explanations, unlike

at-

tachment theory, typically have been

offered

post hoc. Further, several variations

of the

hypothesis

offer

viable

etiological

explana-

tions.

These

variations deserve further spec-

ification

and

testing.

Discipline Practices

A

third mechanism

by

which marital tur-

moil

may

affect

children

is

through

an al-

teration

in

discipline practices. Clinic

and

nonclinic

parents have been differentiated

on

a

variety

of

measures

of

discipline

(Becker,

Peterson, Luria, Shoemaker,

&

Hellmer,

1962);

it is

possible that marital turmoil

leads

to a

change

in the use of

important

discipline

techniques,

to the

detriment

of the

child,

or it may

lead

to

increased inconsis-

tency

in

discipline

either between parents

or

in

the

practices

of a

single parent.

Potential

discipline hypotheses

are nu-

merous because

a

wide range

of

discipline

constructs have

been

offered.

These

con-

structs include such traditional dimensions

of

child rearing

as

warm-hostile, restric-

tive-permissive,

and

calm-anxious (Becker,

1964).

In

addition, reinforcement theories

such

as

coercion

(Patterson

&

Reid, 1970)

and the

positive reinforcer trap (Wahler,

1976) have been

offered.

An

attempt

to ad-

dress each dimension

in

detail

would

be

cum-

bersome; therefore,

the

following

discussion

treats discipline practices

as a

group.

This general approach prohibits

a

detailed

evaluation

in

terms

of the

parameters

of the

relation between child

and

marital problems.

Still,

there

are

congruencies. Discipline

is a

frequent

topic

of

argument,

and

disagree-

ment about discipline

in

front

of a

child

ob-

viously

produces more inconsistent discipline

than when

the

disagreement

is

kept private

(Hetherington

et

al.,

1976; Rutter, 1972).

Problems

of

conduct

and

aggression have

frequently

been related

to

inconsistent dis-

cipline

(Becker, 1964; Becker

et

al.,

1962;

McCord,

&

Howard, 1961; Pat-

terson, 1977); thus,

an

increased frequency

of

undercontrolled behavior might

be ex-

pected.

But

insofar

as

these

are not the

only

aspects

of

child behavior that

are

affected

by

inconsistent discipline, more specific dis-

cipline

predictions need

to be

formulated.

For

example,

a

coercion hypothesis might

predict

that

parents

in

marital turmoil

be-

come

more

likely

to

give

in to a

child's

coer-

cive

demands.

As

with

modeling theory, discipline

ex-

planations

of sex

differences

are of

great

in-

terest. Several accounts

are

plausible.

For

example, evidence suggests that parents

are

more

involved

with disciplining

the

same-sex

child

than

the

opposite-sex child (Baumrind,

1971).

If

marital turmoil causes fathers

to

alter their discipline more radically than

mothers

do, a

greater

effect

would

be

seen

on

sons. Researchers have previously been

warned

not to

ignore

the

father's

effect

on

the

family

(Becker, Peterson, Hellmer,

Shoemaker,

&

Quay,

1959).

The

father's

role

is

also highlighted

if one

hypothesizes

that

sons

receive more

incon-

sistent discipline during times

of

marital tur-

moil

than

do

girls. Research shows that sons

are

disciplined about equally

by

both

par-

ents, whereas daughters

are

disciplined more

by

their mothers (Margolin

&

Patterson,

322

ROBERT

E.

EMERY

1975).

If

parents disagree about discipline,

boys

should

be

affected more than girls

be-

cause boys

are

disciplined more often

by

both

parents.

For

example, fathers

may

counteract maternal discipline

of

their sons

but not

object

to the

mothers' rules

for

their

daughters.

In

short, sons more frequently

than

daughters

may

become

the

focus

of

in-

terparental

conflict

about discipline.

Direct evidence

on the

relation between

marital turmoil

and

discipline practices

is

rare.

One

sound empirical study

(Hether-

ington

et

al.,

1976) documented changes

in

postdivorce

discipline. When compared with

parents

from

intact marriages, divorced par-

ents were

found

to

make

fewer

maturity

de-

mands, have poorer communication,

be

less

affectionate,

and be

more inconsistent with

their

children. Effects

of the

discipline

changes

were

that

children

of

divorce, espe-

cially

boys, were less compliant with paren-

tal

commands than were children

of

intact

marriages. This latter

finding is

particularly

important

to the

discipline hypothesis

be-

cause

it was

also found that boys were

ex-

posed

to

more

ineffective

discipline than

were girls

(Hetherington

et

al.,

1978).

Changes

in

discipline practices were

greatest

1

year after

the

divorce.

By the

sec-

ond

year, differences between intact

and di-

vorced parents were less noticeable

(Heth-

erington

et

al.,

1976).

It is not

known, how-

ever,

whether

the

changes

in

discipline began

before

or

after

the

divorce. This

is a

very

important

distinction

for

questions

of

etiol-

ogy

and

treatment.

Still,

the

study demon-

strated that discipline

was

disrupted

by

mar-

ital turmoil, although when

the

change

oc-

curred

and

what type

of

change most affected

the

child

are

important questions

that

have

been

left

unanswered.

Supporting

the

possibility

that

discipline

may

change

before

a

divorce,

a

study

of in-

tact

marriages (Block

et

al.,

1981) found

that

an

index

of

parental disagreement about

child rearing

was

to

subsequent

marital dissolution

and to

future undercon-

trol

in

boys'

and

overcontrol

in

girls' school

behavior.

Inconsistency

in

discipline

was

thus shown

to be

both

an

aspect

of

marital

turmoil

and a

precursor

of

problems

in

chil-

dren. Such prospective

findings

provide

strong support

for the

hypothesis

that

in-

consistent discipline

is a

mediating factor

between marital turmoil

and

child behavior

problems.

As

such,

the

investigation provides

a

model

for

future research.

In

sum, discipline

is an

important

influ-

ence

on

children

and one

that

is

altered

con-

comitantly

with marital problems. Inconsis-

tent discipline,

in

particular, demands con-

tinued attention, according

to the

results

of

two

well-constructed investigations.

Other Hypotheses

Three additional

hypotheses—stress,

tak-

ing

on the

symptom,

and

child-effects—

make important contributions

to

under-

standing

how

marital turmoil

may

affect

children.

Because

the

data

on

these

hy-

potheses

are

more limited than that

on the

their unique implica-

tions

are

presented below with less critical

evaluation

of

each hypothesis

as a

whole.

Stress.

Marital problems

can be

concep-

tualized

simply

as a

stressor

to

which chil-

dren,

as

well

as

adults,

are

differentially

ex-

posed. Marital disruption

has

been noted

to

have

pathogenic

effects

on

adults (Bloom,

Asher,

&

White,

1978),

and it may

similarly

pose problems

for

children

in

that coping

with

life's

difficulties

takes

a

psychological

toll.

A

stress

hypothesis

is of

interest

primarily

in

explaining

the

differential

sex

effects

of

marital turmoil.

The

simplest stress hypoth-

esis would suggest that exposure

to the

stressor

is

different

for

each

sex.

That

is,

girls

may

be

shielded from interparental

conflict

more than

are

boys, Clinical impressions

that

this

is

indeed

the

case

have been

offered

(Wallerstein

&

Kelly, 1980). Research,

however,

has

shown that parents report

fighting

equally

in

front

of

children

of

both

sexes

(Porter

&

O'Leary,

1980)

and

that

boys

and

girls report similar awareness

of

discord

between

their

parents (Emery

&

O'Leary,

1982). Further,

if

differential

ex-

posure accounted

for the sex

effects, mea-

sures

of

open conflict

or of

children's aware-

ness

of

conflict

should

relate

at

least

to

over-

controlled problems

in

girls,

but

this

has not

been

the

case

(Emery

&

O'Leary,

1982;

Porter

&

O'Leary,

1980).

As an

alternative,

the

sexes

may be ex-

CHILDREN

OF

DISCORD

AND

DIVORCE

323

posed equally

to

interparental

conflict,

but

girls

may be

less susceptible

to the

stressor.

One

such explanation

is

that girls

are

less

susceptible

than

boys

to a

variety

of

physical

stressors; therefore, they

may

also have

a

greater resistance

to

psychological stress

(Erne,

1979;

Rutter,

1970).

A

second such explanation

of the sex

dif-

ference

is

suggested

by a

recent investiga-

tion.

Male

and

female college students were

found

to be

equally accurate

in

recognizing

marital

conflict,

but the two

sexes made dif-

ferent

interpretations

of the

conflict. Specif-

ically,

when judging marital interactions

that were seen

as

equally discordant

by

both

sexes, college women rated

the

couple

as

more

"right

for

each

other"

(Epstein,

Fin-

negan,

&

Bythell,

1979). This

suggests

that

sex-role

socialization

may

explain

the

dif-

ferent effect

on

boys

and

girls. Although

both

sexes

are

equally aware

of the

discord,

parental conflict

may be

interpreted

differ-

ently

by

girls,

who may

view

the

same

stressor

as

being

less

stressful.

A

diathesis-stress model

offers

a final

stress

explanation

of the sex

difference.

Base-rate data

on

childhood disorders

in

the

United

States

suggest that boys

are

predis-

posed

toward undercontrol, whereas girls

are

predisposed

toward overcontrol

(Ross,

1980).

Given

this diathesis

and a

marital turmoil

stressor, each

sex may be at

risk

for

devel-

oping

a

disorder. Because undercontrol

is

more easily recognized

and

less tolerated

than

overcontrol, however,

the finding

that

boys

are

more

affected

than girls

by

marital

turmoil

would

be

expected.

Taking

on the

symptom.

The

hypothesis

that

one

member

of a

family becomes

the

focus

of

larger family conflicts also merits

consideration. Minuchin

and his

colleagues

(Minuchin,

1974; Minuchin

et

al.,

1975)

suggested

that

children

serve

the

function

of

distracting their parents

from

their

own

con-

flicts

by

developing

a

problem

themselves

and

thus redirecting parental concerns.

Chil-

dren's defusion

of

interparental conflict

by

"taking

on the

symptom"

is,

therefore,

a

fifth

etiological

hypothesis.

This hypothesis

offers

yet

another expla-

nation

of the sex

differences

and

also

might

explain

some data

on the

form

of the

child's

behavioral

response.

Boys

may

take

on the

symptom

of

interparental

conflict

in the

sex-

role-appropriate manner

of

becoming

ag-

gressive, noncompliant,

and so on.

This very

noticeable disruptiveness

may

indeed dis-

tract

a

boy's parents

from

their

own

prob-

lems.

Undercontrolled behavior

is

more

ob-

vious

than overcontrolled behavior

and

might

facilitate

the

distraction.

But

why

wouldn't girls serve

the

same

function?

Perhaps they

do, but

girls' behav-

ioral distractions might have

a

very

different

topography.

Rather than being disruptive,

girls

may

become anxious, withdrawn,

or

even

very well behaved. Girls

may

actually

refocus

parental attention

by

their extreme

prosocial behavior.

For

girls,

prosocial

and

overcontrolled

behavior

is

more sex-role

ap-

propriate

than

is

undercontrolled behavior,

therefore being nice

or

worried

may

serve

the

distraction

function

better.

This

hypoth-

esis would explain

the

paradox

in the

Block

et al.

(1981)

data

that

showed

that,

for

girls,

parental disagreement

was

over-

control

in the

expected

direction

but was in-

versely

undercontrol.

At the

least,

a

taking-on-the-symptom

hypothesis sug-

gests that prosocial

or

competent behavior

be

included

as a

dependent measure

in

stud-

ies of

marital turmoil.

Child-effects,

Instead

of the

assumption

that marital turmoil causes problems

in

chil-

dren,

the

reverse argument could

be

true;

namely,

that

a

deviant child places strain

on

a

marriage. Bell

and

Harper's

(1977)

cogent

arguments

concerning such child-effects need

to be

considered. Indeed,

it has

been

found

that children

can

decrease marital satisfac-

tion

as

indexed

by

such

findings

as (a) the

decline

of

marital satisfaction after

the

birth

of

the first

child,

(b) the

negative

effect

on

a

marriage

as a

result

of

rearing

a

handi-

capped child,

and (c)

survey data wherein

parents report

that

their

children

are an

added stress

on

their marriage (Lerner

&

Spanier,

1978).

It is

likely

that

a

conduct

problem child would similarly strain

a

mar-

riage.

Despite

the

consideration

of a

child-ef-

fects

alternative,

the

causal sequence

from

parent

to

child

is

probably

a

more important

pathway.

One

group

of

researchers

(Olt-

manns

et

al.,

1977)

found

that marital sat-

isfaction

did not

increase

concomitantly

with

324

ROBERT

E.

EMERY

parent-rated improvement

in

problem chil-

dren's

behavior,

a

result contrary

to a

child-

effects

prediction. Further,

from

a

strict

child-effects

position,

it

becomes

difficult

to

explain

why

boys disrupt

a

marriage more

or why

open

conflict

is the

principal result

of

a

child's

problem behavior.

Still,

as

others

have

noted (Lerner

&

Spanier, 1978; Mar-

golin,

in

press),

the

best

explanation

of the

relation

between

the two

areas

of

difficulty

is

reciprocal

influence.

Marital

and

child

problems

are

best viewed

as

interactive:

To

an

extent, each causes

and

exacerbates

the

other.

Summary

It is

unlikely that

any

single hypothesis

fully

explains

the

relation between marital

and

child problems,

yet

each

may

prove

to

have

merit.

Parents

involved

in

conflict with

each other

are

probably poorer models,

are

more

inconsistent

in

their

discipline,

and

place more stress

on

their children. Some

children

probably serve

to

distract attention

away

from

parental

conflict,

whereas others

may

aggravate

the

conflict.

Attachment

bonds

are

certainly disrupted

by

separation

and

may

also

be

affected

by

conflict.

These

(and perhaps

other)

processes

are

likely

to

operate collectively

in

affecting

the

children

of

marital turmoil, although,

in any

given

instance,

one

influence

may

predominate.

When

the

hypotheses

are

compared,

mod-

eling

and

discipline

practices

would

seem

to

hold

particular merit

in

light

of the

available

data.

Still,

because

most

of the

data

have

preceded

specific

predictions, even these

two

etiological accounts cannot

be

given strong

support.

This general lack

of

hypothesis test-

ing

is

characteristic

of the

literature

on

mar-

ital

and

child problems

and

should

not

con-

tinue.

It is

hoped that

the

above delineation

of

various

potential

etiological

mechanisms

will

encourage

future

investigators

to

make

specific

predictions based

on

theoretical

ra-

tionales.

Implications

for

Treatment

Although questions

of

etiology

are far

from

being answered,

at

least

one

implica-

tion

of the

research

on

marital

and

child

problems

appears

to be

clear:

Efforts

should

be

made

to

minimize children's involvement

in

interparental conflict. Although this

may

seem

to be

only common sense, there

are

therapists

who

argue

that

one

must increase

family-wide

conflict

following

a

"civilized"

divorce

as a

means

of

allowing

the

members

to

work through their feelings (e.g., Futter-

man,

1980).

At the

level

of

prevention,

four

suggestions

can be

offered.

First, parents should

work

toward

the

difficult

goal

of

keeping their

children

out of

their angry disagreements,

lest

the

children learn

that

differences

are

resolved

by

yelling,

fighting, or

hitting. Sec-

ond, parents should

attempt

always

to

agree

in

front

of the

child about

at

least

one im-

portant

topic: discipline.

Third,

parents

should

make

a

special

effort

to

maintain

their individual relationship with each child,

as

this

may

partially

buffer

the

child from

the

interparental conflict. Finally, more par-

ents need

to be

aware

that

conflicts between

them

can

have negative

effects

on

their chil-

dren.

Parents need

to be

sensitive

to how

their

children react

to

marital turmoil

and

to be

prepared

to

seek outside help

if

these

reactions

are

prolonged.

At the

treatment level,

families

in

which

marital

and

child problems

coexist

present

particularly

difficult

problems. This

is

docu-

mented

both

by

clinical impressions (Kent

&

O'Leary, 1976; Patterson, Cobb,

&

Ray,

1973)

and by

research that suggests that

treatment

is

less

effective

(Strain,

Young,

&

Horowitz, 1979; Clark

&

Baker, Note

5).

Because

of

this

difficulty

therapists

need

to

be

open

to new

approaches

and

roles.

For

example,

the

suggestion that individually

oriented

interventions

may be

less successful

than

treatment that takes place

on the

fam-

ily

level

(e.g.,

Minuchen,

1974) deserves

consideration.

Behavioral

family

therapy

was

shown

in one

study (Oltmanns

et

al.,

1977)

to be

effective

despite

the

presence

of

marital discord.

As an

alternative, recent

evidence suggests that

the

marriage rather

than

the

entire

family

may be the

more

ap-

propriate initial level

for

treatment

(Mar-

golin

&

Christensen,

Note

6).

Families

with

both marital

and

child problems were found

to

improve more

on

self-report

and

obser-

vational measures

of the

family,

the

mar-

CHILDREN

OF

DISCORD

AND

DIVORCE

325

riage,

and the

child when randomly assigned

to

behavioral marital therapy

as

opposed

to

family

treatment

(Margolin

&

Christensen,

Note

6).

Although

these

data

are

prelimi-

nary,

they

point

out the

importance

of

com-

parisons

of

innovative treatments.

Child

and

family therapists must also

be

willing

to

adopt

other,

nontraditional

ap-

proaches. Parents

who are

divorced

or who

come

from

discordant marriages often

do not

receive

the

spousal

support

that

can

main-

tain

positive

efforts

at

child rearing

(Wahler,

1980).

Therapists, therefore, should consider

group

approaches

to

parent training when

family

and

couple options

are

impossible.

Therapists also need

to

take

a

more

active

role

in

legal decisions

involving

marital

and

child

problems.

For

example, Haynes

(1978)

suggested

a new

role

for

mental health

professionals

as

divorce mediators. Rather

than

resolving decisions such

as

child cus-

tody

and

visitation

via the

adversarial legal

process,

it is

possible that more

of

these

de-

cisions

can be

reached with

the

help

of an

impartial

mediator.

The

latter

process would

seem

to

offer

opportunities

for

reducing con-

flict,

whereas

the

former procedure

is

likely

to

exacerbate

it.

Finally,

therapists need

to

keep

in

touch

with

the

emerging

empirical

literature

and

distinguish

it

from opinion.

For

example,

contrary

to

popular

belief,

the

literature

re-

viewed

here

suggests

that

divorce

may

ulti-

mately

result

in a

less detrimental environ-

ment

for a

child's

emotional development,

at

least

in

homes

where

conflict

is

great

and

where

divorce

will

lead

to a

diminution

of

that

conflict.

Such

an

interpretation might

not

only benefit children from

intact

but

con-

tinually

conflicted

families,

but it

might also

relieve

divorced parents' guilt

over

the ir-

reversible damage they

may

believe they

have caused their children. Further,

the fo-

cus on

interparental

conflict

underscores

both parents' continued responsibility

to im-

prove

relations with

the

former spouse

for

the

children's sake. Nevertheless, despite

the

fact

that divorce

may

sometimes

be the

least

detrimental alternative, evidence

does

sug-

gest

that

divorcing parents should

be

pre-

pared

for

three unique reactions

from

their

children: conflict responses, separation

re-

sponses,

and

life-change responses. Parents

should

also know that, unfortunately, psy-

chologists

do not yet

have

all the

answers

to

the

many questions

on how

marital turmoil

affects

children.

Future Trends

If

answers

to

some

of the

questions

are to

be

forthcoming, much more sound research

will

have

to be

done. Given below

are

some

interpretative

and

methodological

consider-

ations that

will

need

to be

addressed

in ad-

dition

to

specific proposals

for

future

re-

search.

Methodological

Considerations

Because marital

and

child problems

can

be

studied only

in a

correlational design

and

because alternative interpretations

are

theo-

retically

relevant

to the

topic,

reverse cau-

sality

and

third variable interpretations need

to be

considered. Reverse causality

was

dis-

cussed earlier

as the

etiological model

of

child-effects;

only third variables

are

consid-

ered now.

A

variety

of

strategies

are

available

for

investigating

the

effects

of

third variables.

The

important point

is to

recognize those

variables

that

are

potential

mediators.

In the

marital-child area, variables related

to

mar-

ital

turmoil such

as

social class

and

parental

psychopathology

can be

controlled

for in

various

research designs.

Indeed,

it has

been

shown

that these

two

variables

do not

fully

account

for the

observed

relation

(Rutter,

1971;

Rutter

et

al.,

1974), although

the fi-

nancial

decline experienced

by

one-parent

families

clearly

produces

important

life-

change responses

in the

children

(Colletta,

1979;

Herzog

&

Sudia, 1973). Other vari-

ables such

as

postdivorce custody arrange-

ments,

visitation

patterns,

and

social support

networks

also need

to be

measured

and

con-

trolled

as

relevant third variables.

One

third variable mentioned earlier

is

worth

discussing again: response bias

in

non-

independent

reports

of

marital

and

child

problems. Recent investigations suggest

that

parents' reports

of

behavior

problems

in

their

children

can

sometimes

reflect

more

about

the

parent than about

the

child.

In one

study (Griest,

Wells,

&

Forehand, 1979),

326

ROBERT

E.

EMERY

maternal

ratings

of

child adjustment were

more strongly related

to

maternal depression

than

to the

child's

observed

behavior.

It was

argued

that

the

depression

had

colored

the

mothers' perceptions.

It is

possible

that

a

similar bias might

ex-

plain

some

of the

association

found

between

marital

turmoil

and

child behavior problems.

One way to

avoid this problem

is to use in-

dependent

raters

or

observational measures.

Because parents have unique

access

to

many

aspects

of

their children's behavior, however,

it is

important

to

determine whether marital

turmoil

does alter parents' perceptions

of

their

children.

The

comparison

of

parental

report with observational measures

(Griest

et

al.,

1979),

the

assessment

of

rating biases

such

as

social desirability (Robinson

& An-

derson,

Note

3), and the

measurement

of

cognitive

set

toward evaluating children's

behavior

(Messe,

Stollak,

Larson,

& Mi-

chaels,

1979)

are

three methods that

can be

used

in

obtaining data

on

this issue.

In

addition

to

third-variable considera-

tions,

methodological improvements

are

needed.

The

taxonomy

of

child

and

marital

problems

and the

instruments available

to

measure their dimensions need

to be

devel-

oped further.

In the

assessment

of

children,

more reliable measures

of

social competence

and

of

overcontrol

are

especially needed.

With regard

to

marital problems, instru-

ments

are

needed that

will

assess

the

process

and the

length

of

conflict independently

along

different

continuua,

within

discrete

Archives

of

General

Psychiatry,

1970,

23,

421-427.

Mednick,

S.

A.,

&

McNeil,

T. F.

Current methodology

in

research

on the

etiology

of

schizophrenia: Serious

difficulties

which

suggest

the use of the

high-risk

group

method. Psychological

Bulletin,

1968,

70,

681-

693.

Messe,

L.

A.,

Stollak,

G.

E.,

Larson,

R.

W.,

&

Michaels,

G. Y.

Interpersonal

consequences

of

person

percep-

tion

processes

in two

social contexts.

Journal

of

Per-

sonality

and

Social

Psychology, 1979,

37,

369-379.

Minuchin,

S.

Families

and

family

therapy,

Cambridge,

Mass.:

Harvard University Press, 1974.

Minuchin,

S. et

at.

A

conceptual model

of

psychoso-

matic

illness

in

children: Family organization

and

family

therapy.

Archives

of

General Psychiatry,

1975,

32,

1031-1038.

Molholm,

L.

H.,

&

Dinitz,

S.

Female

mental patients

and

their normal controls.

Archives

of

General Psy-

chiatry, 1972,

27,

606-610.

Nye,

F. I.

Child

adjustment

in

broken

and

unhappy

unbroken

homes. Marriage

and

Family

Living,

1957,

19.

356-361.

O'Leary,

K.

D.,

&

Emery,

R. E.

Marital discord

and

child

behavior problems.

In M. D.

Levine

& P.

Satz

(Eds.),

Developmental

variation

and

dysfunction.

New

York: Academic Press,

in

press.

O'Leary,

K.

D.,

&

Turkewitz,

H.

Methodological errors

in

marital

and

child

treatment

research.

Journal

of

Consulting

and

Clinical Psychology,

1978,

46,

747-

758.

Olson,

D.

H.,

&

Strauss,

M. A. A

diagnostic tool

for

marital

and

family

therapy. Family Coordinator,

1972,

21,

251-258.

Oltmanns,

T.

F.,

Broderick,

J.

E.,

&

O'Leary,

K. D.

Marital adjustment

and the

efficacy

of

behavior ther-

apy

with

children. Journal

of

Consulting

and

Clinical

Psychology,

1977,

45,

724-729.

Patterson,

G. R.

Accelerating stimuli

for two

classes

of

coercive

behaviors. Journal

of

Abnormal

Child Psy-

chology, 1977,

5,

335-350.

Patterson,

G.

R.,

Cobb,

J.

A.,

&

Ray,

R. S. A

social

engineering

technology

for

retraining

the

families

of

aggressive boys.

In H. E.

Adams

& I. P.

Unikel

(Eds.),

Issues

and

trends

in

behavior therapy. Spring-

field,

111.:

Charles

C

Thomas, 1973.

Patterson,

G.

R.,

&

Reid,

J. B.

Reciprocity

and

coer-

cion:

Two

facets

of

social systems.

In C.

Neuringer

& J.

Michaels

(Eds.),

Behavior modification

in

clin-

ical psychology.

New

York: Appleton-Century-Crofts,

1970.

Patterson,

G.

R.,

Weiss,

R.

L.,

&

Hops,

H.

Training

in

marital skills:

Some

problems

and

concepts.

In H.

Leitenberg

(Ed.),

Handbook

of

behavior

modifica-

330

ROBERT

E.

EMERY

lion

and

behavior therapy.

Englewood

Cliffs,

N.J.:

Prentice-Hall,

1976.

Pemberton,

D.

A.,

&

Benady,

D. R.

Consciously

re-

jected

children.

British

Journal

of

Psychiatry, 1973,

123.

575-578.

Porter,

B., &

O'Leary,

K.

D.

Marital

discord

and

child-

hood

behavior

problems. Journal

of

Abnormal

Child

Psychology,

1980,

80,

287-295.

Power,

M. J.,

Ash,

P. M.,

Schoenberg,

E.,

&

Sorey,

E.

C.

Delinquency

and the

family.

British

Journal

of

Social

Work,

1974,

4,

17-38.

Rosenthal,

R.

Experimenter bias

in

behavioral

re-

search.

New

York:

Appleton-Century-Crofts,

1966.

Ross,

A. O.

Psychological disorders

of

children.

New

York:

McGraw-Hill, 1980.

Rutter,

M. Sex

differences

in

response

to

family

stress.

In

E. J.

Anthony

& C.

Koupernik

(Eds.),

The

child

in

his

family.

New

York:

Wiley,

1970.

Rutter,

M.

Parent-child separation: Psychological

ef-

fects

on the

children.

Journal

of

Child Psychology

and

Psychiatry

and

Allied

Disciplines,

1971,12,

233-

260.

Rutter,

M. The

quality

of

mothering: Maternal depri-

vation reassessed.

Harmondsworth,

Middlesex,

En-

gland:

Penguin,

1972.

Rutter,

M.

Maternal

deprivation,

1972-1978:

New

findings,

new

concepts,

new

approaches. Child

De-

velopment, 1979,

50,

283-305.

Rutter,

M.

Protective

factors

in

children's responses

to

stress

and

disadvantage.

In M.

W.

Kent

& J. E.

Rolf

(Eds.),

Primary

prevention

of

psychopathology:

111.

Promoting social competence

and

coping

in

children.

Hanover,

N.H.:

University

Press

of New

England,

1980.

Rutter,

M.,

Graham,

P.,

Chadwick,

O. F.

D.,

&

Yule,

W.

Adolescent turmoil:

Fact

or fiction?

Journal

of

Child

Psychology

and

Psychiatry

and

Allied

Disci-

plines,

1976,

17,

35-56.

Rutter,

M.

et

al.

Attainment

and

adjustment

in two

geographical

areas: III. Some factors

accounting

for

area

differences.

British Journal

of

Psychiatry, 1974,

125,

520-533.

Santrock,

J.

W.,

&

Warshak,

R. A.

Father

custody

and

social

development

in

boys

and

girls.

Social Issues,

1979,

35,

112-125.

Schaefer,

E. S.

Children's

reports

of

parental

behavior:

An

inventory.

Child Development,

1965,36,

413-421.

Schwarz,

J. C.

Childhood origins

of

psychopathology.

American Psychologist, 1979,

34,

879-885.

Strain,

P.

S.,

Young,

C.

C.,

&

Horowitz,

J. An ex-

amination

of

child

and

family

demographic

variables

treatment

outcomes

during

oppositional

child

training.

Behavior Modification, 1979,

3, 61-

76.

Tuckman,

J.,

&

Regan,

R. A.

Intactness

of

home

and

behavioural

problems

in

children.

Journal

of

Child

Psychology

and

Psychiatry, 1966,

7,

225-233.

U.S. Census

Bureau.

Statistical

abstract

and the

United

States.

Washington,

D.C.:

Author,

1980.

Wahler,

R. G.

Deviant child behavior

within

the

family:

Developmental

speculations

and

behavior

change

strategies.

In H.

Leitenberg

(Ed.),

Handbook

of be-

havior

modification

and

behavior therapy.

Engle-

wood

Cliffs,

N.J.:

Prentice-Hall,

1976.

Wahler,

R. G. The

insular

mother:

Her

problems

in

parent-child

treatment.

Journal

of

Applied Behavior

Analysis, 1980,

13,

207-219.

Wallerstein,

J.

S.,

&

Kelly,

J. B. The

effects

of

parental

divorce:

The

adolescent experience.

In E. J.

Anthony

& C.

Koupernik

(Eds.),

The

child

in his

family: Chil-

dren

at

psychiatric

risk.

New

York:

Wiley,

1974.

Wallerstein,

J.

S.,

&

Kelly,

J. B. The

effects

of

parental

divorce:

Experiences

of

the

preschool

child.

Journal

of

the

American Academy

of

Child Psychiatry, 1975,

14,

600-616.

Wallerstein,

J.

S.,

&

Kelly,

J. B.

Surviving

the

breakup.

New

York: Basic

Books,

1980.

Westman,

J.

D.,

Cline,

D.

W.,

Swift,

W.

J.,

&

Kramer,

D. A. The

role

of

child

psychiatry

in

divorce.

Archives

of

General Psychiatry, 1970,

23.

416-420.

Whitehead,

L. Sex

differences

in

children's

responses

to

family

stress:

A

re-evaluation.

Journal

of

Child

Psychology

and

Psychiatry

and

Allied

Disciplines,

1979,

20,

247-254.

Wolkind,

S.,

&

Rutter,

M.

Children

who

have

been

"in

care"—An

epidemiological

study.

Journal

of

Child

Psychology

and

Psychiatry

and

Allied

Disci-

plines, 1973,

14.

97-105.

Received

May

26,

1981

•

View publication statsView publication stats